diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,27931 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.999892507793185,
+  "global_step": 4651,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 7.0898,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 6.8672,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 6.8906,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.714285714285714e-05,
+      "loss": 6.668,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.142857142857142e-05,
+      "loss": 5.8945,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.571428571428571e-05,
+      "loss": 5.7773,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 5.4805,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011428571428571428,
+      "loss": 5.3086,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012857142857142855,
+      "loss": 5.168,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014285714285714284,
+      "loss": 4.9727,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015714285714285713,
+      "loss": 4.8711,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017142857142857143,
+      "loss": 4.8633,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018571428571428572,
+      "loss": 4.6055,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 4.4727,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00021428571428571427,
+      "loss": 4.3711,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00022857142857142857,
+      "loss": 4.248,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00024285714285714286,
+      "loss": 4.127,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002571428571428571,
+      "loss": 4.1289,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002714285714285714,
+      "loss": 3.9277,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002857142857142857,
+      "loss": 3.8789,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0003,
+      "loss": 3.8105,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00031428571428571427,
+      "loss": 3.834,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00032857142857142856,
+      "loss": 3.8418,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00034285714285714285,
+      "loss": 3.7285,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00035714285714285714,
+      "loss": 3.6172,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00037142857142857143,
+      "loss": 3.5898,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003857142857142857,
+      "loss": 3.4688,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004,
+      "loss": 3.6074,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004142857142857143,
+      "loss": 3.4336,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00042857142857142855,
+      "loss": 3.4648,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00044285714285714284,
+      "loss": 3.4277,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00045714285714285713,
+      "loss": 3.3926,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004714285714285714,
+      "loss": 3.2578,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004857142857142857,
+      "loss": 3.2852,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005,
+      "loss": 3.4043,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005142857142857142,
+      "loss": 3.3125,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005285714285714286,
+      "loss": 3.3008,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005428571428571428,
+      "loss": 3.0801,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005571428571428572,
+      "loss": 3.2988,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005714285714285714,
+      "loss": 3.0898,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005857142857142858,
+      "loss": 3.1523,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006,
+      "loss": 3.2734,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006142857142857143,
+      "loss": 3.1133,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006285714285714285,
+      "loss": 3.3008,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006428571428571429,
+      "loss": 3.166,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006571428571428571,
+      "loss": 3.1973,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006714285714285714,
+      "loss": 3.0312,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006857142857142857,
+      "loss": 3.2461,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007,
+      "loss": 3.1133,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007142857142857143,
+      "loss": 3.1543,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007285714285714286,
+      "loss": 3.0352,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007428571428571429,
+      "loss": 3.1816,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007571428571428572,
+      "loss": 3.2793,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007714285714285715,
+      "loss": 3.2031,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007857142857142857,
+      "loss": 3.084,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008,
+      "loss": 3.0137,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008142857142857143,
+      "loss": 3.0566,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008285714285714286,
+      "loss": 3.1152,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008428571428571429,
+      "loss": 3.1602,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008571428571428571,
+      "loss": 3.1445,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008714285714285715,
+      "loss": 3.1367,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008857142857142857,
+      "loss": 3.1895,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009000000000000001,
+      "loss": 3.0781,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009142857142857143,
+      "loss": 3.0664,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009285714285714287,
+      "loss": 3.082,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009428571428571429,
+      "loss": 3.0996,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009571428571428573,
+      "loss": 2.8984,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009714285714285714,
+      "loss": 3.0117,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009857142857142857,
+      "loss": 3.0254,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001,
+      "loss": 2.9844,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0010142857142857143,
+      "loss": 2.8633,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0010285714285714284,
+      "loss": 2.9941,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001042857142857143,
+      "loss": 2.9844,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0010571428571428572,
+      "loss": 2.7793,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0010714285714285715,
+      "loss": 2.9492,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0010857142857142856,
+      "loss": 2.9336,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0011,
+      "loss": 2.873,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0011142857142857144,
+      "loss": 2.8633,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0011285714285714287,
+      "loss": 2.9453,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0011428571428571427,
+      "loss": 2.8516,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0011571428571428572,
+      "loss": 2.8574,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0011714285714285715,
+      "loss": 2.8359,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0011857142857142858,
+      "loss": 2.8008,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0012,
+      "loss": 2.8145,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0012142857142857142,
+      "loss": 2.9785,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0012285714285714287,
+      "loss": 2.7695,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001242857142857143,
+      "loss": 2.752,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001257142857142857,
+      "loss": 2.6719,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0012714285714285714,
+      "loss": 2.7031,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0012857142857142859,
+      "loss": 2.8613,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0013000000000000002,
+      "loss": 2.709,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0013142857142857142,
+      "loss": 2.7188,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0013285714285714285,
+      "loss": 2.7793,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0013428571428571428,
+      "loss": 2.6387,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0013571428571428573,
+      "loss": 2.7891,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0013714285714285714,
+      "loss": 2.709,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0013857142857142857,
+      "loss": 2.6973,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0014,
+      "loss": 2.625,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0014142857142857145,
+      "loss": 2.5742,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0014285714285714286,
+      "loss": 2.6641,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0014428571428571429,
+      "loss": 2.6484,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0014571428571428572,
+      "loss": 2.6348,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0014714285714285717,
+      "loss": 2.6211,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0014857142857142857,
+      "loss": 2.6895,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0015,
+      "loss": 2.4727,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0015142857142857143,
+      "loss": 2.5762,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0015285714285714284,
+      "loss": 2.6094,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001542857142857143,
+      "loss": 2.8281,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0015571428571428572,
+      "loss": 2.6797,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0015714285714285715,
+      "loss": 2.5762,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0015857142857142856,
+      "loss": 2.6113,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0016,
+      "loss": 2.5449,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0016142857142857144,
+      "loss": 2.6621,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0016285714285714287,
+      "loss": 2.875,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0016428571428571427,
+      "loss": 2.6855,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0016571428571428572,
+      "loss": 2.6367,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0016714285714285715,
+      "loss": 2.627,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0016857142857142858,
+      "loss": 2.6289,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0017,
+      "loss": 2.5586,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0017142857142857142,
+      "loss": 2.6543,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0017285714285714287,
+      "loss": 2.6113,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001742857142857143,
+      "loss": 2.7363,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001757142857142857,
+      "loss": 2.6934,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0017714285714285714,
+      "loss": 2.6777,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0017857142857142859,
+      "loss": 2.6504,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0018000000000000002,
+      "loss": 2.7246,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0018142857142857142,
+      "loss": 2.6348,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0018285714285714285,
+      "loss": 2.6582,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0018428571428571428,
+      "loss": 2.6992,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0018571428571428573,
+      "loss": 2.8379,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0018714285714285714,
+      "loss": 2.6152,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0018857142857142857,
+      "loss": 2.6289,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019,
+      "loss": 2.5469,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019142857142857145,
+      "loss": 2.6113,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019285714285714286,
+      "loss": 2.7461,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019428571428571429,
+      "loss": 2.6543,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001957142857142857,
+      "loss": 2.7148,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019714285714285715,
+      "loss": 2.6074,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001985714285714286,
+      "loss": 2.6289,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.002,
+      "loss": 2.4961,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999999757493112,
+      "loss": 2.6699,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999999029972567,
+      "loss": 2.6309,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999978174387164,
+      "loss": 2.6172,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999996119892149,
+      "loss": 2.584,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999939373336884,
+      "loss": 2.5801,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999991269764393,
+      "loss": 2.6172,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999881171855563,
+      "loss": 2.5352,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999844795987073,
+      "loss": 2.5684,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999803570056106,
+      "loss": 2.6016,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999975749408266,
+      "loss": 2.5508,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999706568089074,
+      "loss": 2.5156,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999650792100056,
+      "loss": 2.6973,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999590166142655,
+      "loss": 2.5039,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999952469024627,
+      "loss": 2.5176,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999945436444267,
+      "loss": 2.6836,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999379188765953,
+      "loss": 2.6719,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999299163252584,
+      "loss": 2.5723,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999214287941378,
+      "loss": 2.6211,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999124562873494,
+      "loss": 2.6113,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999029988092463,
+      "loss": 2.6914,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019998930563644144,
+      "loss": 2.4746,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999882628957676,
+      "loss": 2.6719,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019998717165940886,
+      "loss": 2.6113,
+      "step": 163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019998603192789453,
+      "loss": 2.6895,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999848437017774,
+      "loss": 2.5137,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019998360698163372,
+      "loss": 2.5742,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019998232176806335,
+      "loss": 2.6328,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999809880616896,
+      "loss": 2.6016,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999796058631594,
+      "loss": 2.5547,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019997817517314305,
+      "loss": 2.582,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019997669599233453,
+      "loss": 2.5195,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019997516832145124,
+      "loss": 2.6152,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999735921612341,
+      "loss": 2.584,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019997196751244764,
+      "loss": 2.5547,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999702943758797,
+      "loss": 2.7598,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999685727523419,
+      "loss": 2.625,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999668026426692,
+      "loss": 2.6094,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019996498404772013,
+      "loss": 2.498,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999631169683768,
+      "loss": 2.5078,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019996120140554465,
+      "loss": 2.5527,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019995923736015283,
+      "loss": 2.7148,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019995722483315387,
+      "loss": 2.6367,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019995516382552395,
+      "loss": 2.5117,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999530543382627,
+      "loss": 2.5469,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019995089637239315,
+      "loss": 2.5078,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019994868992896197,
+      "loss": 2.5723,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019994643500903936,
+      "loss": 2.4883,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00199944131613719,
+      "loss": 2.5801,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019994177974411807,
+      "loss": 2.4609,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019993937940137713,
+      "loss": 2.5234,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019993693058666055,
+      "loss": 2.6113,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999344333011559,
+      "loss": 2.4316,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999318875460745,
+      "loss": 2.5762,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019992929332265106,
+      "loss": 2.5352,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999266506321438,
+      "loss": 2.5508,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999239594758344,
+      "loss": 2.5273,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999212198550282,
+      "loss": 2.4707,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999184317710539,
+      "loss": 2.5059,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019991559522526377,
+      "loss": 2.666,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999127102190336,
+      "loss": 2.541,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019990977675376264,
+      "loss": 2.707,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999067948308736,
+      "loss": 2.5449,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999037644518129,
+      "loss": 2.582,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019990068561805017,
+      "loss": 2.5234,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019989755833107873,
+      "loss": 2.4893,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001998943825924154,
+      "loss": 2.4453,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001998911584036005,
+      "loss": 2.5332,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001998878857661976,
+      "loss": 2.4512,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019988456468179416,
+      "loss": 2.4766,
+      "step": 209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019988119515200096,
+      "loss": 2.5488,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998777771784521,
+      "loss": 2.457,
+      "step": 211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998743107628055,
+      "loss": 2.582,
+      "step": 212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998707959067424,
+      "loss": 2.5469,
+      "step": 213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019986723261196755,
+      "loss": 2.5098,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998636208802091,
+      "loss": 2.4629,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998599607132189,
+      "loss": 2.5039,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998562521127721,
+      "loss": 2.6875,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019985249508066754,
+      "loss": 2.5039,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998486896187273,
+      "loss": 2.6426,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019984483572879717,
+      "loss": 2.5117,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998409334127463,
+      "loss": 2.6152,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998369826724674,
+      "loss": 2.5586,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019983298350987654,
+      "loss": 2.5254,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998289359269135,
+      "loss": 2.5547,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019982483992554137,
+      "loss": 2.5938,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998206955077467,
+      "loss": 2.623,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998165026755396,
+      "loss": 2.4648,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998122614309538,
+      "loss": 2.459,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998079717760462,
+      "loss": 2.6074,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998036337128974,
+      "loss": 2.5371,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019979924724361138,
+      "loss": 2.7422,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019979481237031574,
+      "loss": 2.4492,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997903290951613,
+      "loss": 2.6055,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019978579742032264,
+      "loss": 2.543,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019978121734799768,
+      "loss": 2.5645,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997765888804077,
+      "loss": 2.4688,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019977191201979772,
+      "loss": 2.6055,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00199767186768436,
+      "loss": 2.5469,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019976241312861438,
+      "loss": 2.6895,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997575911026481,
+      "loss": 2.3398,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019975272069287595,
+      "loss": 2.5664,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997478019016601,
+      "loss": 2.3535,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997428347313863,
+      "loss": 2.4883,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019973781918446363,
+      "loss": 2.4199,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019973275526332475,
+      "loss": 2.5195,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019972764297042566,
+      "loss": 2.6016,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00199722482308246,
+      "loss": 2.4316,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997172732792887,
+      "loss": 2.5801,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019971201588608015,
+      "loss": 2.5938,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019970671013117038,
+      "loss": 2.5645,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997013560171327,
+      "loss": 2.5605,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001996959535465639,
+      "loss": 2.4668,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001996905027220843,
+      "loss": 2.5254,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019968500354633763,
+      "loss": 2.5938,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019967945602199096,
+      "loss": 2.4492,
+      "step": 255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019967386015173503,
+      "loss": 2.4727,
+      "step": 256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019966821593828394,
+      "loss": 2.4727,
+      "step": 257
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001996625233843751,
+      "loss": 2.5742,
+      "step": 258
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001996567824927695,
+      "loss": 2.4453,
+      "step": 259
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019965099326625163,
+      "loss": 2.5762,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001996451557076293,
+      "loss": 2.6074,
+      "step": 261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019963926981973372,
+      "loss": 2.3574,
+      "step": 262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019963333560541977,
+      "loss": 2.4785,
+      "step": 263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019962735306756556,
+      "loss": 2.459,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019962132220907266,
+      "loss": 2.4121,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019961524303286626,
+      "loss": 2.6016,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019960911554189466,
+      "loss": 2.5781,
+      "step": 267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001996029397391299,
+      "loss": 2.4668,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019959671562756734,
+      "loss": 2.5742,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019959044321022565,
+      "loss": 2.6309,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001995841224901471,
+      "loss": 2.6367,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001995777534703974,
+      "loss": 2.4004,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019957133615406547,
+      "loss": 2.4375,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019956487054426392,
+      "loss": 2.5293,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019955835664412862,
+      "loss": 2.5469,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019955179445681885,
+      "loss": 2.5176,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001995451839855174,
+      "loss": 2.543,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019953852523343046,
+      "loss": 2.5273,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019953181820378764,
+      "loss": 2.5508,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019952506289984185,
+      "loss": 2.5254,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019951825932486957,
+      "loss": 2.5273,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001995114074821706,
+      "loss": 2.5742,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001995045073750682,
+      "loss": 2.3828,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019949755900690903,
+      "loss": 2.5879,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994905623810631,
+      "loss": 2.4844,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019948351750092392,
+      "loss": 2.457,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994764243699083,
+      "loss": 2.6211,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019946928299145656,
+      "loss": 2.4062,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019946209336903233,
+      "loss": 2.3887,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019945485550612266,
+      "loss": 2.4199,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994475694062381,
+      "loss": 2.5254,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994402350729124,
+      "loss": 2.4961,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994328525097029,
+      "loss": 2.6113,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019942542172019014,
+      "loss": 2.584,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019941794270797826,
+      "loss": 2.4609,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019941041547669467,
+      "loss": 2.5195,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994028400299901,
+      "loss": 2.5996,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019939521637153883,
+      "loss": 2.4922,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001993875445050384,
+      "loss": 2.5977,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001993798244342098,
+      "loss": 2.4922,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019937205616279738,
+      "loss": 2.6113,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001993642396945688,
+      "loss": 2.4434,
+      "step": 302
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001993563750333152,
+      "loss": 2.6289,
+      "step": 303
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00199348462182851,
+      "loss": 2.6465,
+      "step": 304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001993405011470141,
+      "loss": 2.5977,
+      "step": 305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001993324919296657,
+      "loss": 2.5352,
+      "step": 306
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019932443453469033,
+      "loss": 2.6445,
+      "step": 307
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00199316328965996,
+      "loss": 2.4805,
+      "step": 308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00199308175227514,
+      "loss": 2.4453,
+      "step": 309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019929997332319904,
+      "loss": 2.6445,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001992917232570291,
+      "loss": 2.3691,
+      "step": 311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019928342503300555,
+      "loss": 2.5986,
+      "step": 312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019927507865515324,
+      "loss": 2.5742,
+      "step": 313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019926668412752024,
+      "loss": 2.5098,
+      "step": 314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019925824145417795,
+      "loss": 2.4688,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019924975063922126,
+      "loss": 2.457,
+      "step": 316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019924121168676833,
+      "loss": 2.5957,
+      "step": 317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019923262460096063,
+      "loss": 2.5176,
+      "step": 318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00199223989385963,
+      "loss": 2.3809,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001992153060459637,
+      "loss": 2.4844,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019920657458517422,
+      "loss": 2.4824,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019919779500782945,
+      "loss": 2.4297,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019918896731818767,
+      "loss": 2.5215,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001991800915205303,
+      "loss": 2.4219,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019917116761916233,
+      "loss": 2.4766,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019916219561841196,
+      "loss": 2.3184,
+      "step": 326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001991531755226307,
+      "loss": 2.4805,
+      "step": 327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001991441073361934,
+      "loss": 2.418,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019913499106349835,
+      "loss": 2.5098,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00199125826708967,
+      "loss": 2.4199,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001991166142770442,
+      "loss": 2.4277,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001991073537721981,
+      "loss": 2.3965,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019909804519892015,
+      "loss": 2.5332,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019908868856172517,
+      "loss": 2.5332,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019907928386515127,
+      "loss": 2.5391,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019906983111375983,
+      "loss": 2.6035,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001990603303121356,
+      "loss": 2.5312,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001990507814648865,
+      "loss": 2.4805,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00199041184576644,
+      "loss": 2.4336,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019903153965206265,
+      "loss": 2.5703,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001990218466958203,
+      "loss": 2.5137,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001990121057126183,
+      "loss": 2.5,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001990023167071811,
+      "loss": 2.4473,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019899247968425647,
+      "loss": 2.5684,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019898259464861557,
+      "loss": 2.5039,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019897266160505273,
+      "loss": 2.3535,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001989626805583856,
+      "loss": 2.4766,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019895265151345515,
+      "loss": 2.416,
+      "step": 348
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019894257447512563,
+      "loss": 2.6348,
+      "step": 349
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019893244944828446,
+      "loss": 2.5898,
+      "step": 350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019892227643784257,
+      "loss": 2.3496,
+      "step": 351
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019891205544873385,
+      "loss": 2.2715,
+      "step": 352
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001989017864859157,
+      "loss": 2.3633,
+      "step": 353
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001988914695543687,
+      "loss": 2.3984,
+      "step": 354
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019888110465909664,
+      "loss": 2.4492,
+      "step": 355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019887069180512676,
+      "loss": 2.4609,
+      "step": 356
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019886023099750932,
+      "loss": 2.4355,
+      "step": 357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019884972224131804,
+      "loss": 2.5234,
+      "step": 358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001988391655416498,
+      "loss": 2.2861,
+      "step": 359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001988285609036247,
+      "loss": 2.4336,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019881790833238613,
+      "loss": 2.4062,
+      "step": 361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019880720783310084,
+      "loss": 2.4395,
+      "step": 362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001987964594109586,
+      "loss": 2.3906,
+      "step": 363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019878566307117256,
+      "loss": 2.4883,
+      "step": 364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019877481881897916,
+      "loss": 2.4277,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019876392665963793,
+      "loss": 2.3691,
+      "step": 366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019875298659843177,
+      "loss": 2.3516,
+      "step": 367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019874199864066675,
+      "loss": 2.5273,
+      "step": 368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001987309627916722,
+      "loss": 2.5156,
+      "step": 369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019871987905680064,
+      "loss": 2.6074,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001987087474414278,
+      "loss": 2.418,
+      "step": 371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001986975679509527,
+      "loss": 2.3066,
+      "step": 372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001986863405907976,
+      "loss": 2.4648,
+      "step": 373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001986750653664078,
+      "loss": 2.5234,
+      "step": 374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019866374228325206,
+      "loss": 2.418,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019865237134682216,
+      "loss": 2.3535,
+      "step": 376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019864095256263317,
+      "loss": 2.5586,
+      "step": 377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019862948593622337,
+      "loss": 2.3809,
+      "step": 378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019861797147315424,
+      "loss": 2.4141,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019860640917901047,
+      "loss": 2.4961,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001985947990593999,
+      "loss": 2.5234,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019858314111995353,
+      "loss": 2.5332,
+      "step": 382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019857143536632575,
+      "loss": 2.4395,
+      "step": 383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00198559681804194,
+      "loss": 2.4141,
+      "step": 384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019854788043925883,
+      "loss": 2.375,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001985360312772441,
+      "loss": 2.459,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019852413432389685,
+      "loss": 2.4492,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019851218958498724,
+      "loss": 2.3477,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019850019706630865,
+      "loss": 2.5117,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019848815677367755,
+      "loss": 2.4121,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019847606871293377,
+      "loss": 2.5059,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019846393288994006,
+      "loss": 2.5,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019845174931058256,
+      "loss": 2.5117,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019843951798077042,
+      "loss": 2.3652,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00198427238906436,
+      "loss": 2.2344,
+      "step": 395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019841491209353487,
+      "loss": 2.4316,
+      "step": 396
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019840253754804566,
+      "loss": 2.46,
+      "step": 397
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019839011527597017,
+      "loss": 2.377,
+      "step": 398
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019837764528333342,
+      "loss": 2.418,
+      "step": 399
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019836512757618356,
+      "loss": 2.4004,
+      "step": 400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019835256216059177,
+      "loss": 2.4062,
+      "step": 401
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019833994904265248,
+      "loss": 2.6426,
+      "step": 402
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019832728822848328,
+      "loss": 2.6191,
+      "step": 403
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019831457972422474,
+      "loss": 2.4062,
+      "step": 404
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001983018235360407,
+      "loss": 2.4883,
+      "step": 405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019828901967011816,
+      "loss": 2.4648,
+      "step": 406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001982761681326671,
+      "loss": 2.418,
+      "step": 407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019826326892992065,
+      "loss": 2.4277,
+      "step": 408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019825032206813517,
+      "loss": 2.4844,
+      "step": 409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001982373275535901,
+      "loss": 2.2812,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019822428539258787,
+      "loss": 2.5137,
+      "step": 411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019821119559145415,
+      "loss": 2.4199,
+      "step": 412
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019819805815653765,
+      "loss": 2.2734,
+      "step": 413
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001981848730942103,
+      "loss": 2.4922,
+      "step": 414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001981716404108669,
+      "loss": 2.5879,
+      "step": 415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001981583601129255,
+      "loss": 2.4648,
+      "step": 416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019814503220682736,
+      "loss": 2.502,
+      "step": 417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001981316566990366,
+      "loss": 2.3672,
+      "step": 418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019811823359604055,
+      "loss": 2.4434,
+      "step": 419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019810476290434953,
+      "loss": 2.3516,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019809124463049713,
+      "loss": 2.4805,
+      "step": 421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001980776787810398,
+      "loss": 2.29,
+      "step": 422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001980640653625572,
+      "loss": 2.3965,
+      "step": 423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019805040438165204,
+      "loss": 2.5508,
+      "step": 424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019803669584495007,
+      "loss": 2.415,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019802293975910014,
+      "loss": 2.5254,
+      "step": 426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019800913613077413,
+      "loss": 2.4805,
+      "step": 427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019799528496666696,
+      "loss": 2.4668,
+      "step": 428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019798138627349663,
+      "loss": 2.2988,
+      "step": 429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019796744005800425,
+      "loss": 2.4512,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019795344632695393,
+      "loss": 2.4375,
+      "step": 431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019793940508713273,
+      "loss": 2.584,
+      "step": 432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019792531634535097,
+      "loss": 2.3438,
+      "step": 433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001979111801084418,
+      "loss": 2.4863,
+      "step": 434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001978969963832615,
+      "loss": 2.2344,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001978827651766894,
+      "loss": 2.4434,
+      "step": 436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001978684864956278,
+      "loss": 2.4453,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001978541603470021,
+      "loss": 2.4531,
+      "step": 438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019783978673776066,
+      "loss": 2.4102,
+      "step": 439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019782536567487484,
+      "loss": 2.4023,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019781089716533907,
+      "loss": 2.4062,
+      "step": 441
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019779638121617082,
+      "loss": 2.4131,
+      "step": 442
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001977818178344105,
+      "loss": 2.4912,
+      "step": 443
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019776720702712153,
+      "loss": 2.3301,
+      "step": 444
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019775254880139035,
+      "loss": 2.4961,
+      "step": 445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019773784316432647,
+      "loss": 2.2988,
+      "step": 446
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019772309012306226,
+      "loss": 2.3926,
+      "step": 447
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019770828968475314,
+      "loss": 2.3496,
+      "step": 448
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001976934418565776,
+      "loss": 2.3828,
+      "step": 449
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019767854664573693,
+      "loss": 2.4951,
+      "step": 450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019766360405945563,
+      "loss": 2.5176,
+      "step": 451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00197648614104981,
+      "loss": 2.4414,
+      "step": 452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001976335767895834,
+      "loss": 2.2861,
+      "step": 453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019761849212055604,
+      "loss": 2.374,
+      "step": 454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019760336010521532,
+      "loss": 2.2363,
+      "step": 455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001975881807509004,
+      "loss": 2.3027,
+      "step": 456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019757295406497353,
+      "loss": 2.4043,
+      "step": 457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001975576800548198,
+      "loss": 2.418,
+      "step": 458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001975423587278474,
+      "loss": 2.4385,
+      "step": 459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001975269900914873,
+      "loss": 2.3213,
+      "step": 460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001975115741531935,
+      "loss": 2.4629,
+      "step": 461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019749611092044306,
+      "loss": 2.3027,
+      "step": 462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019748060040073573,
+      "loss": 2.4375,
+      "step": 463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019746504260159435,
+      "loss": 2.4102,
+      "step": 464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001974494375305647,
+      "loss": 2.3516,
+      "step": 465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019743378519521543,
+      "loss": 2.3496,
+      "step": 466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001974180856031382,
+      "loss": 2.4414,
+      "step": 467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001974023387619475,
+      "loss": 2.4238,
+      "step": 468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019738654467928063,
+      "loss": 2.5,
+      "step": 469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019737070336279814,
+      "loss": 2.2861,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019735481482018318,
+      "loss": 2.4492,
+      "step": 471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001973388790591419,
+      "loss": 2.2988,
+      "step": 472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019732289608740345,
+      "loss": 2.4277,
+      "step": 473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001973068659127197,
+      "loss": 2.498,
+      "step": 474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019729078854286556,
+      "loss": 2.4941,
+      "step": 475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019727466398563875,
+      "loss": 2.3594,
+      "step": 476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001972584922488599,
+      "loss": 2.3809,
+      "step": 477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019724227334037255,
+      "loss": 2.373,
+      "step": 478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019722600726804307,
+      "loss": 2.4902,
+      "step": 479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019720969403976074,
+      "loss": 2.3467,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001971933336634377,
+      "loss": 2.5098,
+      "step": 481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00197176926147009,
+      "loss": 2.3848,
+      "step": 482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001971604714984324,
+      "loss": 2.5312,
+      "step": 483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001971439697256887,
+      "loss": 2.3438,
+      "step": 484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001971274208367815,
+      "loss": 2.3164,
+      "step": 485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019711082483973726,
+      "loss": 2.3926,
+      "step": 486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001970941817426052,
+      "loss": 2.4551,
+      "step": 487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001970774915534575,
+      "loss": 2.4424,
+      "step": 488
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001970607542803891,
+      "loss": 2.4473,
+      "step": 489
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019704396993151787,
+      "loss": 2.4141,
+      "step": 490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019702713851498434,
+      "loss": 2.541,
+      "step": 491
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019701026003895207,
+      "loss": 2.457,
+      "step": 492
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019699333451160734,
+      "loss": 2.3047,
+      "step": 493
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001969763619411592,
+      "loss": 2.3486,
+      "step": 494
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001969593423358397,
+      "loss": 2.5117,
+      "step": 495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001969422757039035,
+      "loss": 2.3535,
+      "step": 496
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001969251620536281,
+      "loss": 2.4824,
+      "step": 497
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00196908001393314,
+      "loss": 2.4258,
+      "step": 498
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001968907937312842,
+      "loss": 2.3359,
+      "step": 499
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001968735390758848,
+      "loss": 2.5312,
+      "step": 500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001968562374354845,
+      "loss": 2.4766,
+      "step": 501
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019683888881847473,
+      "loss": 2.3008,
+      "step": 502
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001968214932332699,
+      "loss": 2.2949,
+      "step": 503
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019680405068830714,
+      "loss": 2.3271,
+      "step": 504
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001967865611920463,
+      "loss": 2.249,
+      "step": 505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019676902475296997,
+      "loss": 2.2363,
+      "step": 506
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019675144137958358,
+      "loss": 2.4961,
+      "step": 507
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019673381108041535,
+      "loss": 2.3145,
+      "step": 508
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001967161338640162,
+      "loss": 2.3301,
+      "step": 509
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019669840973895987,
+      "loss": 2.3008,
+      "step": 510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001966806387138427,
+      "loss": 2.4453,
+      "step": 511
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019666282079728397,
+      "loss": 2.3574,
+      "step": 512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001966449559979256,
+      "loss": 2.4297,
+      "step": 513
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001966270443244322,
+      "loss": 2.3828,
+      "step": 514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019660908578549123,
+      "loss": 2.4277,
+      "step": 515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019659108038981286,
+      "loss": 2.4082,
+      "step": 516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001965730281461299,
+      "loss": 2.3594,
+      "step": 517
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019655492906319794,
+      "loss": 2.4473,
+      "step": 518
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019653678314979534,
+      "loss": 2.2402,
+      "step": 519
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001965185904147231,
+      "loss": 2.4688,
+      "step": 520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001965003508668049,
+      "loss": 2.4082,
+      "step": 521
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019648206451488718,
+      "loss": 2.2334,
+      "step": 522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001964637313678391,
+      "loss": 2.4297,
+      "step": 523
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019644535143455246,
+      "loss": 2.4707,
+      "step": 524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019642692472394184,
+      "loss": 2.2227,
+      "step": 525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001964084512449444,
+      "loss": 2.3496,
+      "step": 526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019638993100652004,
+      "loss": 2.3223,
+      "step": 527
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001963713640176513,
+      "loss": 2.3613,
+      "step": 528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001963527502873435,
+      "loss": 2.4609,
+      "step": 529
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019633408982462452,
+      "loss": 2.25,
+      "step": 530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019631538263854493,
+      "loss": 2.5234,
+      "step": 531
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019629662873817794,
+      "loss": 2.375,
+      "step": 532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001962778281326195,
+      "loss": 2.3809,
+      "step": 533
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019625898083098815,
+      "loss": 2.4482,
+      "step": 534
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019624008684242513,
+      "loss": 2.4961,
+      "step": 535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001962211461760942,
+      "loss": 2.3379,
+      "step": 536
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019620215884118193,
+      "loss": 2.4102,
+      "step": 537
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019618312484689735,
+      "loss": 2.4043,
+      "step": 538
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001961640442024723,
+      "loss": 2.3086,
+      "step": 539
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001961449169171611,
+      "loss": 2.4219,
+      "step": 540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001961257430002408,
+      "loss": 2.3525,
+      "step": 541
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001961065224610109,
+      "loss": 2.4297,
+      "step": 542
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019608725530879377,
+      "loss": 2.3086,
+      "step": 543
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001960679415529341,
+      "loss": 2.377,
+      "step": 544
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001960485812027995,
+      "loss": 2.25,
+      "step": 545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001960291742677798,
+      "loss": 2.2559,
+      "step": 546
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001960097207572878,
+      "loss": 2.3867,
+      "step": 547
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019599022068075866,
+      "loss": 2.3369,
+      "step": 548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001959706740476502,
+      "loss": 2.4609,
+      "step": 549
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019595108086744276,
+      "loss": 2.4219,
+      "step": 550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019593144114963936,
+      "loss": 2.4062,
+      "step": 551
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001959117549037655,
+      "loss": 2.2324,
+      "step": 552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001958920221393693,
+      "loss": 2.2217,
+      "step": 553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001958722428660214,
+      "loss": 2.4668,
+      "step": 554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019585241709331506,
+      "loss": 2.4814,
+      "step": 555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00195832544830866,
+      "loss": 2.4492,
+      "step": 556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019581262608831256,
+      "loss": 2.3809,
+      "step": 557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001957926608753156,
+      "loss": 2.4375,
+      "step": 558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019577264920155853,
+      "loss": 2.4102,
+      "step": 559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019575259107674735,
+      "loss": 2.5332,
+      "step": 560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001957324865106104,
+      "loss": 2.2227,
+      "step": 561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019571233551289877,
+      "loss": 2.2158,
+      "step": 562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019569213809338595,
+      "loss": 2.2969,
+      "step": 563
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019567189426186794,
+      "loss": 2.3867,
+      "step": 564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001956516040281633,
+      "loss": 2.4258,
+      "step": 565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019563126740211306,
+      "loss": 2.4512,
+      "step": 566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019561088439358078,
+      "loss": 2.3418,
+      "step": 567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001955904550124525,
+      "loss": 2.4014,
+      "step": 568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019556997926863675,
+      "loss": 2.3145,
+      "step": 569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001955494571720645,
+      "loss": 2.4902,
+      "step": 570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019552888873268933,
+      "loss": 2.2832,
+      "step": 571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019550827396048716,
+      "loss": 2.2773,
+      "step": 572
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019548761286545644,
+      "loss": 2.3887,
+      "step": 573
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019546690545761813,
+      "loss": 2.3262,
+      "step": 574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001954461517470156,
+      "loss": 2.5469,
+      "step": 575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001954253517437146,
+      "loss": 2.293,
+      "step": 576
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019540450545780354,
+      "loss": 2.6523,
+      "step": 577
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019538361289939306,
+      "loss": 2.4258,
+      "step": 578
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019536267407861637,
+      "loss": 2.4131,
+      "step": 579
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001953416890056291,
+      "loss": 2.3945,
+      "step": 580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001953206576906093,
+      "loss": 2.3672,
+      "step": 581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019529958014375746,
+      "loss": 2.3633,
+      "step": 582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019527845637529644,
+      "loss": 2.4531,
+      "step": 583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019525728639547158,
+      "loss": 2.4043,
+      "step": 584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019523607021455062,
+      "loss": 2.4785,
+      "step": 585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019521480784282371,
+      "loss": 2.3906,
+      "step": 586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019519349929060334,
+      "loss": 2.3281,
+      "step": 587
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001951721445682245,
+      "loss": 2.5391,
+      "step": 588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001951507436860445,
+      "loss": 2.4512,
+      "step": 589
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019512929665444307,
+      "loss": 2.5293,
+      "step": 590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019510780348382234,
+      "loss": 2.5352,
+      "step": 591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019508626418460679,
+      "loss": 2.416,
+      "step": 592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019506467876724322,
+      "loss": 2.2969,
+      "step": 593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001950430472422009,
+      "loss": 2.375,
+      "step": 594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001950213696199714,
+      "loss": 2.4707,
+      "step": 595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019499964591106872,
+      "loss": 2.4336,
+      "step": 596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001949778761260291,
+      "loss": 2.2754,
+      "step": 597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001949560602754112,
+      "loss": 2.4102,
+      "step": 598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00194934198369796,
+      "loss": 2.2383,
+      "step": 599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019491229041978682,
+      "loss": 2.3994,
+      "step": 600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019489033643600938,
+      "loss": 2.2422,
+      "step": 601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019486833642911154,
+      "loss": 2.2822,
+      "step": 602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019484629040976374,
+      "loss": 2.3828,
+      "step": 603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019482419838865856,
+      "loss": 2.3887,
+      "step": 604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019480206037651086,
+      "loss": 2.3398,
+      "step": 605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019477987638405801,
+      "loss": 2.4102,
+      "step": 606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019475764642205944,
+      "loss": 2.4043,
+      "step": 607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019473537050129703,
+      "loss": 2.1895,
+      "step": 608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019471304863257495,
+      "loss": 2.5312,
+      "step": 609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019469068082671953,
+      "loss": 2.3838,
+      "step": 610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019466826709457954,
+      "loss": 2.3203,
+      "step": 611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019464580744702588,
+      "loss": 2.2236,
+      "step": 612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019462330189495185,
+      "loss": 2.3613,
+      "step": 613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019460075044927293,
+      "loss": 2.3584,
+      "step": 614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019457815312092686,
+      "loss": 2.4824,
+      "step": 615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001945555099208737,
+      "loss": 2.3809,
+      "step": 616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019453282086009565,
+      "loss": 2.2568,
+      "step": 617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019451008594959728,
+      "loss": 2.3965,
+      "step": 618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019448730520040534,
+      "loss": 2.5898,
+      "step": 619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019446447862356876,
+      "loss": 2.3652,
+      "step": 620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019444160623015873,
+      "loss": 2.3613,
+      "step": 621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019441868803126874,
+      "loss": 2.4258,
+      "step": 622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019439572403801438,
+      "loss": 2.3145,
+      "step": 623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019437271426153353,
+      "loss": 2.4473,
+      "step": 624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019434965871298624,
+      "loss": 2.375,
+      "step": 625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019432655740355478,
+      "loss": 2.3691,
+      "step": 626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001943034103444436,
+      "loss": 2.4668,
+      "step": 627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019428021754687931,
+      "loss": 2.4648,
+      "step": 628
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019425697902211078,
+      "loss": 2.2363,
+      "step": 629
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00194233694781409,
+      "loss": 2.3301,
+      "step": 630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019421036483606713,
+      "loss": 2.3574,
+      "step": 631
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019418698919740054,
+      "loss": 2.4434,
+      "step": 632
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019416356787674673,
+      "loss": 2.3496,
+      "step": 633
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019414010088546535,
+      "loss": 2.3145,
+      "step": 634
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019411658823493823,
+      "loss": 2.3857,
+      "step": 635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019409302993656933,
+      "loss": 2.4082,
+      "step": 636
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019406942600178473,
+      "loss": 2.4062,
+      "step": 637
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019404577644203268,
+      "loss": 2.4355,
+      "step": 638
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019402208126878353,
+      "loss": 2.4434,
+      "step": 639
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019399834049352977,
+      "loss": 2.4014,
+      "step": 640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00193974554127786,
+      "loss": 2.3438,
+      "step": 641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019395072218308898,
+      "loss": 2.3359,
+      "step": 642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019392684467099746,
+      "loss": 2.3809,
+      "step": 643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019390292160309242,
+      "loss": 2.4014,
+      "step": 644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019387895299097684,
+      "loss": 2.4219,
+      "step": 645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019385493884627584,
+      "loss": 2.3965,
+      "step": 646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001938308791806366,
+      "loss": 2.3633,
+      "step": 647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019380677400572842,
+      "loss": 2.2773,
+      "step": 648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019378262333324262,
+      "loss": 2.3555,
+      "step": 649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019375842717489259,
+      "loss": 2.3711,
+      "step": 650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019373418554241382,
+      "loss": 2.3262,
+      "step": 651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019370989844756385,
+      "loss": 2.248,
+      "step": 652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019368556590212225,
+      "loss": 2.3047,
+      "step": 653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019366118791789063,
+      "loss": 2.3281,
+      "step": 654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019363676450669263,
+      "loss": 2.3555,
+      "step": 655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019361229568037397,
+      "loss": 2.2949,
+      "step": 656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019358778145080237,
+      "loss": 2.3457,
+      "step": 657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019356322182986754,
+      "loss": 2.3262,
+      "step": 658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019353861682948126,
+      "loss": 2.3125,
+      "step": 659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019351396646157728,
+      "loss": 2.3574,
+      "step": 660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019348927073811139,
+      "loss": 2.3711,
+      "step": 661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019346452967106134,
+      "loss": 2.3213,
+      "step": 662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019343974327242686,
+      "loss": 2.4121,
+      "step": 663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019341491155422978,
+      "loss": 2.3672,
+      "step": 664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019339003452851372,
+      "loss": 2.4092,
+      "step": 665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019336511220734444,
+      "loss": 2.4238,
+      "step": 666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001933401446028096,
+      "loss": 2.4482,
+      "step": 667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019331513172701883,
+      "loss": 2.3398,
+      "step": 668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019329007359210372,
+      "loss": 2.3047,
+      "step": 669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001932649702102178,
+      "loss": 2.3027,
+      "step": 670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019323982159353656,
+      "loss": 2.4355,
+      "step": 671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019321462775425747,
+      "loss": 2.2383,
+      "step": 672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019318938870459984,
+      "loss": 2.3027,
+      "step": 673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019316410445680498,
+      "loss": 2.3496,
+      "step": 674
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019313877502313604,
+      "loss": 2.3164,
+      "step": 675
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001931134004158782,
+      "loss": 2.293,
+      "step": 676
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001930879806473385,
+      "loss": 2.3809,
+      "step": 677
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019306251572984588,
+      "loss": 2.375,
+      "step": 678
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001930370056757511,
+      "loss": 2.2676,
+      "step": 679
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00193011450497427,
+      "loss": 2.3984,
+      "step": 680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019298585020726814,
+      "loss": 2.3477,
+      "step": 681
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019296020481769098,
+      "loss": 2.4004,
+      "step": 682
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019293451434113393,
+      "loss": 2.4648,
+      "step": 683
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019290877879005724,
+      "loss": 2.375,
+      "step": 684
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019288299817694296,
+      "loss": 2.4824,
+      "step": 685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019285717251429507,
+      "loss": 2.4082,
+      "step": 686
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019283130181463935,
+      "loss": 2.2852,
+      "step": 687
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001928053860905235,
+      "loss": 2.4297,
+      "step": 688
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019277942535451693,
+      "loss": 2.3223,
+      "step": 689
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019275341961921097,
+      "loss": 2.4062,
+      "step": 690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019272736889721883,
+      "loss": 2.4805,
+      "step": 691
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001927012732011754,
+      "loss": 2.3535,
+      "step": 692
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019267513254373747,
+      "loss": 2.3184,
+      "step": 693
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019264894693758361,
+      "loss": 2.46,
+      "step": 694
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019262271639541421,
+      "loss": 2.4258,
+      "step": 695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019259644092995147,
+      "loss": 2.293,
+      "step": 696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019257012055393932,
+      "loss": 2.3418,
+      "step": 697
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001925437552801435,
+      "loss": 2.3301,
+      "step": 698
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019251734512135154,
+      "loss": 2.3467,
+      "step": 699
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019249089009037272,
+      "loss": 2.3301,
+      "step": 700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019246439020003813,
+      "loss": 2.3223,
+      "step": 701
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019243784546320055,
+      "loss": 2.3379,
+      "step": 702
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019241125589273456,
+      "loss": 2.2695,
+      "step": 703
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019238462150153646,
+      "loss": 2.4668,
+      "step": 704
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001923579423025243,
+      "loss": 2.3887,
+      "step": 705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019233121830863782,
+      "loss": 2.2354,
+      "step": 706
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001923044495328386,
+      "loss": 2.3965,
+      "step": 707
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019227763598810982,
+      "loss": 2.4629,
+      "step": 708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019225077768745642,
+      "loss": 2.3818,
+      "step": 709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019222387464390507,
+      "loss": 2.3281,
+      "step": 710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019219692687050408,
+      "loss": 2.2402,
+      "step": 711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001921699343803235,
+      "loss": 2.4385,
+      "step": 712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019214289718645508,
+      "loss": 2.207,
+      "step": 713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001921158153020122,
+      "loss": 2.4512,
+      "step": 714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019208868874013,
+      "loss": 2.4336,
+      "step": 715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001920615175139652,
+      "loss": 2.3154,
+      "step": 716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001920343016366962,
+      "loss": 2.3486,
+      "step": 717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019200704112152308,
+      "loss": 2.293,
+      "step": 718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001919797359816676,
+      "loss": 2.3438,
+      "step": 719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019195238623037315,
+      "loss": 2.2734,
+      "step": 720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019192499188090465,
+      "loss": 2.3398,
+      "step": 721
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001918975529465488,
+      "loss": 2.2969,
+      "step": 722
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019187006944061385,
+      "loss": 2.4277,
+      "step": 723
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019184254137642968,
+      "loss": 2.3145,
+      "step": 724
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019181496876734777,
+      "loss": 2.2842,
+      "step": 725
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019178735162674121,
+      "loss": 2.4434,
+      "step": 726
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001917596899680047,
+      "loss": 2.3369,
+      "step": 727
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019173198380455455,
+      "loss": 2.2773,
+      "step": 728
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019170423314982862,
+      "loss": 2.2383,
+      "step": 729
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001916764380172863,
+      "loss": 2.2979,
+      "step": 730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019164859842040873,
+      "loss": 2.3867,
+      "step": 731
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019162071437269837,
+      "loss": 2.3887,
+      "step": 732
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019159278588767945,
+      "loss": 2.2725,
+      "step": 733
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019156481297889765,
+      "loss": 2.3213,
+      "step": 734
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019153679565992025,
+      "loss": 2.3262,
+      "step": 735
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019150873394433595,
+      "loss": 2.373,
+      "step": 736
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019148062784575513,
+      "loss": 2.3984,
+      "step": 737
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019145247737780962,
+      "loss": 2.3447,
+      "step": 738
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019142428255415277,
+      "loss": 2.5039,
+      "step": 739
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019139604338845953,
+      "loss": 2.3086,
+      "step": 740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001913677598944262,
+      "loss": 2.3809,
+      "step": 741
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019133943208577068,
+      "loss": 2.248,
+      "step": 742
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001913110599762324,
+      "loss": 2.4004,
+      "step": 743
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019128264357957212,
+      "loss": 2.4766,
+      "step": 744
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001912541829095723,
+      "loss": 2.3125,
+      "step": 745
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001912256779800367,
+      "loss": 2.3652,
+      "step": 746
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001911971288047906,
+      "loss": 2.3418,
+      "step": 747
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019116853539768077,
+      "loss": 2.457,
+      "step": 748
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019113989777257538,
+      "loss": 2.502,
+      "step": 749
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019111121594336407,
+      "loss": 2.3457,
+      "step": 750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019108248992395795,
+      "loss": 2.3906,
+      "step": 751
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001910537197282895,
+      "loss": 2.332,
+      "step": 752
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019102490537031272,
+      "loss": 2.3223,
+      "step": 753
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001909960468640029,
+      "loss": 2.3516,
+      "step": 754
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019096714422335683,
+      "loss": 2.2715,
+      "step": 755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019093819746239273,
+      "loss": 2.3418,
+      "step": 756
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019090920659515015,
+      "loss": 2.4648,
+      "step": 757
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019088017163569006,
+      "loss": 2.4805,
+      "step": 758
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019085109259809481,
+      "loss": 2.4629,
+      "step": 759
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019082196949646816,
+      "loss": 2.3359,
+      "step": 760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019079280234493518,
+      "loss": 2.1777,
+      "step": 761
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019076359115764235,
+      "loss": 2.4824,
+      "step": 762
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001907343359487575,
+      "loss": 2.3496,
+      "step": 763
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019070503673246982,
+      "loss": 2.3301,
+      "step": 764
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019067569352298984,
+      "loss": 2.3281,
+      "step": 765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019064630633454937,
+      "loss": 2.4473,
+      "step": 766
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001906168751814017,
+      "loss": 2.3164,
+      "step": 767
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019058740007782125,
+      "loss": 2.4062,
+      "step": 768
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019055788103810388,
+      "loss": 2.2803,
+      "step": 769
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019052831807656674,
+      "loss": 2.377,
+      "step": 770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019049871120754827,
+      "loss": 2.4189,
+      "step": 771
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001904690604454082,
+      "loss": 2.332,
+      "step": 772
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001904393658045276,
+      "loss": 2.5625,
+      "step": 773
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001904096272993087,
+      "loss": 2.4531,
+      "step": 774
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019037984494417516,
+      "loss": 2.3867,
+      "step": 775
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001903500187535718,
+      "loss": 2.3613,
+      "step": 776
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019032014874196476,
+      "loss": 2.1582,
+      "step": 777
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019029023492384135,
+      "loss": 2.4492,
+      "step": 778
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019026027731371023,
+      "loss": 2.3281,
+      "step": 779
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019023027592610124,
+      "loss": 2.377,
+      "step": 780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019020023077556548,
+      "loss": 2.2266,
+      "step": 781
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001901701418766752,
+      "loss": 2.2969,
+      "step": 782
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019014000924402403,
+      "loss": 2.293,
+      "step": 783
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019010983289222665,
+      "loss": 2.3398,
+      "step": 784
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019007961283591904,
+      "loss": 2.5215,
+      "step": 785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001900493490897583,
+      "loss": 2.4004,
+      "step": 786
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019001904166842279,
+      "loss": 2.459,
+      "step": 787
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018998869058661203,
+      "loss": 2.3027,
+      "step": 788
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018995829585904667,
+      "loss": 2.4492,
+      "step": 789
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018992785750046863,
+      "loss": 2.3398,
+      "step": 790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018989737552564093,
+      "loss": 2.3027,
+      "step": 791
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001898668499493477,
+      "loss": 2.4082,
+      "step": 792
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018983628078639433,
+      "loss": 2.4746,
+      "step": 793
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001898056680516072,
+      "loss": 2.3516,
+      "step": 794
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00189775011759834,
+      "loss": 2.3486,
+      "step": 795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018974431192594339,
+      "loss": 2.3105,
+      "step": 796
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018971356856482525,
+      "loss": 2.373,
+      "step": 797
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001896827816913905,
+      "loss": 2.4141,
+      "step": 798
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001896519513205712,
+      "loss": 2.2607,
+      "step": 799
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018962107746732053,
+      "loss": 2.3359,
+      "step": 800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018959016014661275,
+      "loss": 2.2617,
+      "step": 801
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018955919937344312,
+      "loss": 2.3486,
+      "step": 802
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001895281951628281,
+      "loss": 2.3516,
+      "step": 803
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018949714752980514,
+      "loss": 2.3066,
+      "step": 804
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018946605648943273,
+      "loss": 2.4766,
+      "step": 805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018943492205679053,
+      "loss": 2.251,
+      "step": 806
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018940374424697914,
+      "loss": 2.3984,
+      "step": 807
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001893725230751202,
+      "loss": 2.2686,
+      "step": 808
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018934125855635644,
+      "loss": 2.3984,
+      "step": 809
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018930995070585157,
+      "loss": 2.3555,
+      "step": 810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001892785995387903,
+      "loss": 2.2383,
+      "step": 811
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018924720507037846,
+      "loss": 2.3389,
+      "step": 812
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001892157673158427,
+      "loss": 2.3242,
+      "step": 813
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001891842862904308,
+      "loss": 2.3047,
+      "step": 814
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018915276200941154,
+      "loss": 2.4531,
+      "step": 815
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001891211944880746,
+      "loss": 2.3594,
+      "step": 816
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001890895837417306,
+      "loss": 2.2119,
+      "step": 817
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018905792978571125,
+      "loss": 2.375,
+      "step": 818
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018902623263536919,
+      "loss": 2.3555,
+      "step": 819
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018899449230607788,
+      "loss": 2.2666,
+      "step": 820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001889627088132319,
+      "loss": 2.2393,
+      "step": 821
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018893088217224662,
+      "loss": 2.416,
+      "step": 822
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018889901239855847,
+      "loss": 2.2979,
+      "step": 823
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018886709950762467,
+      "loss": 2.3262,
+      "step": 824
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001888351435149234,
+      "loss": 2.293,
+      "step": 825
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018880314443595384,
+      "loss": 2.3008,
+      "step": 826
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001887711022862359,
+      "loss": 2.4346,
+      "step": 827
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018873901708131048,
+      "loss": 2.2949,
+      "step": 828
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018870688883673937,
+      "loss": 2.2334,
+      "step": 829
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018867471756810522,
+      "loss": 2.2715,
+      "step": 830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018864250329101151,
+      "loss": 2.374,
+      "step": 831
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001886102460210826,
+      "loss": 2.373,
+      "step": 832
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018857794577396376,
+      "loss": 2.334,
+      "step": 833
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018854560256532098,
+      "loss": 2.3809,
+      "step": 834
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018851321641084125,
+      "loss": 2.2686,
+      "step": 835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018848078732623223,
+      "loss": 2.3105,
+      "step": 836
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018844831532722252,
+      "loss": 2.3799,
+      "step": 837
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018841580042956144,
+      "loss": 2.4258,
+      "step": 838
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018838324264901921,
+      "loss": 2.3105,
+      "step": 839
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001883506420013868,
+      "loss": 2.3809,
+      "step": 840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001883179985024759,
+      "loss": 2.3818,
+      "step": 841
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018828531216811913,
+      "loss": 2.2871,
+      "step": 842
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018825258301416983,
+      "loss": 2.3955,
+      "step": 843
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00188219811056502,
+      "loss": 2.498,
+      "step": 844
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001881869963110106,
+      "loss": 2.2275,
+      "step": 845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018815413879361112,
+      "loss": 2.373,
+      "step": 846
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018812123852024,
+      "loss": 2.3086,
+      "step": 847
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001880882955068543,
+      "loss": 2.3066,
+      "step": 848
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018805530976943182,
+      "loss": 2.3213,
+      "step": 849
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001880222813239711,
+      "loss": 2.3223,
+      "step": 850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001879892101864914,
+      "loss": 2.293,
+      "step": 851
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018795609637303268,
+      "loss": 2.3555,
+      "step": 852
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001879229398996556,
+      "loss": 2.3125,
+      "step": 853
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018788974078244145,
+      "loss": 2.3691,
+      "step": 854
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018785649903749234,
+      "loss": 2.4043,
+      "step": 855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018782321468093094,
+      "loss": 2.502,
+      "step": 856
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001877898877289006,
+      "loss": 2.3184,
+      "step": 857
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018775651819756538,
+      "loss": 2.3047,
+      "step": 858
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018772310610310995,
+      "loss": 2.3691,
+      "step": 859
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018768965146173965,
+      "loss": 2.3027,
+      "step": 860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018765615428968042,
+      "loss": 2.2861,
+      "step": 861
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018762261460317886,
+      "loss": 2.4082,
+      "step": 862
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001875890324185022,
+      "loss": 2.4062,
+      "step": 863
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018755540775193822,
+      "loss": 2.375,
+      "step": 864
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018752174061979542,
+      "loss": 2.3418,
+      "step": 865
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001874880310384027,
+      "loss": 2.457,
+      "step": 866
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001874542790241098,
+      "loss": 2.2773,
+      "step": 867
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018742048459328682,
+      "loss": 2.2793,
+      "step": 868
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001873866477623246,
+      "loss": 2.375,
+      "step": 869
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001873527685476344,
+      "loss": 2.3721,
+      "step": 870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001873188469656481,
+      "loss": 2.3477,
+      "step": 871
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018728488303281826,
+      "loss": 2.375,
+      "step": 872
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018725087676561768,
+      "loss": 2.5645,
+      "step": 873
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018721682818054002,
+      "loss": 2.4512,
+      "step": 874
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001871827372940992,
+      "loss": 2.3594,
+      "step": 875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018714860412282984,
+      "loss": 2.2852,
+      "step": 876
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018711442868328697,
+      "loss": 2.3418,
+      "step": 877
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018708021099204616,
+      "loss": 2.2109,
+      "step": 878
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018704595106570343,
+      "loss": 2.3848,
+      "step": 879
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018701164892087534,
+      "loss": 2.3652,
+      "step": 880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018697730457419893,
+      "loss": 2.3135,
+      "step": 881
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018694291804233164,
+      "loss": 2.3252,
+      "step": 882
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018690848934195144,
+      "loss": 2.4453,
+      "step": 883
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001868740184897567,
+      "loss": 2.4199,
+      "step": 884
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018683950550246627,
+      "loss": 2.5117,
+      "step": 885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018680495039681939,
+      "loss": 2.4492,
+      "step": 886
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018677035318957582,
+      "loss": 2.2852,
+      "step": 887
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001867357138975157,
+      "loss": 2.4785,
+      "step": 888
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018670103253743946,
+      "loss": 2.416,
+      "step": 889
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018666630912616813,
+      "loss": 2.2217,
+      "step": 890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018663154368054298,
+      "loss": 2.3672,
+      "step": 891
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001865967362174258,
+      "loss": 2.2559,
+      "step": 892
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018656188675369864,
+      "loss": 2.2539,
+      "step": 893
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018652699530626398,
+      "loss": 2.4395,
+      "step": 894
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018649206189204463,
+      "loss": 2.2451,
+      "step": 895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018645708652798383,
+      "loss": 2.293,
+      "step": 896
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018642206923104507,
+      "loss": 2.2539,
+      "step": 897
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018638701001821225,
+      "loss": 2.2773,
+      "step": 898
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018635190890648951,
+      "loss": 2.4492,
+      "step": 899
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018631676591290148,
+      "loss": 2.4258,
+      "step": 900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018628158105449288,
+      "loss": 2.3389,
+      "step": 901
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018624635434832896,
+      "loss": 2.2656,
+      "step": 902
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018621108581149506,
+      "loss": 2.2168,
+      "step": 903
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018617577546109695,
+      "loss": 2.2266,
+      "step": 904
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018614042331426067,
+      "loss": 2.3711,
+      "step": 905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018610502938813244,
+      "loss": 2.418,
+      "step": 906
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018606959369987883,
+      "loss": 2.1934,
+      "step": 907
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018603411626668663,
+      "loss": 2.332,
+      "step": 908
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001859985971057629,
+      "loss": 2.208,
+      "step": 909
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018596303623433488,
+      "loss": 2.4434,
+      "step": 910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018592743366965016,
+      "loss": 2.3613,
+      "step": 911
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018589178942897638,
+      "loss": 2.1494,
+      "step": 912
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018585610352960154,
+      "loss": 2.3262,
+      "step": 913
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018582037598883381,
+      "loss": 2.3262,
+      "step": 914
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001857846068240015,
+      "loss": 2.3242,
+      "step": 915
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018574879605245318,
+      "loss": 2.4414,
+      "step": 916
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018571294369155754,
+      "loss": 2.4473,
+      "step": 917
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018567704975870347,
+      "loss": 2.3018,
+      "step": 918
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018564111427130001,
+      "loss": 2.5352,
+      "step": 919
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018560513724677644,
+      "loss": 2.3125,
+      "step": 920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018556911870258204,
+      "loss": 2.3145,
+      "step": 921
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001855330586561863,
+      "loss": 2.4062,
+      "step": 922
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001854969571250789,
+      "loss": 2.3613,
+      "step": 923
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018546081412676952,
+      "loss": 2.2812,
+      "step": 924
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018542462967878804,
+      "loss": 2.3594,
+      "step": 925
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018538840379868438,
+      "loss": 2.4082,
+      "step": 926
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018535213650402865,
+      "loss": 2.3125,
+      "step": 927
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018531582781241092,
+      "loss": 2.2373,
+      "step": 928
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018527947774144148,
+      "loss": 2.2441,
+      "step": 929
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018524308630875055,
+      "loss": 2.4297,
+      "step": 930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001852066535319885,
+      "loss": 2.3799,
+      "step": 931
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018517017942882576,
+      "loss": 2.1895,
+      "step": 932
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018513366401695275,
+      "loss": 2.291,
+      "step": 933
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001850971073140799,
+      "loss": 2.3184,
+      "step": 934
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001850605093379378,
+      "loss": 2.252,
+      "step": 935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018502387010627687,
+      "loss": 2.377,
+      "step": 936
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018498718963686777,
+      "loss": 2.2803,
+      "step": 937
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018495046794750093,
+      "loss": 2.2021,
+      "step": 938
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001849137050559869,
+      "loss": 2.4082,
+      "step": 939
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018487690098015622,
+      "loss": 2.1709,
+      "step": 940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018484005573785934,
+      "loss": 2.3008,
+      "step": 941
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018480316934696673,
+      "loss": 2.2998,
+      "step": 942
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001847662418253688,
+      "loss": 2.1885,
+      "step": 943
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018472927319097587,
+      "loss": 2.3281,
+      "step": 944
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018469226346171826,
+      "loss": 2.3613,
+      "step": 945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001846552126555462,
+      "loss": 2.3887,
+      "step": 946
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018461812079042987,
+      "loss": 2.2539,
+      "step": 947
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001845809878843593,
+      "loss": 2.418,
+      "step": 948
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018454381395534448,
+      "loss": 2.3613,
+      "step": 949
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018450659902141525,
+      "loss": 2.2549,
+      "step": 950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018446934310062137,
+      "loss": 2.4199,
+      "step": 951
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001844320462110325,
+      "loss": 2.3418,
+      "step": 952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018439470837073816,
+      "loss": 2.3965,
+      "step": 953
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018435732959784766,
+      "loss": 2.1816,
+      "step": 954
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018431990991049025,
+      "loss": 2.3643,
+      "step": 955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00184282449326815,
+      "loss": 2.2598,
+      "step": 956
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001842449478649908,
+      "loss": 2.334,
+      "step": 957
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018420740554320639,
+      "loss": 2.2627,
+      "step": 958
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018416982237967029,
+      "loss": 2.418,
+      "step": 959
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018413219839261085,
+      "loss": 2.4609,
+      "step": 960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018409453360027626,
+      "loss": 2.4121,
+      "step": 961
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018405682802093443,
+      "loss": 2.2295,
+      "step": 962
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018401908167287309,
+      "loss": 2.2129,
+      "step": 963
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018398129457439976,
+      "loss": 2.3418,
+      "step": 964
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018394346674384164,
+      "loss": 2.3906,
+      "step": 965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018390559819954585,
+      "loss": 2.377,
+      "step": 966
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018386768895987907,
+      "loss": 2.252,
+      "step": 967
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018382973904322783,
+      "loss": 2.3066,
+      "step": 968
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018379174846799838,
+      "loss": 2.4531,
+      "step": 969
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018375371725261665,
+      "loss": 2.2363,
+      "step": 970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018371564541552832,
+      "loss": 2.2539,
+      "step": 971
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018367753297519873,
+      "loss": 2.3486,
+      "step": 972
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018363937995011298,
+      "loss": 2.3447,
+      "step": 973
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018360118635877578,
+      "loss": 2.4062,
+      "step": 974
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018356295221971155,
+      "loss": 2.2607,
+      "step": 975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018352467755146438,
+      "loss": 2.4785,
+      "step": 976
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00183486362372598,
+      "loss": 2.3135,
+      "step": 977
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018344800670169585,
+      "loss": 2.4062,
+      "step": 978
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001834096105573609,
+      "loss": 2.2832,
+      "step": 979
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018337117395821583,
+      "loss": 2.4141,
+      "step": 980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001833326969229029,
+      "loss": 2.2637,
+      "step": 981
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018329417947008408,
+      "loss": 2.3418,
+      "step": 982
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018325562161844076,
+      "loss": 2.2393,
+      "step": 983
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018321702338667408,
+      "loss": 2.1924,
+      "step": 984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001831783847935047,
+      "loss": 2.3398,
+      "step": 985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001831397058576729,
+      "loss": 2.332,
+      "step": 986
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018310098659793845,
+      "loss": 2.248,
+      "step": 987
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001830622270330808,
+      "loss": 2.3496,
+      "step": 988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018302342718189877,
+      "loss": 2.3945,
+      "step": 989
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001829845870632109,
+      "loss": 2.3691,
+      "step": 990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018294570669585518,
+      "loss": 2.2725,
+      "step": 991
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018290678609868907,
+      "loss": 2.3418,
+      "step": 992
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018286782529058963,
+      "loss": 2.3574,
+      "step": 993
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001828288242904534,
+      "loss": 2.3887,
+      "step": 994
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001827897831171964,
+      "loss": 2.2246,
+      "step": 995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001827507017897541,
+      "loss": 2.4307,
+      "step": 996
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018271158032708153,
+      "loss": 2.3457,
+      "step": 997
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018267241874815312,
+      "loss": 2.3018,
+      "step": 998
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018263321707196276,
+      "loss": 2.2871,
+      "step": 999
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018259397531752383,
+      "loss": 2.375,
+      "step": 1000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018255469350386907,
+      "loss": 2.2881,
+      "step": 1001
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018251537165005077,
+      "loss": 2.3145,
+      "step": 1002
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018247600977514052,
+      "loss": 2.2031,
+      "step": 1003
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001824366078982294,
+      "loss": 2.418,
+      "step": 1004
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018239716603842783,
+      "loss": 2.2402,
+      "step": 1005
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018235768421486572,
+      "loss": 2.373,
+      "step": 1006
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001823181624466922,
+      "loss": 2.2441,
+      "step": 1007
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00182278600753076,
+      "loss": 2.3008,
+      "step": 1008
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018223899915320496,
+      "loss": 2.3555,
+      "step": 1009
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018219935766628646,
+      "loss": 2.25,
+      "step": 1010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018215967631154718,
+      "loss": 2.3037,
+      "step": 1011
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001821199551082331,
+      "loss": 2.2217,
+      "step": 1012
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018208019407560956,
+      "loss": 2.5488,
+      "step": 1013
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001820403932329612,
+      "loss": 2.2578,
+      "step": 1014
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018200055259959198,
+      "loss": 2.4873,
+      "step": 1015
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018196067219482517,
+      "loss": 2.3223,
+      "step": 1016
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001819207520380033,
+      "loss": 2.375,
+      "step": 1017
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001818807921484882,
+      "loss": 2.4248,
+      "step": 1018
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018184079254566097,
+      "loss": 2.4014,
+      "step": 1019
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018180075324892198,
+      "loss": 2.3086,
+      "step": 1020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018176067427769082,
+      "loss": 2.166,
+      "step": 1021
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018172055565140633,
+      "loss": 2.4297,
+      "step": 1022
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018168039738952664,
+      "loss": 2.2842,
+      "step": 1023
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018164019951152901,
+      "loss": 2.2627,
+      "step": 1024
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018159996203691003,
+      "loss": 2.3809,
+      "step": 1025
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018155968498518537,
+      "loss": 2.4023,
+      "step": 1026
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018151936837588996,
+      "loss": 2.4307,
+      "step": 1027
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018147901222857793,
+      "loss": 2.3066,
+      "step": 1028
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018143861656282256,
+      "loss": 2.334,
+      "step": 1029
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018139818139821632,
+      "loss": 2.541,
+      "step": 1030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018135770675437085,
+      "loss": 2.3105,
+      "step": 1031
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018131719265091681,
+      "loss": 2.291,
+      "step": 1032
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018127663910750419,
+      "loss": 2.2441,
+      "step": 1033
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018123604614380196,
+      "loss": 2.3047,
+      "step": 1034
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001811954137794983,
+      "loss": 2.4121,
+      "step": 1035
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018115474203430047,
+      "loss": 2.2988,
+      "step": 1036
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001811140309279348,
+      "loss": 2.2832,
+      "step": 1037
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018107328048014674,
+      "loss": 2.416,
+      "step": 1038
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018103249071070087,
+      "loss": 2.332,
+      "step": 1039
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018099166163938075,
+      "loss": 2.4355,
+      "step": 1040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00180950793285989,
+      "loss": 2.3506,
+      "step": 1041
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001809098856703474,
+      "loss": 2.2188,
+      "step": 1042
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001808689388122967,
+      "loss": 2.3945,
+      "step": 1043
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018082795273169662,
+      "loss": 2.2793,
+      "step": 1044
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018078692744842608,
+      "loss": 2.2539,
+      "step": 1045
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018074586298238288,
+      "loss": 2.3184,
+      "step": 1046
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018070475935348376,
+      "loss": 2.2598,
+      "step": 1047
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018066361658166464,
+      "loss": 2.1367,
+      "step": 1048
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018062243468688031,
+      "loss": 2.3301,
+      "step": 1049
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018058121368910455,
+      "loss": 2.3086,
+      "step": 1050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001805399536083301,
+      "loss": 2.4707,
+      "step": 1051
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018049865446456872,
+      "loss": 2.3203,
+      "step": 1052
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018045731627785097,
+      "loss": 2.252,
+      "step": 1053
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018041593906822655,
+      "loss": 2.2588,
+      "step": 1054
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018037452285576388,
+      "loss": 2.3691,
+      "step": 1055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018033306766055044,
+      "loss": 2.334,
+      "step": 1056
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018029157350269256,
+      "loss": 2.4375,
+      "step": 1057
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018025004040231553,
+      "loss": 2.3447,
+      "step": 1058
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018020846837956336,
+      "loss": 2.377,
+      "step": 1059
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018016685745459915,
+      "loss": 2.4199,
+      "step": 1060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018012520764760476,
+      "loss": 2.3096,
+      "step": 1061
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018008351897878087,
+      "loss": 2.3535,
+      "step": 1062
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018004179146834712,
+      "loss": 2.4492,
+      "step": 1063
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001800000251365419,
+      "loss": 2.5371,
+      "step": 1064
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017995822000362244,
+      "loss": 2.4023,
+      "step": 1065
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017991637608986487,
+      "loss": 2.375,
+      "step": 1066
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017987449341556397,
+      "loss": 2.2725,
+      "step": 1067
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017983257200103347,
+      "loss": 2.4619,
+      "step": 1068
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017979061186660583,
+      "loss": 2.3008,
+      "step": 1069
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017974861303263227,
+      "loss": 2.3369,
+      "step": 1070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017970657551948284,
+      "loss": 2.2988,
+      "step": 1071
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001796644993475463,
+      "loss": 2.3311,
+      "step": 1072
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017962238453723017,
+      "loss": 2.4014,
+      "step": 1073
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001795802311089607,
+      "loss": 2.2578,
+      "step": 1074
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017953803908318288,
+      "loss": 2.3223,
+      "step": 1075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017949580848036044,
+      "loss": 2.4102,
+      "step": 1076
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017945353932097578,
+      "loss": 2.2178,
+      "step": 1077
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001794112316255301,
+      "loss": 2.2803,
+      "step": 1078
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017936888541454313,
+      "loss": 2.4258,
+      "step": 1079
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017932650070855338,
+      "loss": 2.4473,
+      "step": 1080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017928407752811806,
+      "loss": 2.2803,
+      "step": 1081
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017924161589381296,
+      "loss": 2.1543,
+      "step": 1082
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001791991158262326,
+      "loss": 2.502,
+      "step": 1083
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017915657734599,
+      "loss": 2.2686,
+      "step": 1084
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017911400047371702,
+      "loss": 2.3848,
+      "step": 1085
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017907138523006396,
+      "loss": 2.2832,
+      "step": 1086
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001790287316356998,
+      "loss": 2.2441,
+      "step": 1087
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001789860397113122,
+      "loss": 2.4473,
+      "step": 1088
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017894330947760726,
+      "loss": 2.375,
+      "step": 1089
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017890054095530976,
+      "loss": 2.252,
+      "step": 1090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017885773416516297,
+      "loss": 2.4062,
+      "step": 1091
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017881488912792885,
+      "loss": 2.3154,
+      "step": 1092
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001787720058643878,
+      "loss": 2.2773,
+      "step": 1093
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017872908439533876,
+      "loss": 2.3086,
+      "step": 1094
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017868612474159927,
+      "loss": 2.4941,
+      "step": 1095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017864312692400536,
+      "loss": 2.332,
+      "step": 1096
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017860009096341155,
+      "loss": 2.2734,
+      "step": 1097
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017855701688069088,
+      "loss": 2.4629,
+      "step": 1098
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017851390469673485,
+      "loss": 2.3555,
+      "step": 1099
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017847075443245348,
+      "loss": 2.2061,
+      "step": 1100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017842756610877526,
+      "loss": 2.3564,
+      "step": 1101
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017838433974664712,
+      "loss": 2.2246,
+      "step": 1102
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017834107536703438,
+      "loss": 2.502,
+      "step": 1103
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017829777299092095,
+      "loss": 2.4102,
+      "step": 1104
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017825443263930906,
+      "loss": 2.2754,
+      "step": 1105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001782110543332193,
+      "loss": 2.3926,
+      "step": 1106
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017816763809369084,
+      "loss": 2.3213,
+      "step": 1107
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001781241839417811,
+      "loss": 2.4082,
+      "step": 1108
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017808069189856593,
+      "loss": 2.2793,
+      "step": 1109
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017803716198513961,
+      "loss": 2.3711,
+      "step": 1110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017799359422261474,
+      "loss": 2.3389,
+      "step": 1111
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017794998863212232,
+      "loss": 2.3447,
+      "step": 1112
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017790634523481158,
+      "loss": 2.2891,
+      "step": 1113
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001778626640518502,
+      "loss": 2.1904,
+      "step": 1114
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017781894510442418,
+      "loss": 2.3135,
+      "step": 1115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017777518841373778,
+      "loss": 2.4688,
+      "step": 1116
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017773139400101363,
+      "loss": 2.2695,
+      "step": 1117
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017768756188749264,
+      "loss": 2.4688,
+      "step": 1118
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017764369209443394,
+      "loss": 2.2129,
+      "step": 1119
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017759978464311496,
+      "loss": 2.3594,
+      "step": 1120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001775558395548315,
+      "loss": 2.2842,
+      "step": 1121
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001775118568508975,
+      "loss": 2.1865,
+      "step": 1122
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017746783655264516,
+      "loss": 2.2949,
+      "step": 1123
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017742377868142493,
+      "loss": 2.3008,
+      "step": 1124
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001773796832586055,
+      "loss": 2.3965,
+      "step": 1125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017733555030557376,
+      "loss": 2.2783,
+      "step": 1126
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017729137984373478,
+      "loss": 2.3291,
+      "step": 1127
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001772471718945119,
+      "loss": 2.3066,
+      "step": 1128
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017720292647934648,
+      "loss": 2.2539,
+      "step": 1129
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017715864361969827,
+      "loss": 2.1436,
+      "step": 1130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00177114323337045,
+      "loss": 2.2529,
+      "step": 1131
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001770699656528826,
+      "loss": 2.2334,
+      "step": 1132
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017702557058872522,
+      "loss": 2.2549,
+      "step": 1133
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017698113816610505,
+      "loss": 2.4785,
+      "step": 1134
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017693666840657239,
+      "loss": 2.2031,
+      "step": 1135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017689216133169575,
+      "loss": 2.252,
+      "step": 1136
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017684761696306164,
+      "loss": 2.3809,
+      "step": 1137
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001768030353222747,
+      "loss": 2.2832,
+      "step": 1138
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001767584164309576,
+      "loss": 2.4609,
+      "step": 1139
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017671376031075119,
+      "loss": 2.2959,
+      "step": 1140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017666906698331427,
+      "loss": 2.21,
+      "step": 1141
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001766243364703237,
+      "loss": 2.3223,
+      "step": 1142
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001765795687934744,
+      "loss": 2.1396,
+      "step": 1143
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017653476397447934,
+      "loss": 2.3242,
+      "step": 1144
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017648992203506946,
+      "loss": 2.2988,
+      "step": 1145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017644504299699371,
+      "loss": 2.2568,
+      "step": 1146
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017640012688201906,
+      "loss": 2.1914,
+      "step": 1147
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017635517371193039,
+      "loss": 2.2988,
+      "step": 1148
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017631018350853069,
+      "loss": 2.1582,
+      "step": 1149
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017626515629364076,
+      "loss": 2.3027,
+      "step": 1150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017622009208909945,
+      "loss": 2.334,
+      "step": 1151
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017617499091676352,
+      "loss": 2.3076,
+      "step": 1152
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017612985279850766,
+      "loss": 2.416,
+      "step": 1153
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017608467775622446,
+      "loss": 2.3691,
+      "step": 1154
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017603946581182446,
+      "loss": 2.208,
+      "step": 1155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017599421698723606,
+      "loss": 2.4023,
+      "step": 1156
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017594893130440557,
+      "loss": 2.2588,
+      "step": 1157
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001759036087852972,
+      "loss": 2.3564,
+      "step": 1158
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017585824945189293,
+      "loss": 2.3926,
+      "step": 1159
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017581285332619273,
+      "loss": 2.3809,
+      "step": 1160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017576742043021426,
+      "loss": 2.3887,
+      "step": 1161
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001757219507859932,
+      "loss": 2.4023,
+      "step": 1162
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001756764444155829,
+      "loss": 2.3965,
+      "step": 1163
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017563090134105458,
+      "loss": 2.1768,
+      "step": 1164
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017558532158449727,
+      "loss": 2.3184,
+      "step": 1165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017553970516801778,
+      "loss": 2.293,
+      "step": 1166
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001754940521137407,
+      "loss": 2.3125,
+      "step": 1167
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017544836244380835,
+      "loss": 2.3711,
+      "step": 1168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017540263618038092,
+      "loss": 2.3643,
+      "step": 1169
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017535687334563623,
+      "loss": 2.3828,
+      "step": 1170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017531107396176991,
+      "loss": 2.3555,
+      "step": 1171
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017526523805099525,
+      "loss": 2.3945,
+      "step": 1172
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017521936563554333,
+      "loss": 2.4102,
+      "step": 1173
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017517345673766292,
+      "loss": 2.3379,
+      "step": 1174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017512751137962045,
+      "loss": 2.2949,
+      "step": 1175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017508152958370003,
+      "loss": 2.1787,
+      "step": 1176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001750355113722035,
+      "loss": 2.293,
+      "step": 1177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017498945676745028,
+      "loss": 2.3203,
+      "step": 1178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017494336579177757,
+      "loss": 2.4102,
+      "step": 1179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017489723846754002,
+      "loss": 2.2393,
+      "step": 1180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001748510748171101,
+      "loss": 2.3975,
+      "step": 1181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001748048748628778,
+      "loss": 2.4961,
+      "step": 1182
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017475863862725072,
+      "loss": 2.3115,
+      "step": 1183
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017471236613265408,
+      "loss": 2.3594,
+      "step": 1184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017466605740153068,
+      "loss": 2.2402,
+      "step": 1185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017461971245634085,
+      "loss": 2.4043,
+      "step": 1186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017457333131956259,
+      "loss": 2.4375,
+      "step": 1187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017452691401369136,
+      "loss": 2.3203,
+      "step": 1188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001744804605612402,
+      "loss": 2.3125,
+      "step": 1189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017443397098473965,
+      "loss": 2.2637,
+      "step": 1190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017438744530673784,
+      "loss": 2.2988,
+      "step": 1191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017434088354980034,
+      "loss": 2.1973,
+      "step": 1192
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017429428573651022,
+      "loss": 2.3633,
+      "step": 1193
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017424765188946809,
+      "loss": 2.2246,
+      "step": 1194
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017420098203129201,
+      "loss": 2.3291,
+      "step": 1195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001741542761846175,
+      "loss": 2.4004,
+      "step": 1196
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017410753437209754,
+      "loss": 2.3301,
+      "step": 1197
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017406075661640253,
+      "loss": 2.1426,
+      "step": 1198
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017401394294022036,
+      "loss": 2.2627,
+      "step": 1199
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001739670933662563,
+      "loss": 2.5,
+      "step": 1200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017392020791723302,
+      "loss": 2.2578,
+      "step": 1201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001738732866158906,
+      "loss": 2.3242,
+      "step": 1202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017382632948498658,
+      "loss": 2.3438,
+      "step": 1203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017377933654729575,
+      "loss": 2.3311,
+      "step": 1204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017373230782561037,
+      "loss": 2.375,
+      "step": 1205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017368524334273995,
+      "loss": 2.3066,
+      "step": 1206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017363814312151154,
+      "loss": 2.3047,
+      "step": 1207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017359100718476927,
+      "loss": 2.3496,
+      "step": 1208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017354383555537479,
+      "loss": 2.3242,
+      "step": 1209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017349662825620698,
+      "loss": 2.2705,
+      "step": 1210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017344938531016199,
+      "loss": 2.3574,
+      "step": 1211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017340210674015334,
+      "loss": 2.2832,
+      "step": 1212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017335479256911182,
+      "loss": 2.165,
+      "step": 1213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017330744281998536,
+      "loss": 2.377,
+      "step": 1214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001732600575157393,
+      "loss": 2.2979,
+      "step": 1215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017321263667935617,
+      "loss": 2.332,
+      "step": 1216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017316518033383573,
+      "loss": 2.3418,
+      "step": 1217
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017311768850219495,
+      "loss": 2.2988,
+      "step": 1218
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017307016120746799,
+      "loss": 2.3633,
+      "step": 1219
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017302259847270629,
+      "loss": 2.3242,
+      "step": 1220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017297500032097842,
+      "loss": 2.2188,
+      "step": 1221
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017292736677537012,
+      "loss": 2.2197,
+      "step": 1222
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017287969785898434,
+      "loss": 2.3047,
+      "step": 1223
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017283199359494118,
+      "loss": 2.3691,
+      "step": 1224
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001727842540063778,
+      "loss": 2.3535,
+      "step": 1225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001727364791164486,
+      "loss": 2.2041,
+      "step": 1226
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017268866894832506,
+      "loss": 2.3281,
+      "step": 1227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017264082352519577,
+      "loss": 2.3496,
+      "step": 1228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017259294287026643,
+      "loss": 2.3789,
+      "step": 1229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017254502700675975,
+      "loss": 2.2266,
+      "step": 1230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017249707595791569,
+      "loss": 2.2656,
+      "step": 1231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017244908974699109,
+      "loss": 2.3887,
+      "step": 1232
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017240106839725995,
+      "loss": 2.2734,
+      "step": 1233
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001723530119320133,
+      "loss": 2.3496,
+      "step": 1234
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017230492037455913,
+      "loss": 2.2617,
+      "step": 1235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001722567937482226,
+      "loss": 2.2891,
+      "step": 1236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001722086320763457,
+      "loss": 2.3154,
+      "step": 1237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017216043538228756,
+      "loss": 2.2656,
+      "step": 1238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001721122036894242,
+      "loss": 2.3008,
+      "step": 1239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017206393702114868,
+      "loss": 2.3652,
+      "step": 1240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017201563540087102,
+      "loss": 2.1396,
+      "step": 1241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017196729885201813,
+      "loss": 2.375,
+      "step": 1242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017191892739803392,
+      "loss": 2.1738,
+      "step": 1243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017187052106237921,
+      "loss": 2.457,
+      "step": 1244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017182207986853177,
+      "loss": 2.292,
+      "step": 1245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017177360383998615,
+      "loss": 2.3809,
+      "step": 1246
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017172509300025398,
+      "loss": 2.2871,
+      "step": 1247
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001716765473728637,
+      "loss": 2.3066,
+      "step": 1248
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017162796698136056,
+      "loss": 2.3359,
+      "step": 1249
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001715793518493067,
+      "loss": 2.3145,
+      "step": 1250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001715307020002812,
+      "loss": 2.2832,
+      "step": 1251
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017148201745787984,
+      "loss": 2.4102,
+      "step": 1252
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017143329824571532,
+      "loss": 2.4453,
+      "step": 1253
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017138454438741712,
+      "loss": 2.2891,
+      "step": 1254
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017133575590663158,
+      "loss": 2.2773,
+      "step": 1255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017128693282702173,
+      "loss": 2.2432,
+      "step": 1256
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017123807517226745,
+      "loss": 2.2217,
+      "step": 1257
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017118918296606537,
+      "loss": 2.3623,
+      "step": 1258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017114025623212888,
+      "loss": 2.3613,
+      "step": 1259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017109129499418816,
+      "loss": 2.2969,
+      "step": 1260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017104229927599004,
+      "loss": 2.2832,
+      "step": 1261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017099326910129814,
+      "loss": 2.4248,
+      "step": 1262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017094420449389278,
+      "loss": 2.21,
+      "step": 1263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017089510547757091,
+      "loss": 2.3926,
+      "step": 1264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017084597207614634,
+      "loss": 2.2822,
+      "step": 1265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017079680431344934,
+      "loss": 2.3564,
+      "step": 1266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017074760221332701,
+      "loss": 2.3105,
+      "step": 1267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017069836579964304,
+      "loss": 2.2061,
+      "step": 1268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017064909509627772,
+      "loss": 2.459,
+      "step": 1269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017059979012712812,
+      "loss": 2.2598,
+      "step": 1270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001705504509161077,
+      "loss": 2.3594,
+      "step": 1271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017050107748714676,
+      "loss": 2.1768,
+      "step": 1272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017045166986419207,
+      "loss": 2.332,
+      "step": 1273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017040222807120698,
+      "loss": 2.1309,
+      "step": 1274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017035275213217146,
+      "loss": 2.3594,
+      "step": 1275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00170303242071082,
+      "loss": 2.4238,
+      "step": 1276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017025369791195172,
+      "loss": 2.3574,
+      "step": 1277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017020411967881015,
+      "loss": 2.3643,
+      "step": 1278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017015450739570346,
+      "loss": 2.3574,
+      "step": 1279
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0017010486108669427,
+      "loss": 2.3027,
+      "step": 1280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0017005518077586174,
+      "loss": 2.418,
+      "step": 1281
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0017000546648730146,
+      "loss": 2.3359,
+      "step": 1282
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016995571824512563,
+      "loss": 2.3066,
+      "step": 1283
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016990593607346277,
+      "loss": 2.3633,
+      "step": 1284
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016985611999645793,
+      "loss": 2.2695,
+      "step": 1285
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016980627003827257,
+      "loss": 2.3379,
+      "step": 1286
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016975638622308466,
+      "loss": 2.3965,
+      "step": 1287
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016970646857508852,
+      "loss": 2.3145,
+      "step": 1288
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016965651711849485,
+      "loss": 2.418,
+      "step": 1289
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016960653187753087,
+      "loss": 2.377,
+      "step": 1290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016955651287644006,
+      "loss": 2.3398,
+      "step": 1291
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016950646013948233,
+      "loss": 2.3145,
+      "step": 1292
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016945637369093394,
+      "loss": 2.252,
+      "step": 1293
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016940625355508754,
+      "loss": 2.2832,
+      "step": 1294
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016935609975625206,
+      "loss": 2.3477,
+      "step": 1295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016930591231875281,
+      "loss": 2.3418,
+      "step": 1296
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016925569126693134,
+      "loss": 2.3047,
+      "step": 1297
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001692054366251456,
+      "loss": 2.3438,
+      "step": 1298
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016915514841776976,
+      "loss": 2.2959,
+      "step": 1299
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001691048266691943,
+      "loss": 2.2461,
+      "step": 1300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016905447140382595,
+      "loss": 2.1885,
+      "step": 1301
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016900408264608772,
+      "loss": 2.4062,
+      "step": 1302
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016895366042041884,
+      "loss": 2.3652,
+      "step": 1303
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016890320475127481,
+      "loss": 2.373,
+      "step": 1304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001688527156631273,
+      "loss": 2.3203,
+      "step": 1305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016880219318046422,
+      "loss": 2.3135,
+      "step": 1306
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016875163732778967,
+      "loss": 2.2207,
+      "step": 1307
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016870104812962393,
+      "loss": 2.2441,
+      "step": 1308
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016865042561050349,
+      "loss": 2.375,
+      "step": 1309
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016859976979498091,
+      "loss": 2.2373,
+      "step": 1310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016854908070762498,
+      "loss": 2.3262,
+      "step": 1311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016849835837302062,
+      "loss": 2.2471,
+      "step": 1312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001684476028157689,
+      "loss": 2.2871,
+      "step": 1313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016839681406048688,
+      "loss": 2.4863,
+      "step": 1314
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016834599213180787,
+      "loss": 2.2627,
+      "step": 1315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016829513705438118,
+      "loss": 2.3848,
+      "step": 1316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016824424885287222,
+      "loss": 2.2891,
+      "step": 1317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016819332755196247,
+      "loss": 2.2891,
+      "step": 1318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016814237317634949,
+      "loss": 2.2539,
+      "step": 1319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001680913857507468,
+      "loss": 2.3613,
+      "step": 1320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016804036529988406,
+      "loss": 2.291,
+      "step": 1321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016798931184850683,
+      "loss": 2.3848,
+      "step": 1322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016793822542137682,
+      "loss": 2.3945,
+      "step": 1323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016788710604327157,
+      "loss": 2.2432,
+      "step": 1324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016783595373898474,
+      "loss": 2.1543,
+      "step": 1325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016778476853332588,
+      "loss": 2.2295,
+      "step": 1326
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001677335504511205,
+      "loss": 2.2207,
+      "step": 1327
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016768229951721013,
+      "loss": 2.2783,
+      "step": 1328
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001676310157564521,
+      "loss": 2.3262,
+      "step": 1329
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016757969919371982,
+      "loss": 2.3594,
+      "step": 1330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016752834985390246,
+      "loss": 2.2646,
+      "step": 1331
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001674769677619052,
+      "loss": 2.1826,
+      "step": 1332
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016742555294264906,
+      "loss": 2.2461,
+      "step": 1333
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016737410542107093,
+      "loss": 2.3574,
+      "step": 1334
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016732262522212357,
+      "loss": 2.2725,
+      "step": 1335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001672711123707756,
+      "loss": 2.334,
+      "step": 1336
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016721956689201144,
+      "loss": 2.2812,
+      "step": 1337
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016716798881083134,
+      "loss": 2.2832,
+      "step": 1338
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016711637815225138,
+      "loss": 2.293,
+      "step": 1339
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001670647349413035,
+      "loss": 2.2041,
+      "step": 1340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001670130592030353,
+      "loss": 2.3877,
+      "step": 1341
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016696135096251029,
+      "loss": 2.3223,
+      "step": 1342
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001669096102448076,
+      "loss": 2.2803,
+      "step": 1343
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016685783707502228,
+      "loss": 2.3594,
+      "step": 1344
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016680603147826499,
+      "loss": 2.4102,
+      "step": 1345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016675419347966212,
+      "loss": 2.3848,
+      "step": 1346
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016670232310435585,
+      "loss": 2.2285,
+      "step": 1347
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00166650420377504,
+      "loss": 2.3721,
+      "step": 1348
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001665984853242802,
+      "loss": 2.4512,
+      "step": 1349
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016654651796987354,
+      "loss": 2.2764,
+      "step": 1350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00166494518339489,
+      "loss": 2.3672,
+      "step": 1351
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016644248645834706,
+      "loss": 2.4082,
+      "step": 1352
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016639042235168388,
+      "loss": 2.3535,
+      "step": 1353
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016633832604475136,
+      "loss": 2.123,
+      "step": 1354
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016628619756281684,
+      "loss": 2.2852,
+      "step": 1355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016623403693116342,
+      "loss": 2.2617,
+      "step": 1356
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016618184417508965,
+      "loss": 2.2236,
+      "step": 1357
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016612961931990978,
+      "loss": 2.3066,
+      "step": 1358
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001660773623909536,
+      "loss": 2.3398,
+      "step": 1359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016602507341356638,
+      "loss": 2.3281,
+      "step": 1360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016597275241310907,
+      "loss": 2.3281,
+      "step": 1361
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016592039941495803,
+      "loss": 2.3594,
+      "step": 1362
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016586801444450519,
+      "loss": 2.2842,
+      "step": 1363
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016581559752715798,
+      "loss": 2.3418,
+      "step": 1364
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016576314868833933,
+      "loss": 2.2939,
+      "step": 1365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016571066795348765,
+      "loss": 2.3311,
+      "step": 1366
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001656581553480568,
+      "loss": 2.335,
+      "step": 1367
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016560561089751622,
+      "loss": 2.3262,
+      "step": 1368
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016555303462735056,
+      "loss": 2.3613,
+      "step": 1369
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016550042656306011,
+      "loss": 2.4004,
+      "step": 1370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016544778673016044,
+      "loss": 2.3027,
+      "step": 1371
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016539511515418265,
+      "loss": 2.3398,
+      "step": 1372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016534241186067318,
+      "loss": 2.2988,
+      "step": 1373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016528967687519382,
+      "loss": 2.292,
+      "step": 1374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016523691022332183,
+      "loss": 2.3262,
+      "step": 1375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016518411193064968,
+      "loss": 2.293,
+      "step": 1376
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016513128202278531,
+      "loss": 2.3633,
+      "step": 1377
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016507842052535199,
+      "loss": 2.4082,
+      "step": 1378
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016502552746398816,
+      "loss": 2.2939,
+      "step": 1379
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001649726028643478,
+      "loss": 2.3174,
+      "step": 1380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016491964675210006,
+      "loss": 2.3594,
+      "step": 1381
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016486665915292928,
+      "loss": 2.3828,
+      "step": 1382
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001648136400925353,
+      "loss": 2.2607,
+      "step": 1383
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016476058959663303,
+      "loss": 2.291,
+      "step": 1384
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001647075076909527,
+      "loss": 2.4414,
+      "step": 1385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016465439440123974,
+      "loss": 2.5391,
+      "step": 1386
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001646012497532549,
+      "loss": 2.2227,
+      "step": 1387
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00164548073772774,
+      "loss": 2.1562,
+      "step": 1388
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016449486648558813,
+      "loss": 2.3047,
+      "step": 1389
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001644416279175036,
+      "loss": 2.2217,
+      "step": 1390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016438835809434182,
+      "loss": 2.3535,
+      "step": 1391
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016433505704193938,
+      "loss": 2.3555,
+      "step": 1392
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016428172478614802,
+      "loss": 2.3574,
+      "step": 1393
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016422836135283469,
+      "loss": 2.4355,
+      "step": 1394
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016417496676788127,
+      "loss": 2.4297,
+      "step": 1395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016412154105718492,
+      "loss": 2.2148,
+      "step": 1396
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016406808424665794,
+      "loss": 2.3105,
+      "step": 1397
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016401459636222744,
+      "loss": 2.2939,
+      "step": 1398
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016396107742983595,
+      "loss": 2.1514,
+      "step": 1399
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016390752747544078,
+      "loss": 2.4375,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016385394652501445,
+      "loss": 2.2441,
+      "step": 1401
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016380033460454448,
+      "loss": 2.4102,
+      "step": 1402
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016374669174003327,
+      "loss": 2.3164,
+      "step": 1403
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016369301795749846,
+      "loss": 2.1562,
+      "step": 1404
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016363931328297256,
+      "loss": 2.377,
+      "step": 1405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016358557774250305,
+      "loss": 2.3008,
+      "step": 1406
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016353181136215245,
+      "loss": 2.2803,
+      "step": 1407
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016347801416799813,
+      "loss": 2.4619,
+      "step": 1408
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001634241861861325,
+      "loss": 2.2988,
+      "step": 1409
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001633703274426629,
+      "loss": 2.3281,
+      "step": 1410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016331643796371152,
+      "loss": 2.291,
+      "step": 1411
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001632625177754155,
+      "loss": 2.3135,
+      "step": 1412
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016320856690392694,
+      "loss": 2.2832,
+      "step": 1413
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001631545853754127,
+      "loss": 2.3398,
+      "step": 1414
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016310057321605456,
+      "loss": 2.3633,
+      "step": 1415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016304653045204915,
+      "loss": 2.3203,
+      "step": 1416
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016299245710960802,
+      "loss": 2.3672,
+      "step": 1417
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016293835321495745,
+      "loss": 2.3594,
+      "step": 1418
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016288421879433854,
+      "loss": 2.4141,
+      "step": 1419
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016283005387400726,
+      "loss": 2.168,
+      "step": 1420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016277585848023435,
+      "loss": 2.2109,
+      "step": 1421
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016272163263930533,
+      "loss": 2.1592,
+      "step": 1422
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016266737637752045,
+      "loss": 2.3545,
+      "step": 1423
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016261308972119475,
+      "loss": 2.3457,
+      "step": 1424
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016255877269665802,
+      "loss": 2.1572,
+      "step": 1425
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016250442533025477,
+      "loss": 2.2549,
+      "step": 1426
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016245004764834422,
+      "loss": 2.3496,
+      "step": 1427
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016239563967730027,
+      "loss": 2.1836,
+      "step": 1428
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016234120144351155,
+      "loss": 2.2988,
+      "step": 1429
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016228673297338137,
+      "loss": 2.1562,
+      "step": 1430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016223223429332764,
+      "loss": 2.4746,
+      "step": 1431
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016217770542978306,
+      "loss": 2.3701,
+      "step": 1432
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001621231464091948,
+      "loss": 2.3799,
+      "step": 1433
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016206855725802475,
+      "loss": 2.5293,
+      "step": 1434
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016201393800274939,
+      "loss": 2.1611,
+      "step": 1435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016195928866985987,
+      "loss": 2.3535,
+      "step": 1436
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016190460928586182,
+      "loss": 2.4023,
+      "step": 1437
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001618498998772755,
+      "loss": 2.2637,
+      "step": 1438
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016179516047063575,
+      "loss": 2.3594,
+      "step": 1439
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001617403910924919,
+      "loss": 2.3398,
+      "step": 1440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016168559176940787,
+      "loss": 2.0938,
+      "step": 1441
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001616307625279621,
+      "loss": 2.3076,
+      "step": 1442
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016157590339474753,
+      "loss": 2.2695,
+      "step": 1443
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016152101439637155,
+      "loss": 2.3252,
+      "step": 1444
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016146609555945613,
+      "loss": 2.4453,
+      "step": 1445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016141114691063762,
+      "loss": 2.4199,
+      "step": 1446
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016135616847656694,
+      "loss": 2.2988,
+      "step": 1447
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001613011602839093,
+      "loss": 2.3027,
+      "step": 1448
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001612461223593445,
+      "loss": 2.2354,
+      "step": 1449
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016119105472956665,
+      "loss": 2.3906,
+      "step": 1450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016113595742128432,
+      "loss": 2.2461,
+      "step": 1451
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016108083046122053,
+      "loss": 2.3555,
+      "step": 1452
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001610256738761125,
+      "loss": 2.1631,
+      "step": 1453
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016097048769271197,
+      "loss": 2.334,
+      "step": 1454
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016091527193778507,
+      "loss": 2.2432,
+      "step": 1455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001608600266381121,
+      "loss": 2.3828,
+      "step": 1456
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016080475182048786,
+      "loss": 2.2617,
+      "step": 1457
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016074944751172135,
+      "loss": 2.3887,
+      "step": 1458
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00160694113738636,
+      "loss": 2.3154,
+      "step": 1459
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016063875052806937,
+      "loss": 2.334,
+      "step": 1460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016058335790687342,
+      "loss": 2.1973,
+      "step": 1461
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016052793590191432,
+      "loss": 2.2559,
+      "step": 1462
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016047248454007252,
+      "loss": 2.293,
+      "step": 1463
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001604170038482427,
+      "loss": 2.3066,
+      "step": 1464
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016036149385333372,
+      "loss": 2.2852,
+      "step": 1465
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0016030595458226872,
+      "loss": 2.3359,
+      "step": 1466
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00160250386061985,
+      "loss": 2.1914,
+      "step": 1467
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0016019478831943408,
+      "loss": 2.2354,
+      "step": 1468
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0016013916138158163,
+      "loss": 2.1914,
+      "step": 1469
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0016008350527540744,
+      "loss": 2.292,
+      "step": 1470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0016002782002790554,
+      "loss": 2.3857,
+      "step": 1471
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015997210566608402,
+      "loss": 2.0469,
+      "step": 1472
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001599163622169651,
+      "loss": 2.2822,
+      "step": 1473
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015986058970758514,
+      "loss": 2.165,
+      "step": 1474
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015980478816499458,
+      "loss": 2.3115,
+      "step": 1475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001597489576162579,
+      "loss": 2.3418,
+      "step": 1476
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001596930980884537,
+      "loss": 2.4268,
+      "step": 1477
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015963720960867466,
+      "loss": 2.2793,
+      "step": 1478
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015958129220402745,
+      "loss": 1.9375,
+      "step": 1479
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015952534590163272,
+      "loss": 2.3379,
+      "step": 1480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015946937072862531,
+      "loss": 2.2578,
+      "step": 1481
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015941336671215383,
+      "loss": 2.291,
+      "step": 1482
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015935733387938105,
+      "loss": 2.373,
+      "step": 1483
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015930127225748368,
+      "loss": 2.2793,
+      "step": 1484
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015924518187365236,
+      "loss": 2.3711,
+      "step": 1485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015918906275509171,
+      "loss": 2.3848,
+      "step": 1486
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015913291492902029,
+      "loss": 2.3652,
+      "step": 1487
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015907673842267052,
+      "loss": 2.4336,
+      "step": 1488
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015902053326328882,
+      "loss": 2.1182,
+      "step": 1489
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015896429947813546,
+      "loss": 2.334,
+      "step": 1490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015890803709448461,
+      "loss": 2.3018,
+      "step": 1491
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015885174613962426,
+      "loss": 2.4688,
+      "step": 1492
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015879542664085633,
+      "loss": 2.3438,
+      "step": 1493
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015873907862549654,
+      "loss": 2.2529,
+      "step": 1494
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001586827021208745,
+      "loss": 2.2598,
+      "step": 1495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001586262971543335,
+      "loss": 2.3086,
+      "step": 1496
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015856986375323086,
+      "loss": 2.4336,
+      "step": 1497
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015851340194493742,
+      "loss": 2.2891,
+      "step": 1498
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00158456911756838,
+      "loss": 2.4062,
+      "step": 1499
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001584003932163311,
+      "loss": 2.3281,
+      "step": 1500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00158343846350829,
+      "loss": 2.4707,
+      "step": 1501
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015828727118775774,
+      "loss": 2.3906,
+      "step": 1502
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015823066775455703,
+      "loss": 2.21,
+      "step": 1503
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015817403607868027,
+      "loss": 2.2441,
+      "step": 1504
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015811737618759468,
+      "loss": 2.2402,
+      "step": 1505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015806068810878102,
+      "loss": 2.3145,
+      "step": 1506
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015800397186973383,
+      "loss": 2.2354,
+      "step": 1507
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015794722749796124,
+      "loss": 2.2822,
+      "step": 1508
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015789045502098507,
+      "loss": 2.2734,
+      "step": 1509
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001578336544663408,
+      "loss": 2.3535,
+      "step": 1510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015777682586157736,
+      "loss": 2.207,
+      "step": 1511
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001577199692342575,
+      "loss": 2.293,
+      "step": 1512
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015766308461195747,
+      "loss": 2.2617,
+      "step": 1513
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015760617202226705,
+      "loss": 2.2559,
+      "step": 1514
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015754923149278969,
+      "loss": 2.334,
+      "step": 1515
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015749226305114227,
+      "loss": 2.208,
+      "step": 1516
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015743526672495527,
+      "loss": 2.2783,
+      "step": 1517
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015737824254187275,
+      "loss": 2.0928,
+      "step": 1518
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015732119052955214,
+      "loss": 2.2891,
+      "step": 1519
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015726411071566454,
+      "loss": 2.3613,
+      "step": 1520
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001572070031278944,
+      "loss": 2.3984,
+      "step": 1521
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015714986779393966,
+      "loss": 2.3926,
+      "step": 1522
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015709270474151182,
+      "loss": 2.167,
+      "step": 1523
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015703551399833572,
+      "loss": 2.2314,
+      "step": 1524
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015697829559214959,
+      "loss": 2.2578,
+      "step": 1525
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015692104955070524,
+      "loss": 2.2617,
+      "step": 1526
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015686377590176772,
+      "loss": 2.3223,
+      "step": 1527
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015680647467311557,
+      "loss": 2.4434,
+      "step": 1528
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001567491458925407,
+      "loss": 2.249,
+      "step": 1529
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001566917895878483,
+      "loss": 2.2617,
+      "step": 1530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015663440578685703,
+      "loss": 2.4219,
+      "step": 1531
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015657699451739877,
+      "loss": 2.2871,
+      "step": 1532
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001565195558073188,
+      "loss": 2.3574,
+      "step": 1533
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015646208968447567,
+      "loss": 2.4316,
+      "step": 1534
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001564045961767413,
+      "loss": 2.3408,
+      "step": 1535
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015634707531200075,
+      "loss": 2.1416,
+      "step": 1536
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001562895271181525,
+      "loss": 2.3418,
+      "step": 1537
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015623195162310815,
+      "loss": 2.3105,
+      "step": 1538
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015617434885479267,
+      "loss": 2.3389,
+      "step": 1539
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015611671884114419,
+      "loss": 2.0625,
+      "step": 1540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015605906161011399,
+      "loss": 2.3203,
+      "step": 1541
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001560013771896667,
+      "loss": 2.1953,
+      "step": 1542
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015594366560778004,
+      "loss": 2.2461,
+      "step": 1543
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001558859268924449,
+      "loss": 2.1865,
+      "step": 1544
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015582816107166538,
+      "loss": 2.3594,
+      "step": 1545
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015577036817345869,
+      "loss": 2.2383,
+      "step": 1546
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015571254822585514,
+      "loss": 2.3184,
+      "step": 1547
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015565470125689829,
+      "loss": 2.2695,
+      "step": 1548
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015559682729464463,
+      "loss": 2.2539,
+      "step": 1549
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015553892636716387,
+      "loss": 2.2666,
+      "step": 1550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015548099850253875,
+      "loss": 2.3809,
+      "step": 1551
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015542304372886508,
+      "loss": 2.375,
+      "step": 1552
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015536506207425177,
+      "loss": 2.3574,
+      "step": 1553
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015530705356682061,
+      "loss": 2.332,
+      "step": 1554
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015524901823470666,
+      "loss": 2.4414,
+      "step": 1555
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015519095610605773,
+      "loss": 2.3008,
+      "step": 1556
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015513286720903485,
+      "loss": 2.3164,
+      "step": 1557
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015507475157181187,
+      "loss": 2.3105,
+      "step": 1558
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015501660922257572,
+      "loss": 2.3047,
+      "step": 1559
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015495844018952622,
+      "loss": 2.3965,
+      "step": 1560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015490024450087617,
+      "loss": 2.334,
+      "step": 1561
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015484202218485122,
+      "loss": 2.2617,
+      "step": 1562
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015478377326969008,
+      "loss": 2.3223,
+      "step": 1563
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001547254977836442,
+      "loss": 2.291,
+      "step": 1564
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015466719575497807,
+      "loss": 2.3145,
+      "step": 1565
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015460886721196893,
+      "loss": 2.1895,
+      "step": 1566
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001545505121829069,
+      "loss": 2.1533,
+      "step": 1567
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00154492130696095,
+      "loss": 2.377,
+      "step": 1568
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015443372277984912,
+      "loss": 2.2441,
+      "step": 1569
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001543752884624978,
+      "loss": 2.3652,
+      "step": 1570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015431682777238259,
+      "loss": 2.207,
+      "step": 1571
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015425834073785761,
+      "loss": 2.2734,
+      "step": 1572
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015419982738728996,
+      "loss": 2.2939,
+      "step": 1573
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015414128774905943,
+      "loss": 2.3066,
+      "step": 1574
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001540827218515585,
+      "loss": 2.3125,
+      "step": 1575
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015402412972319243,
+      "loss": 2.4238,
+      "step": 1576
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001539655113923793,
+      "loss": 2.4219,
+      "step": 1577
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001539068668875497,
+      "loss": 2.2471,
+      "step": 1578
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001538481962371471,
+      "loss": 2.4629,
+      "step": 1579
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015378949946962754,
+      "loss": 2.4043,
+      "step": 1580
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015373077661345975,
+      "loss": 2.3096,
+      "step": 1581
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001536720276971252,
+      "loss": 2.1914,
+      "step": 1582
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015361325274911779,
+      "loss": 2.3271,
+      "step": 1583
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015355445179794428,
+      "loss": 2.418,
+      "step": 1584
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015349562487212394,
+      "loss": 2.2754,
+      "step": 1585
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015343677200018858,
+      "loss": 2.2754,
+      "step": 1586
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001533778932106827,
+      "loss": 2.3789,
+      "step": 1587
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015331898853216328,
+      "loss": 2.1875,
+      "step": 1588
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015326005799319998,
+      "loss": 2.1875,
+      "step": 1589
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015320110162237485,
+      "loss": 2.0781,
+      "step": 1590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015314211944828255,
+      "loss": 2.3975,
+      "step": 1591
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001530831114995303,
+      "loss": 2.3535,
+      "step": 1592
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015302407780473772,
+      "loss": 2.3643,
+      "step": 1593
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015296501839253697,
+      "loss": 2.2715,
+      "step": 1594
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001529059332915727,
+      "loss": 2.3906,
+      "step": 1595
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015284682253050197,
+      "loss": 2.4688,
+      "step": 1596
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015278768613799431,
+      "loss": 2.333,
+      "step": 1597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015272852414273173,
+      "loss": 2.3457,
+      "step": 1598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001526693365734086,
+      "loss": 2.4121,
+      "step": 1599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015261012345873166,
+      "loss": 2.3262,
+      "step": 1600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001525508848274201,
+      "loss": 2.2109,
+      "step": 1601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015249162070820554,
+      "loss": 2.2285,
+      "step": 1602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015243233112983182,
+      "loss": 2.2988,
+      "step": 1603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015237301612105521,
+      "loss": 2.3066,
+      "step": 1604
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015231367571064433,
+      "loss": 2.2314,
+      "step": 1605
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001522543099273801,
+      "loss": 2.2754,
+      "step": 1606
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001521949188000557,
+      "loss": 2.3477,
+      "step": 1607
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015213550235747672,
+      "loss": 2.5,
+      "step": 1608
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001520760606284609,
+      "loss": 2.2393,
+      "step": 1609
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001520165936418383,
+      "loss": 2.3359,
+      "step": 1610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015195710142645123,
+      "loss": 2.2666,
+      "step": 1611
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015189758401115423,
+      "loss": 2.2598,
+      "step": 1612
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001518380414248141,
+      "loss": 2.2559,
+      "step": 1613
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015177847369630979,
+      "loss": 2.3008,
+      "step": 1614
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015171888085453245,
+      "loss": 2.2422,
+      "step": 1615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015165926292838546,
+      "loss": 2.3516,
+      "step": 1616
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001515996199467843,
+      "loss": 2.2461,
+      "step": 1617
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015153995193865667,
+      "loss": 2.292,
+      "step": 1618
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001514802589329424,
+      "loss": 2.4434,
+      "step": 1619
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015142054095859336,
+      "loss": 2.2793,
+      "step": 1620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015136079804457358,
+      "loss": 2.3193,
+      "step": 1621
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015130103021985927,
+      "loss": 2.4004,
+      "step": 1622
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015124123751343863,
+      "loss": 2.1826,
+      "step": 1623
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015118141995431192,
+      "loss": 2.373,
+      "step": 1624
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015112157757149146,
+      "loss": 2.3096,
+      "step": 1625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015106171039400169,
+      "loss": 2.2646,
+      "step": 1626
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015100181845087897,
+      "loss": 2.373,
+      "step": 1627
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001509419017711717,
+      "loss": 2.3418,
+      "step": 1628
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015088196038394036,
+      "loss": 2.2188,
+      "step": 1629
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015082199431825734,
+      "loss": 2.292,
+      "step": 1630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015076200360320694,
+      "loss": 2.2383,
+      "step": 1631
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015070198826788552,
+      "loss": 2.2783,
+      "step": 1632
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015064194834140138,
+      "loss": 2.3125,
+      "step": 1633
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015058188385287466,
+      "loss": 2.2441,
+      "step": 1634
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001505217948314375,
+      "loss": 2.2197,
+      "step": 1635
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015046168130623389,
+      "loss": 2.2168,
+      "step": 1636
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015040154330641972,
+      "loss": 2.2412,
+      "step": 1637
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015034138086116272,
+      "loss": 2.3223,
+      "step": 1638
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015028119399964255,
+      "loss": 2.2793,
+      "step": 1639
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015022098275105065,
+      "loss": 2.208,
+      "step": 1640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001501607471445903,
+      "loss": 2.2881,
+      "step": 1641
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015010048720947658,
+      "loss": 2.3955,
+      "step": 1642
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001500402029749364,
+      "loss": 2.2891,
+      "step": 1643
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0014997989447020853,
+      "loss": 2.3008,
+      "step": 1644
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0014991956172454328,
+      "loss": 2.1494,
+      "step": 1645
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0014985920476720295,
+      "loss": 2.1416,
+      "step": 1646
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001497988236274615,
+      "loss": 2.3145,
+      "step": 1647
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0014973841833460456,
+      "loss": 2.4375,
+      "step": 1648
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0014967798891792957,
+      "loss": 2.2031,
+      "step": 1649
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001496175354067456,
+      "loss": 2.3867,
+      "step": 1650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001495570578303735,
+      "loss": 2.2881,
+      "step": 1651
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014949655621814566,
+      "loss": 2.2412,
+      "step": 1652
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014943603059940623,
+      "loss": 2.3066,
+      "step": 1653
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014937548100351094,
+      "loss": 2.2432,
+      "step": 1654
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014931490745982718,
+      "loss": 2.1338,
+      "step": 1655
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014925430999773402,
+      "loss": 2.3672,
+      "step": 1656
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00149193688646622,
+      "loss": 2.292,
+      "step": 1657
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001491330434358933,
+      "loss": 2.2803,
+      "step": 1658
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014907237439496172,
+      "loss": 2.3594,
+      "step": 1659
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014901168155325255,
+      "loss": 2.3828,
+      "step": 1660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014895096494020274,
+      "loss": 2.2939,
+      "step": 1661
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014889022458526053,
+      "loss": 2.2441,
+      "step": 1662
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014882946051788595,
+      "loss": 2.3223,
+      "step": 1663
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001487686727675504,
+      "loss": 2.3975,
+      "step": 1664
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001487078613637367,
+      "loss": 2.3525,
+      "step": 1665
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014864702633593928,
+      "loss": 2.2783,
+      "step": 1666
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014858616771366397,
+      "loss": 2.2969,
+      "step": 1667
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014852528552642802,
+      "loss": 2.3145,
+      "step": 1668
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014846437980376016,
+      "loss": 2.3242,
+      "step": 1669
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014840345057520045,
+      "loss": 2.2949,
+      "step": 1670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014834249787030044,
+      "loss": 2.2031,
+      "step": 1671
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014828152171862303,
+      "loss": 2.3184,
+      "step": 1672
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014822052214974247,
+      "loss": 2.2373,
+      "step": 1673
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014815949919324443,
+      "loss": 2.3096,
+      "step": 1674
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014809845287872588,
+      "loss": 2.2852,
+      "step": 1675
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014803738323579507,
+      "loss": 2.2598,
+      "step": 1676
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001479762902940717,
+      "loss": 2.3789,
+      "step": 1677
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001479151740831866,
+      "loss": 2.3135,
+      "step": 1678
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014785403463278203,
+      "loss": 2.3008,
+      "step": 1679
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014779287197251147,
+      "loss": 2.2129,
+      "step": 1680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014773168613203965,
+      "loss": 2.2842,
+      "step": 1681
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014767047714104248,
+      "loss": 2.3477,
+      "step": 1682
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014760924502920728,
+      "loss": 2.332,
+      "step": 1683
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014754798982623237,
+      "loss": 2.3613,
+      "step": 1684
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014748671156182742,
+      "loss": 2.3105,
+      "step": 1685
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014742541026571319,
+      "loss": 2.1426,
+      "step": 1686
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001473640859676217,
+      "loss": 2.2812,
+      "step": 1687
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014730273869729606,
+      "loss": 2.2861,
+      "step": 1688
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014724136848449053,
+      "loss": 2.2617,
+      "step": 1689
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001471799753589705,
+      "loss": 2.3594,
+      "step": 1690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001471185593505125,
+      "loss": 2.2266,
+      "step": 1691
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014705712048890417,
+      "loss": 2.3457,
+      "step": 1692
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014699565880394415,
+      "loss": 2.3223,
+      "step": 1693
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001469341743254422,
+      "loss": 2.4219,
+      "step": 1694
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001468726670832192,
+      "loss": 2.1914,
+      "step": 1695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00146811137107107,
+      "loss": 2.2461,
+      "step": 1696
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014674958442694838,
+      "loss": 2.3203,
+      "step": 1697
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014668800907259739,
+      "loss": 2.3271,
+      "step": 1698
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014662641107391883,
+      "loss": 2.1738,
+      "step": 1699
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001465647904607886,
+      "loss": 2.3691,
+      "step": 1700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014650314726309356,
+      "loss": 2.2793,
+      "step": 1701
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014644148151073148,
+      "loss": 2.375,
+      "step": 1702
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014637979323361113,
+      "loss": 2.248,
+      "step": 1703
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014631808246165217,
+      "loss": 2.0879,
+      "step": 1704
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014625634922478517,
+      "loss": 2.2949,
+      "step": 1705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014619459355295157,
+      "loss": 2.1523,
+      "step": 1706
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014613281547610376,
+      "loss": 2.3457,
+      "step": 1707
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014607101502420493,
+      "loss": 2.2656,
+      "step": 1708
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014600919222722921,
+      "loss": 2.2627,
+      "step": 1709
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014594734711516142,
+      "loss": 2.2295,
+      "step": 1710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014588547971799734,
+      "loss": 2.4043,
+      "step": 1711
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014582359006574356,
+      "loss": 2.2598,
+      "step": 1712
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001457616781884173,
+      "loss": 2.252,
+      "step": 1713
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014569974411604676,
+      "loss": 2.2803,
+      "step": 1714
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014563778787867077,
+      "loss": 2.4004,
+      "step": 1715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00145575809506339,
+      "loss": 2.3047,
+      "step": 1716
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001455138090291118,
+      "loss": 2.3496,
+      "step": 1717
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014545178647706024,
+      "loss": 2.3408,
+      "step": 1718
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014538974188026612,
+      "loss": 2.3887,
+      "step": 1719
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014532767526882194,
+      "loss": 2.2793,
+      "step": 1720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014526558667283082,
+      "loss": 2.2246,
+      "step": 1721
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014520347612240663,
+      "loss": 2.2363,
+      "step": 1722
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014514134364767384,
+      "loss": 2.3633,
+      "step": 1723
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001450791892787675,
+      "loss": 2.0947,
+      "step": 1724
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014501701304583342,
+      "loss": 2.252,
+      "step": 1725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014495481497902786,
+      "loss": 2.1758,
+      "step": 1726
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014489259510851778,
+      "loss": 2.3789,
+      "step": 1727
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014483035346448065,
+      "loss": 2.3701,
+      "step": 1728
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014476809007710454,
+      "loss": 2.1953,
+      "step": 1729
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014470580497658804,
+      "loss": 2.2783,
+      "step": 1730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014464349819314029,
+      "loss": 2.3105,
+      "step": 1731
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001445811697569809,
+      "loss": 2.2012,
+      "step": 1732
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014451881969834014,
+      "loss": 2.252,
+      "step": 1733
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014445644804745848,
+      "loss": 2.1084,
+      "step": 1734
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014439405483458714,
+      "loss": 2.293,
+      "step": 1735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014433164008998767,
+      "loss": 2.2549,
+      "step": 1736
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014426920384393205,
+      "loss": 2.1807,
+      "step": 1737
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014420674612670274,
+      "loss": 2.1328,
+      "step": 1738
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014414426696859259,
+      "loss": 2.3262,
+      "step": 1739
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001440817663999049,
+      "loss": 2.4092,
+      "step": 1740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014401924445095323,
+      "loss": 2.208,
+      "step": 1741
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001439567011520616,
+      "loss": 2.3643,
+      "step": 1742
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014389413653356443,
+      "loss": 2.3447,
+      "step": 1743
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014383155062580636,
+      "loss": 2.2646,
+      "step": 1744
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014376894345914243,
+      "loss": 2.2227,
+      "step": 1745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014370631506393798,
+      "loss": 2.2793,
+      "step": 1746
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014364366547056872,
+      "loss": 2.2324,
+      "step": 1747
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014358099470942042,
+      "loss": 2.3184,
+      "step": 1748
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014351830281088937,
+      "loss": 2.3711,
+      "step": 1749
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014345558980538198,
+      "loss": 2.3398,
+      "step": 1750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014339285572331493,
+      "loss": 2.3428,
+      "step": 1751
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014333010059511505,
+      "loss": 2.2773,
+      "step": 1752
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014326732445121952,
+      "loss": 2.3672,
+      "step": 1753
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014320452732207562,
+      "loss": 2.252,
+      "step": 1754
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001431417092381408,
+      "loss": 2.2363,
+      "step": 1755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014307887022988268,
+      "loss": 2.2002,
+      "step": 1756
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014301601032777912,
+      "loss": 2.3926,
+      "step": 1757
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014295312956231795,
+      "loss": 2.252,
+      "step": 1758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001428902279639973,
+      "loss": 2.1924,
+      "step": 1759
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014282730556332522,
+      "loss": 2.2598,
+      "step": 1760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014276436239081996,
+      "loss": 2.3516,
+      "step": 1761
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014270139847700986,
+      "loss": 2.1875,
+      "step": 1762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014263841385243326,
+      "loss": 2.1934,
+      "step": 1763
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001425754085476386,
+      "loss": 2.2197,
+      "step": 1764
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014251238259318428,
+      "loss": 2.3965,
+      "step": 1765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014244933601963879,
+      "loss": 2.3135,
+      "step": 1766
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014238626885758057,
+      "loss": 2.2031,
+      "step": 1767
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014232318113759808,
+      "loss": 2.0117,
+      "step": 1768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001422600728902897,
+      "loss": 2.1357,
+      "step": 1769
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014219694414626383,
+      "loss": 2.1807,
+      "step": 1770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014213379493613878,
+      "loss": 2.3359,
+      "step": 1771
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014207062529054276,
+      "loss": 2.2568,
+      "step": 1772
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014200743524011394,
+      "loss": 2.2871,
+      "step": 1773
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014194422481550036,
+      "loss": 2.3291,
+      "step": 1774
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014188099404735994,
+      "loss": 2.2471,
+      "step": 1775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014181774296636047,
+      "loss": 2.1357,
+      "step": 1776
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014175447160317965,
+      "loss": 2.2461,
+      "step": 1777
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001416911799885049,
+      "loss": 2.2324,
+      "step": 1778
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001416278681530335,
+      "loss": 2.2744,
+      "step": 1779
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014156453612747262,
+      "loss": 2.1211,
+      "step": 1780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014150118394253921,
+      "loss": 2.2832,
+      "step": 1781
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014143781162895984,
+      "loss": 2.127,
+      "step": 1782
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014137441921747104,
+      "loss": 2.3623,
+      "step": 1783
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014131100673881894,
+      "loss": 2.1465,
+      "step": 1784
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014124757422375946,
+      "loss": 2.2607,
+      "step": 1785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014118412170305837,
+      "loss": 2.2344,
+      "step": 1786
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014112064920749083,
+      "loss": 2.2734,
+      "step": 1787
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014105715676784208,
+      "loss": 2.291,
+      "step": 1788
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014099364441490664,
+      "loss": 2.2812,
+      "step": 1789
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014093011217948895,
+      "loss": 2.3867,
+      "step": 1790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014086656009240307,
+      "loss": 2.2402,
+      "step": 1791
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014080298818447256,
+      "loss": 2.2988,
+      "step": 1792
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001407393964865307,
+      "loss": 2.3936,
+      "step": 1793
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014067578502942038,
+      "loss": 2.1201,
+      "step": 1794
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00140612153843994,
+      "loss": 2.1533,
+      "step": 1795
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014054850296111353,
+      "loss": 2.2266,
+      "step": 1796
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014048483241165053,
+      "loss": 2.1055,
+      "step": 1797
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014042114222648614,
+      "loss": 2.1689,
+      "step": 1798
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014035743243651097,
+      "loss": 2.375,
+      "step": 1799
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001402937030726251,
+      "loss": 2.3574,
+      "step": 1800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014022995416573818,
+      "loss": 2.2959,
+      "step": 1801
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014016618574676934,
+      "loss": 2.3398,
+      "step": 1802
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014010239784664708,
+      "loss": 2.3906,
+      "step": 1803
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014003859049630943,
+      "loss": 2.291,
+      "step": 1804
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013997476372670386,
+      "loss": 2.1484,
+      "step": 1805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001399109175687872,
+      "loss": 2.3408,
+      "step": 1806
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013984705205352574,
+      "loss": 2.1855,
+      "step": 1807
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013978316721189512,
+      "loss": 2.3223,
+      "step": 1808
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013971926307488039,
+      "loss": 2.2314,
+      "step": 1809
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001396553396734759,
+      "loss": 2.2178,
+      "step": 1810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001395913970386854,
+      "loss": 2.3486,
+      "step": 1811
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013952743520152199,
+      "loss": 2.2246,
+      "step": 1812
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013946345419300794,
+      "loss": 2.1797,
+      "step": 1813
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00139399454044175,
+      "loss": 2.2773,
+      "step": 1814
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013933543478606412,
+      "loss": 2.165,
+      "step": 1815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013927139644972551,
+      "loss": 2.3477,
+      "step": 1816
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013920733906621862,
+      "loss": 2.4424,
+      "step": 1817
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013914326266661222,
+      "loss": 2.2012,
+      "step": 1818
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013907916728198417,
+      "loss": 2.1953,
+      "step": 1819
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001390150529434217,
+      "loss": 2.3467,
+      "step": 1820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013895091968202107,
+      "loss": 2.2607,
+      "step": 1821
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013888676752888786,
+      "loss": 2.165,
+      "step": 1822
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013882259651513671,
+      "loss": 2.2402,
+      "step": 1823
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013875840667189143,
+      "loss": 2.167,
+      "step": 1824
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013869419803028502,
+      "loss": 2.0596,
+      "step": 1825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013862997062145954,
+      "loss": 2.3945,
+      "step": 1826
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013856572447656617,
+      "loss": 2.21,
+      "step": 1827
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013850145962676517,
+      "loss": 2.1885,
+      "step": 1828
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013843717610322588,
+      "loss": 2.2441,
+      "step": 1829
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013837287393712666,
+      "loss": 2.3066,
+      "step": 1830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013830855315965502,
+      "loss": 2.2979,
+      "step": 1831
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013824421380200739,
+      "loss": 2.2871,
+      "step": 1832
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001381798558953892,
+      "loss": 2.2041,
+      "step": 1833
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013811547947101496,
+      "loss": 2.2725,
+      "step": 1834
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001380510845601081,
+      "loss": 2.2109,
+      "step": 1835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013798667119390112,
+      "loss": 2.3828,
+      "step": 1836
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013792223940363529,
+      "loss": 2.2832,
+      "step": 1837
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013785778922056095,
+      "loss": 2.3574,
+      "step": 1838
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013779332067593738,
+      "loss": 2.1885,
+      "step": 1839
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001377288338010326,
+      "loss": 2.2715,
+      "step": 1840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013766432862712375,
+      "loss": 2.373,
+      "step": 1841
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013759980518549663,
+      "loss": 2.2314,
+      "step": 1842
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001375352635074461,
+      "loss": 2.2236,
+      "step": 1843
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013747070362427568,
+      "loss": 2.2637,
+      "step": 1844
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001374061255672978,
+      "loss": 2.4004,
+      "step": 1845
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013734152936783375,
+      "loss": 2.4727,
+      "step": 1846
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013727691505721357,
+      "loss": 2.0674,
+      "step": 1847
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013721228266677609,
+      "loss": 2.2139,
+      "step": 1848
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001371476322278689,
+      "loss": 2.3154,
+      "step": 1849
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013708296377184835,
+      "loss": 2.1963,
+      "step": 1850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013701827733007953,
+      "loss": 2.376,
+      "step": 1851
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013695357293393626,
+      "loss": 2.2734,
+      "step": 1852
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001368888506148011,
+      "loss": 2.2451,
+      "step": 1853
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001368241104040652,
+      "loss": 2.2041,
+      "step": 1854
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001367593523331285,
+      "loss": 2.3457,
+      "step": 1855
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013669457643339955,
+      "loss": 2.2178,
+      "step": 1856
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013662978273629553,
+      "loss": 2.2539,
+      "step": 1857
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013656497127324232,
+      "loss": 2.3516,
+      "step": 1858
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013650014207567433,
+      "loss": 2.25,
+      "step": 1859
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013643529517503464,
+      "loss": 2.1582,
+      "step": 1860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013637043060277486,
+      "loss": 2.2246,
+      "step": 1861
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001363055483903552,
+      "loss": 2.4199,
+      "step": 1862
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001362406485692445,
+      "loss": 2.2334,
+      "step": 1863
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013617573117092,
+      "loss": 2.3555,
+      "step": 1864
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013611079622686752,
+      "loss": 2.2715,
+      "step": 1865
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013604584376858144,
+      "loss": 2.2266,
+      "step": 1866
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001359808738275646,
+      "loss": 2.3115,
+      "step": 1867
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001359158864353283,
+      "loss": 2.2646,
+      "step": 1868
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013585088162339231,
+      "loss": 2.1865,
+      "step": 1869
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013578585942328489,
+      "loss": 2.2422,
+      "step": 1870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001357208198665427,
+      "loss": 2.2402,
+      "step": 1871
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013565576298471076,
+      "loss": 2.2617,
+      "step": 1872
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013559068880934265,
+      "loss": 2.1621,
+      "step": 1873
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013552559737200016,
+      "loss": 2.252,
+      "step": 1874
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013546048870425357,
+      "loss": 2.2002,
+      "step": 1875
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013539536283768147,
+      "loss": 2.2754,
+      "step": 1876
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013533021980387083,
+      "loss": 2.2178,
+      "step": 1877
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013526505963441689,
+      "loss": 2.2861,
+      "step": 1878
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013519988236092321,
+      "loss": 2.1582,
+      "step": 1879
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013513468801500173,
+      "loss": 2.3926,
+      "step": 1880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013506947662827256,
+      "loss": 2.1133,
+      "step": 1881
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013500424823236412,
+      "loss": 2.0146,
+      "step": 1882
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013493900285891306,
+      "loss": 2.2676,
+      "step": 1883
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013487374053956437,
+      "loss": 2.168,
+      "step": 1884
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013480846130597111,
+      "loss": 2.1074,
+      "step": 1885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013474316518979458,
+      "loss": 2.2725,
+      "step": 1886
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013467785222270434,
+      "loss": 2.3184,
+      "step": 1887
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013461252243637811,
+      "loss": 2.1973,
+      "step": 1888
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013454717586250167,
+      "loss": 2.2637,
+      "step": 1889
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013448181253276903,
+      "loss": 2.1875,
+      "step": 1890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013441643247888233,
+      "loss": 2.2871,
+      "step": 1891
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001343510357325518,
+      "loss": 2.2012,
+      "step": 1892
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013428562232549565,
+      "loss": 2.2803,
+      "step": 1893
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013422019228944045,
+      "loss": 2.3281,
+      "step": 1894
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013415474565612057,
+      "loss": 2.2441,
+      "step": 1895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013408928245727857,
+      "loss": 2.3525,
+      "step": 1896
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013402380272466497,
+      "loss": 2.2734,
+      "step": 1897
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013395830649003836,
+      "loss": 2.1631,
+      "step": 1898
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001338927937851653,
+      "loss": 2.2451,
+      "step": 1899
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013382726464182038,
+      "loss": 2.2012,
+      "step": 1900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013376171909178613,
+      "loss": 2.2451,
+      "step": 1901
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013369615716685304,
+      "loss": 2.2119,
+      "step": 1902
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013363057889881954,
+      "loss": 2.1064,
+      "step": 1903
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00133564984319492,
+      "loss": 2.3037,
+      "step": 1904
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001334993734606847,
+      "loss": 2.1719,
+      "step": 1905
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013343374635421978,
+      "loss": 2.0967,
+      "step": 1906
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013336810303192733,
+      "loss": 2.1494,
+      "step": 1907
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013330244352564528,
+      "loss": 2.293,
+      "step": 1908
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013323676786721932,
+      "loss": 2.2949,
+      "step": 1909
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013317107608850313,
+      "loss": 2.127,
+      "step": 1910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013310536822135806,
+      "loss": 2.2217,
+      "step": 1911
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001330396442976534,
+      "loss": 2.3389,
+      "step": 1912
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013297390434926608,
+      "loss": 2.3711,
+      "step": 1913
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013290814840808095,
+      "loss": 2.1699,
+      "step": 1914
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013284237650599052,
+      "loss": 2.3115,
+      "step": 1915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013277658867489506,
+      "loss": 2.3516,
+      "step": 1916
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013271078494670257,
+      "loss": 2.2891,
+      "step": 1917
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001326449653533288,
+      "loss": 2.2988,
+      "step": 1918
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013257912992669712,
+      "loss": 2.335,
+      "step": 1919
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013251327869873864,
+      "loss": 2.3711,
+      "step": 1920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013244741170139209,
+      "loss": 2.0742,
+      "step": 1921
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013238152896660393,
+      "loss": 2.3213,
+      "step": 1922
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013231563052632813,
+      "loss": 2.1592,
+      "step": 1923
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013224971641252635,
+      "loss": 2.2139,
+      "step": 1924
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013218378665716787,
+      "loss": 2.3057,
+      "step": 1925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013211784129222954,
+      "loss": 2.2031,
+      "step": 1926
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001320518803496957,
+      "loss": 2.416,
+      "step": 1927
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013198590386155843,
+      "loss": 2.291,
+      "step": 1928
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001319199118598171,
+      "loss": 2.2578,
+      "step": 1929
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013185390437647883,
+      "loss": 2.165,
+      "step": 1930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013178788144355815,
+      "loss": 2.3193,
+      "step": 1931
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013172184309307709,
+      "loss": 2.2363,
+      "step": 1932
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013165578935706512,
+      "loss": 2.2109,
+      "step": 1933
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013158972026755926,
+      "loss": 2.5781,
+      "step": 1934
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013152363585660386,
+      "loss": 2.2559,
+      "step": 1935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001314575361562509,
+      "loss": 2.1963,
+      "step": 1936
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013139142119855953,
+      "loss": 2.1367,
+      "step": 1937
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013132529101559643,
+      "loss": 2.2598,
+      "step": 1938
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001312591456394357,
+      "loss": 2.2871,
+      "step": 1939
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001311929851021587,
+      "loss": 2.1445,
+      "step": 1940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013112680943585424,
+      "loss": 2.4238,
+      "step": 1941
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013106061867261842,
+      "loss": 2.1387,
+      "step": 1942
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001309944128445547,
+      "loss": 2.2939,
+      "step": 1943
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013092819198377373,
+      "loss": 2.332,
+      "step": 1944
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013086195612239365,
+      "loss": 2.3125,
+      "step": 1945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001307957052925397,
+      "loss": 2.3086,
+      "step": 1946
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013072943952634447,
+      "loss": 2.2383,
+      "step": 1947
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013066315885594774,
+      "loss": 2.3457,
+      "step": 1948
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013059686331349657,
+      "loss": 2.0928,
+      "step": 1949
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013053055293114522,
+      "loss": 2.3984,
+      "step": 1950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001304642277410551,
+      "loss": 2.3193,
+      "step": 1951
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013039788777539489,
+      "loss": 2.3691,
+      "step": 1952
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013033153306634038,
+      "loss": 2.2676,
+      "step": 1953
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013026516364607447,
+      "loss": 2.2744,
+      "step": 1954
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001301987795467873,
+      "loss": 2.1729,
+      "step": 1955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013013238080067607,
+      "loss": 2.0986,
+      "step": 1956
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013006596743994504,
+      "loss": 2.375,
+      "step": 1957
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012999953949680563,
+      "loss": 2.2285,
+      "step": 1958
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001299330970034763,
+      "loss": 2.4141,
+      "step": 1959
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012986663999218263,
+      "loss": 2.2178,
+      "step": 1960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001298001684951571,
+      "loss": 2.3301,
+      "step": 1961
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012973368254463934,
+      "loss": 2.2832,
+      "step": 1962
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012966718217287596,
+      "loss": 2.1104,
+      "step": 1963
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012960066741212054,
+      "loss": 2.4199,
+      "step": 1964
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012953413829463365,
+      "loss": 2.3066,
+      "step": 1965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012946759485268288,
+      "loss": 2.2871,
+      "step": 1966
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012940103711854267,
+      "loss": 2.2432,
+      "step": 1967
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012933446512449446,
+      "loss": 2.2646,
+      "step": 1968
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012926787890282655,
+      "loss": 2.2734,
+      "step": 1969
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012920127848583419,
+      "loss": 2.166,
+      "step": 1970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001291346639058195,
+      "loss": 2.4043,
+      "step": 1971
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012906803519509148,
+      "loss": 2.3145,
+      "step": 1972
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012900139238596598,
+      "loss": 2.2441,
+      "step": 1973
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012893473551076568,
+      "loss": 2.2822,
+      "step": 1974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012886806460182003,
+      "loss": 2.125,
+      "step": 1975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012880137969146542,
+      "loss": 2.0918,
+      "step": 1976
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012873468081204488,
+      "loss": 2.3467,
+      "step": 1977
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001286679679959083,
+      "loss": 2.2871,
+      "step": 1978
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012860124127541236,
+      "loss": 2.2559,
+      "step": 1979
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001285345006829204,
+      "loss": 2.2793,
+      "step": 1980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012846774625080253,
+      "loss": 2.1953,
+      "step": 1981
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001284009780114356,
+      "loss": 2.1816,
+      "step": 1982
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012833419599720304,
+      "loss": 2.2148,
+      "step": 1983
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012826740024049518,
+      "loss": 2.2803,
+      "step": 1984
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012820059077370877,
+      "loss": 2.166,
+      "step": 1985
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012813376762924734,
+      "loss": 2.2012,
+      "step": 1986
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012806693083952112,
+      "loss": 2.3262,
+      "step": 1987
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012800008043694676,
+      "loss": 2.1074,
+      "step": 1988
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012793321645394767,
+      "loss": 2.2656,
+      "step": 1989
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012786633892295383,
+      "loss": 2.1992,
+      "step": 1990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012779944787640173,
+      "loss": 2.3301,
+      "step": 1991
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012773254334673449,
+      "loss": 2.333,
+      "step": 1992
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012766562536640166,
+      "loss": 2.2812,
+      "step": 1993
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012759869396785945,
+      "loss": 2.2236,
+      "step": 1994
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001275317491835705,
+      "loss": 2.2773,
+      "step": 1995
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012746479104600388,
+      "loss": 2.3359,
+      "step": 1996
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001273978195876353,
+      "loss": 2.293,
+      "step": 1997
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001273308348409468,
+      "loss": 2.1777,
+      "step": 1998
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012726383683842688,
+      "loss": 2.3066,
+      "step": 1999
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012719682561257058,
+      "loss": 2.1211,
+      "step": 2000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012712980119587917,
+      "loss": 2.2051,
+      "step": 2001
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012706276362086048,
+      "loss": 2.2139,
+      "step": 2002
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001269957129200286,
+      "loss": 2.2236,
+      "step": 2003
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012692864912590412,
+      "loss": 2.3613,
+      "step": 2004
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012686157227101383,
+      "loss": 2.2461,
+      "step": 2005
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012679448238789096,
+      "loss": 2.1885,
+      "step": 2006
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012672737950907502,
+      "loss": 2.2783,
+      "step": 2007
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012666026366711187,
+      "loss": 2.3486,
+      "step": 2008
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012659313489455357,
+      "loss": 2.1113,
+      "step": 2009
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001265259932239585,
+      "loss": 2.2373,
+      "step": 2010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012645883868789135,
+      "loss": 2.2354,
+      "step": 2011
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012639167131892294,
+      "loss": 2.332,
+      "step": 2012
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012632449114963035,
+      "loss": 2.168,
+      "step": 2013
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012625729821259695,
+      "loss": 2.332,
+      "step": 2014
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012619009254041223,
+      "loss": 2.2314,
+      "step": 2015
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012612287416567183,
+      "loss": 2.2793,
+      "step": 2016
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001260556431209776,
+      "loss": 2.2568,
+      "step": 2017
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012598839943893751,
+      "loss": 2.1318,
+      "step": 2018
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012592114315216575,
+      "loss": 2.3662,
+      "step": 2019
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012585387429328244,
+      "loss": 2.2236,
+      "step": 2020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012578659289491396,
+      "loss": 2.0635,
+      "step": 2021
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012571929898969273,
+      "loss": 2.1201,
+      "step": 2022
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012565199261025716,
+      "loss": 2.2383,
+      "step": 2023
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012558467378925183,
+      "loss": 2.2139,
+      "step": 2024
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012551734255932727,
+      "loss": 2.4492,
+      "step": 2025
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012544999895314005,
+      "loss": 2.3633,
+      "step": 2026
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012538264300335277,
+      "loss": 2.1885,
+      "step": 2027
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012531527474263396,
+      "loss": 2.1475,
+      "step": 2028
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012524789420365818,
+      "loss": 2.2441,
+      "step": 2029
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001251805014191059,
+      "loss": 2.2539,
+      "step": 2030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012511309642166356,
+      "loss": 2.2275,
+      "step": 2031
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012504567924402354,
+      "loss": 2.3164,
+      "step": 2032
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012497824991888405,
+      "loss": 2.2695,
+      "step": 2033
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012491080847894923,
+      "loss": 2.4375,
+      "step": 2034
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012484335495692922,
+      "loss": 2.1523,
+      "step": 2035
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001247758893855398,
+      "loss": 2.1484,
+      "step": 2036
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012470841179750272,
+      "loss": 2.127,
+      "step": 2037
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012464092222554553,
+      "loss": 2.1738,
+      "step": 2038
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012457342070240165,
+      "loss": 2.208,
+      "step": 2039
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012450590726081022,
+      "loss": 2.1611,
+      "step": 2040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012443838193351615,
+      "loss": 2.2256,
+      "step": 2041
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012437084475327026,
+      "loss": 2.3535,
+      "step": 2042
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012430329575282891,
+      "loss": 2.1475,
+      "step": 2043
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012423573496495432,
+      "loss": 2.3477,
+      "step": 2044
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012416816242241445,
+      "loss": 2.3574,
+      "step": 2045
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012410057815798285,
+      "loss": 2.1621,
+      "step": 2046
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012403298220443886,
+      "loss": 2.3691,
+      "step": 2047
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012396537459456742,
+      "loss": 2.3965,
+      "step": 2048
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001238977553611592,
+      "loss": 2.1807,
+      "step": 2049
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001238301245370104,
+      "loss": 2.2461,
+      "step": 2050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012376248215492296,
+      "loss": 2.2881,
+      "step": 2051
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012369482824770434,
+      "loss": 2.3389,
+      "step": 2052
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001236271628481676,
+      "loss": 2.2988,
+      "step": 2053
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012355948598913136,
+      "loss": 2.2314,
+      "step": 2054
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001234917977034199,
+      "loss": 2.1582,
+      "step": 2055
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00123424098023863,
+      "loss": 2.209,
+      "step": 2056
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012335638698329583,
+      "loss": 2.2266,
+      "step": 2057
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012328866461455924,
+      "loss": 2.1689,
+      "step": 2058
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012322093095049952,
+      "loss": 2.1777,
+      "step": 2059
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012315318602396836,
+      "loss": 2.2012,
+      "step": 2060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001230854298678231,
+      "loss": 2.1338,
+      "step": 2061
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012301766251492628,
+      "loss": 2.2158,
+      "step": 2062
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001229498839981461,
+      "loss": 2.293,
+      "step": 2063
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012288209435035605,
+      "loss": 2.1396,
+      "step": 2064
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012281429360443499,
+      "loss": 2.1885,
+      "step": 2065
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001227464817932673,
+      "loss": 2.2402,
+      "step": 2066
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012267865894974258,
+      "loss": 2.2246,
+      "step": 2067
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012261082510675583,
+      "loss": 1.9902,
+      "step": 2068
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012254298029720748,
+      "loss": 2.2734,
+      "step": 2069
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001224751245540031,
+      "loss": 2.3516,
+      "step": 2070
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012240725791005374,
+      "loss": 2.3301,
+      "step": 2071
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012233938039827562,
+      "loss": 2.1992,
+      "step": 2072
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001222714920515903,
+      "loss": 2.207,
+      "step": 2073
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012220359290292448,
+      "loss": 2.2002,
+      "step": 2074
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012213568298521027,
+      "loss": 2.1562,
+      "step": 2075
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012206776233138488,
+      "loss": 2.252,
+      "step": 2076
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012199983097439077,
+      "loss": 2.251,
+      "step": 2077
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012193188894717556,
+      "loss": 2.2197,
+      "step": 2078
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001218639362826921,
+      "loss": 2.1113,
+      "step": 2079
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001217959730138984,
+      "loss": 2.3438,
+      "step": 2080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001217279991737575,
+      "loss": 2.3018,
+      "step": 2081
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012166001479523769,
+      "loss": 2.0273,
+      "step": 2082
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012159201991131234,
+      "loss": 2.2578,
+      "step": 2083
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012152401455495989,
+      "loss": 2.4609,
+      "step": 2084
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001214559987591639,
+      "loss": 2.375,
+      "step": 2085
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001213879725569129,
+      "loss": 2.2012,
+      "step": 2086
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012131993598120062,
+      "loss": 2.3223,
+      "step": 2087
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001212518890650257,
+      "loss": 2.2109,
+      "step": 2088
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012118383184139179,
+      "loss": 2.2334,
+      "step": 2089
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012111576434330766,
+      "loss": 2.2051,
+      "step": 2090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012104768660378693,
+      "loss": 2.2725,
+      "step": 2091
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012097959865584825,
+      "loss": 2.3477,
+      "step": 2092
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012091150053251524,
+      "loss": 2.2441,
+      "step": 2093
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001208433922668164,
+      "loss": 2.208,
+      "step": 2094
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012077527389178517,
+      "loss": 2.2959,
+      "step": 2095
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012070714544045993,
+      "loss": 2.0459,
+      "step": 2096
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012063900694588392,
+      "loss": 2.1992,
+      "step": 2097
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001205708584411052,
+      "loss": 2.3447,
+      "step": 2098
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012050269995917677,
+      "loss": 2.2656,
+      "step": 2099
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012043453153315644,
+      "loss": 2.2217,
+      "step": 2100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012036635319610683,
+      "loss": 2.3398,
+      "step": 2101
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012029816498109534,
+      "loss": 2.1455,
+      "step": 2102
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012022996692119424,
+      "loss": 2.3428,
+      "step": 2103
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012016175904948048,
+      "loss": 2.0479,
+      "step": 2104
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012009354139903585,
+      "loss": 2.3496,
+      "step": 2105
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012002531400294688,
+      "loss": 2.3379,
+      "step": 2106
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011995707689430473,
+      "loss": 2.1826,
+      "step": 2107
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011988883010620537,
+      "loss": 2.251,
+      "step": 2108
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011982057367174942,
+      "loss": 2.1113,
+      "step": 2109
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011975230762404224,
+      "loss": 2.377,
+      "step": 2110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011968403199619378,
+      "loss": 2.2676,
+      "step": 2111
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011961574682131859,
+      "loss": 2.3584,
+      "step": 2112
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00119547452132536,
+      "loss": 2.2197,
+      "step": 2113
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011947914796296985,
+      "loss": 2.3262,
+      "step": 2114
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011941083434574861,
+      "loss": 2.3887,
+      "step": 2115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001193425113140053,
+      "loss": 2.1104,
+      "step": 2116
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001192741789008776,
+      "loss": 2.1221,
+      "step": 2117
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001192058371395076,
+      "loss": 2.3066,
+      "step": 2118
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00119137486063042,
+      "loss": 2.1758,
+      "step": 2119
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011906912570463204,
+      "loss": 2.1934,
+      "step": 2120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011900075609743346,
+      "loss": 2.2334,
+      "step": 2121
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001189323772746064,
+      "loss": 2.2227,
+      "step": 2122
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011886398926931557,
+      "loss": 2.1621,
+      "step": 2123
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011879559211473009,
+      "loss": 2.2295,
+      "step": 2124
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011872718584402349,
+      "loss": 2.2559,
+      "step": 2125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011865877049037378,
+      "loss": 2.3164,
+      "step": 2126
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001185903460869634,
+      "loss": 2.3447,
+      "step": 2127
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011852191266697905,
+      "loss": 2.2344,
+      "step": 2128
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001184534702636119,
+      "loss": 2.2363,
+      "step": 2129
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011838501891005748,
+      "loss": 2.3574,
+      "step": 2130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011831655863951563,
+      "loss": 2.0234,
+      "step": 2131
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001182480894851905,
+      "loss": 2.1182,
+      "step": 2132
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011817961148029065,
+      "loss": 2.1885,
+      "step": 2133
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011811112465802873,
+      "loss": 2.3711,
+      "step": 2134
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011804262905162192,
+      "loss": 2.2861,
+      "step": 2135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011797412469429145,
+      "loss": 2.2344,
+      "step": 2136
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011790561161926288,
+      "loss": 2.3291,
+      "step": 2137
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011783708985976605,
+      "loss": 2.25,
+      "step": 2138
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011776855944903492,
+      "loss": 2.3652,
+      "step": 2139
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011770002042030766,
+      "loss": 2.2588,
+      "step": 2140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001176314728068267,
+      "loss": 2.0986,
+      "step": 2141
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011756291664183859,
+      "loss": 2.1152,
+      "step": 2142
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011749435195859393,
+      "loss": 2.2637,
+      "step": 2143
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011742577879034755,
+      "loss": 2.1289,
+      "step": 2144
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011735719717035843,
+      "loss": 2.1113,
+      "step": 2145
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001172886071318896,
+      "loss": 2.2305,
+      "step": 2146
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011722000870820813,
+      "loss": 2.2109,
+      "step": 2147
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011715140193258524,
+      "loss": 2.291,
+      "step": 2148
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011708278683829616,
+      "loss": 2.2832,
+      "step": 2149
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001170141634586201,
+      "loss": 2.2451,
+      "step": 2150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011694553182684043,
+      "loss": 2.0439,
+      "step": 2151
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011687689197624437,
+      "loss": 2.2402,
+      "step": 2152
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001168082439401232,
+      "loss": 2.2256,
+      "step": 2153
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001167395877517722,
+      "loss": 2.1924,
+      "step": 2154
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011667092344449053,
+      "loss": 2.2275,
+      "step": 2155
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011660225105158135,
+      "loss": 2.1494,
+      "step": 2156
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001165335706063517,
+      "loss": 2.1338,
+      "step": 2157
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011646488214211255,
+      "loss": 2.3789,
+      "step": 2158
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001163961856921788,
+      "loss": 2.3613,
+      "step": 2159
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011632748128986906,
+      "loss": 2.21,
+      "step": 2160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011625876896850598,
+      "loss": 2.1816,
+      "step": 2161
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011619004876141602,
+      "loss": 2.3652,
+      "step": 2162
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011612132070192936,
+      "loss": 1.9531,
+      "step": 2163
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001160525848233801,
+      "loss": 2.2695,
+      "step": 2164
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011598384115910606,
+      "loss": 2.2812,
+      "step": 2165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011591508974244887,
+      "loss": 2.083,
+      "step": 2166
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011584633060675391,
+      "loss": 2.3359,
+      "step": 2167
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011577756378537032,
+      "loss": 2.3193,
+      "step": 2168
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011570878931165096,
+      "loss": 2.1904,
+      "step": 2169
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001156400072189524,
+      "loss": 2.3311,
+      "step": 2170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011557121754063487,
+      "loss": 2.2666,
+      "step": 2171
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011550242031006235,
+      "loss": 2.1982,
+      "step": 2172
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011543361556060243,
+      "loss": 2.2109,
+      "step": 2173
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011536480332562634,
+      "loss": 2.2598,
+      "step": 2174
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011529598363850897,
+      "loss": 2.2129,
+      "step": 2175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011522715653262887,
+      "loss": 2.2793,
+      "step": 2176
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011515832204136804,
+      "loss": 2.1797,
+      "step": 2177
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011508948019811225,
+      "loss": 2.2041,
+      "step": 2178
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011502063103625072,
+      "loss": 2.3008,
+      "step": 2179
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011495177458917618,
+      "loss": 2.1641,
+      "step": 2180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00114882910890285,
+      "loss": 2.2393,
+      "step": 2181
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011481403997297699,
+      "loss": 2.2979,
+      "step": 2182
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011474516187065561,
+      "loss": 2.1592,
+      "step": 2183
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011467627661672752,
+      "loss": 2.2295,
+      "step": 2184
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011460738424460314,
+      "loss": 2.1309,
+      "step": 2185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011453848478769617,
+      "loss": 2.1797,
+      "step": 2186
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011446957827942381,
+      "loss": 2.2031,
+      "step": 2187
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011440066475320667,
+      "loss": 2.1865,
+      "step": 2188
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001143317442424687,
+      "loss": 2.4707,
+      "step": 2189
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011426281678063742,
+      "loss": 2.2686,
+      "step": 2190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011419388240114347,
+      "loss": 2.291,
+      "step": 2191
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011412494113742105,
+      "loss": 2.1455,
+      "step": 2192
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001140559930229076,
+      "loss": 2.2559,
+      "step": 2193
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011398703809104391,
+      "loss": 2.2441,
+      "step": 2194
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011391807637527406,
+      "loss": 2.1934,
+      "step": 2195
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011384910790904544,
+      "loss": 2.1035,
+      "step": 2196
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011378013272580873,
+      "loss": 2.0303,
+      "step": 2197
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001137111508590178,
+      "loss": 2.2637,
+      "step": 2198
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011364216234212984,
+      "loss": 2.1934,
+      "step": 2199
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001135731672086052,
+      "loss": 2.3164,
+      "step": 2200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011350416549190752,
+      "loss": 2.2285,
+      "step": 2201
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001134351572255035,
+      "loss": 2.1396,
+      "step": 2202
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011336614244286318,
+      "loss": 2.4023,
+      "step": 2203
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011329712117745968,
+      "loss": 2.1982,
+      "step": 2204
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001132280934627692,
+      "loss": 2.2852,
+      "step": 2205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011315905933227115,
+      "loss": 2.373,
+      "step": 2206
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011309001881944808,
+      "loss": 2.3027,
+      "step": 2207
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011302097195778555,
+      "loss": 2.1318,
+      "step": 2208
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001129519187807723,
+      "loss": 2.1016,
+      "step": 2209
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011288285932189994,
+      "loss": 2.0146,
+      "step": 2210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011281379361466341,
+      "loss": 2.2559,
+      "step": 2211
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011274472169256044,
+      "loss": 2.1963,
+      "step": 2212
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011267564358909189,
+      "loss": 2.3125,
+      "step": 2213
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011260655933776158,
+      "loss": 2.2803,
+      "step": 2214
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011253746897207633,
+      "loss": 2.2881,
+      "step": 2215
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011246837252554592,
+      "loss": 2.3027,
+      "step": 2216
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011239927003168309,
+      "loss": 2.1475,
+      "step": 2217
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011233016152400349,
+      "loss": 2.2432,
+      "step": 2218
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011226104703602566,
+      "loss": 2.3066,
+      "step": 2219
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011219192660127115,
+      "loss": 2.2646,
+      "step": 2220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011212280025326429,
+      "loss": 2.1914,
+      "step": 2221
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011205366802553229,
+      "loss": 2.2344,
+      "step": 2222
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011198452995160528,
+      "loss": 2.2266,
+      "step": 2223
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011191538606501617,
+      "loss": 2.3379,
+      "step": 2224
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011184623639930065,
+      "loss": 2.0918,
+      "step": 2225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001117770809879973,
+      "loss": 2.0479,
+      "step": 2226
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001117079198646474,
+      "loss": 2.0928,
+      "step": 2227
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011163875306279515,
+      "loss": 2.1934,
+      "step": 2228
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011156958061598725,
+      "loss": 2.0547,
+      "step": 2229
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011150040255777343,
+      "loss": 2.3555,
+      "step": 2230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011143121892170597,
+      "loss": 2.2002,
+      "step": 2231
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011136202974133982,
+      "loss": 2.168,
+      "step": 2232
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011129283505023273,
+      "loss": 2.1455,
+      "step": 2233
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001112236348819451,
+      "loss": 2.3125,
+      "step": 2234
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011115442927003994,
+      "loss": 2.2295,
+      "step": 2235
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011108521824808291,
+      "loss": 2.2852,
+      "step": 2236
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011101600184964234,
+      "loss": 2.2334,
+      "step": 2237
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011094678010828913,
+      "loss": 2.248,
+      "step": 2238
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011087755305759675,
+      "loss": 2.2529,
+      "step": 2239
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011080832073114128,
+      "loss": 2.2842,
+      "step": 2240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001107390831625014,
+      "loss": 2.207,
+      "step": 2241
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011066984038525826,
+      "loss": 2.1377,
+      "step": 2242
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011060059243299554,
+      "loss": 2.0527,
+      "step": 2243
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011053133933929943,
+      "loss": 2.1768,
+      "step": 2244
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011046208113775872,
+      "loss": 2.3398,
+      "step": 2245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011039281786196454,
+      "loss": 2.3672,
+      "step": 2246
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011032354954551046,
+      "loss": 2.1523,
+      "step": 2247
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011025427622199271,
+      "loss": 2.1504,
+      "step": 2248
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001101849979250097,
+      "loss": 2.1738,
+      "step": 2249
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011011571468816242,
+      "loss": 2.2109,
+      "step": 2250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011004642654505415,
+      "loss": 2.3076,
+      "step": 2251
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001099771335292906,
+      "loss": 2.2363,
+      "step": 2252
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0010990783567447987,
+      "loss": 2.2168,
+      "step": 2253
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0010983853301423238,
+      "loss": 2.252,
+      "step": 2254
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001097692255821608,
+      "loss": 2.2471,
+      "step": 2255
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010969991341188024,
+      "loss": 2.2305,
+      "step": 2256
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010963059653700806,
+      "loss": 2.252,
+      "step": 2257
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001095612749911639,
+      "loss": 2.1836,
+      "step": 2258
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010949194880796966,
+      "loss": 2.3496,
+      "step": 2259
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010942261802104952,
+      "loss": 2.3477,
+      "step": 2260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001093532826640298,
+      "loss": 2.3262,
+      "step": 2261
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001092839427705392,
+      "loss": 2.1162,
+      "step": 2262
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001092145983742084,
+      "loss": 2.2617,
+      "step": 2263
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010914524950867046,
+      "loss": 2.207,
+      "step": 2264
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010907589620756056,
+      "loss": 2.0928,
+      "step": 2265
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010900653850451597,
+      "loss": 2.2402,
+      "step": 2266
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001089371764331761,
+      "loss": 2.0508,
+      "step": 2267
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010886781002718258,
+      "loss": 2.1553,
+      "step": 2268
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010879843932017905,
+      "loss": 2.3379,
+      "step": 2269
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010872906434581124,
+      "loss": 2.1953,
+      "step": 2270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010865968513772695,
+      "loss": 2.2451,
+      "step": 2271
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001085903017295761,
+      "loss": 2.2422,
+      "step": 2272
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010852091415501058,
+      "loss": 2.0654,
+      "step": 2273
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001084515224476843,
+      "loss": 2.3555,
+      "step": 2274
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010838212664125322,
+      "loss": 2.2383,
+      "step": 2275
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010831272676937525,
+      "loss": 2.2969,
+      "step": 2276
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010824332286571029,
+      "loss": 2.0977,
+      "step": 2277
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010817391496392018,
+      "loss": 2.2441,
+      "step": 2278
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001081045030976687,
+      "loss": 2.4434,
+      "step": 2279
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001080350873006216,
+      "loss": 2.167,
+      "step": 2280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010796566760644646,
+      "loss": 2.209,
+      "step": 2281
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001078962440488128,
+      "loss": 2.127,
+      "step": 2282
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010782681666139202,
+      "loss": 2.4043,
+      "step": 2283
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001077573854778573,
+      "loss": 2.1611,
+      "step": 2284
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010768795053188378,
+      "loss": 2.1631,
+      "step": 2285
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001076185118571484,
+      "loss": 2.2266,
+      "step": 2286
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010754906948732977,
+      "loss": 2.1816,
+      "step": 2287
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010747962345610841,
+      "loss": 2.2754,
+      "step": 2288
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010741017379716671,
+      "loss": 2.1133,
+      "step": 2289
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010734072054418861,
+      "loss": 2.3076,
+      "step": 2290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010727126373085993,
+      "loss": 2.208,
+      "step": 2291
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010720180339086817,
+      "loss": 2.3613,
+      "step": 2292
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010713233955790258,
+      "loss": 2.1426,
+      "step": 2293
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010706287226565404,
+      "loss": 2.3633,
+      "step": 2294
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010699340154781513,
+      "loss": 2.1582,
+      "step": 2295
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010692392743808018,
+      "loss": 2.3135,
+      "step": 2296
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010685444997014502,
+      "loss": 2.2334,
+      "step": 2297
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010678496917770719,
+      "loss": 2.2549,
+      "step": 2298
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010671548509446585,
+      "loss": 2.1885,
+      "step": 2299
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010664599775412173,
+      "loss": 2.2529,
+      "step": 2300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010657650719037716,
+      "loss": 2.3418,
+      "step": 2301
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00106507013436936,
+      "loss": 2.2188,
+      "step": 2302
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001064375165275037,
+      "loss": 2.1865,
+      "step": 2303
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010636801649578718,
+      "loss": 2.2275,
+      "step": 2304
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010629851337549494,
+      "loss": 2.3154,
+      "step": 2305
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010622900720033694,
+      "loss": 2.1318,
+      "step": 2306
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001061594980040247,
+      "loss": 2.2285,
+      "step": 2307
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010608998582027102,
+      "loss": 2.1475,
+      "step": 2308
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010602047068279032,
+      "loss": 2.2061,
+      "step": 2309
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010595095262529845,
+      "loss": 2.2256,
+      "step": 2310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010588143168151257,
+      "loss": 2.2021,
+      "step": 2311
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010581190788515127,
+      "loss": 2.1523,
+      "step": 2312
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001057423812699346,
+      "loss": 2.1221,
+      "step": 2313
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010567285186958394,
+      "loss": 2.0596,
+      "step": 2314
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010560331971782196,
+      "loss": 2.0664,
+      "step": 2315
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001055337848483727,
+      "loss": 2.2158,
+      "step": 2316
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001054642472949616,
+      "loss": 2.3057,
+      "step": 2317
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010539470709131527,
+      "loss": 2.3711,
+      "step": 2318
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010532516427116168,
+      "loss": 2.0918,
+      "step": 2319
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010525561886823008,
+      "loss": 2.124,
+      "step": 2320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010518607091625093,
+      "loss": 2.1738,
+      "step": 2321
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010511652044895591,
+      "loss": 2.2637,
+      "step": 2322
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010504696750007804,
+      "loss": 2.123,
+      "step": 2323
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010497741210335138,
+      "loss": 2.3027,
+      "step": 2324
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010490785429251128,
+      "loss": 2.2207,
+      "step": 2325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010483829410129425,
+      "loss": 2.2412,
+      "step": 2326
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010476873156343796,
+      "loss": 2.3711,
+      "step": 2327
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010469916671268114,
+      "loss": 2.3203,
+      "step": 2328
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010462959958276372,
+      "loss": 2.1025,
+      "step": 2329
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010456003020742673,
+      "loss": 2.3047,
+      "step": 2330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010449045862041231,
+      "loss": 2.3252,
+      "step": 2331
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010442088485546354,
+      "loss": 2.1387,
+      "step": 2332
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010435130894632476,
+      "loss": 2.2617,
+      "step": 2333
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010428173092674122,
+      "loss": 2.2061,
+      "step": 2334
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010421215083045916,
+      "loss": 2.3262,
+      "step": 2335
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010414256869122594,
+      "loss": 2.292,
+      "step": 2336
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010407298454278983,
+      "loss": 2.2041,
+      "step": 2337
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001040033984189001,
+      "loss": 2.3379,
+      "step": 2338
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00103933810353307,
+      "loss": 2.207,
+      "step": 2339
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010386422037976168,
+      "loss": 2.3047,
+      "step": 2340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010379462853201626,
+      "loss": 2.2734,
+      "step": 2341
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010372503484382374,
+      "loss": 2.208,
+      "step": 2342
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010365543934893798,
+      "loss": 2.3691,
+      "step": 2343
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010358584208111378,
+      "loss": 2.1904,
+      "step": 2344
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010351624307410679,
+      "loss": 2.1807,
+      "step": 2345
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010344664236167345,
+      "loss": 2.1895,
+      "step": 2346
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001033770399775711,
+      "loss": 2.1104,
+      "step": 2347
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010330743595555788,
+      "loss": 2.209,
+      "step": 2348
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001032378303293926,
+      "loss": 2.0547,
+      "step": 2349
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010316822313283503,
+      "loss": 2.1963,
+      "step": 2350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001030986143996456,
+      "loss": 2.2148,
+      "step": 2351
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010302900416358552,
+      "loss": 2.1416,
+      "step": 2352
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010295939245841665,
+      "loss": 2.2793,
+      "step": 2353
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010288977931790168,
+      "loss": 2.2061,
+      "step": 2354
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010282016477580398,
+      "loss": 2.1855,
+      "step": 2355
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010275054886588748,
+      "loss": 2.2949,
+      "step": 2356
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010268093162191687,
+      "loss": 2.1875,
+      "step": 2357
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010261131307765755,
+      "loss": 2.1699,
+      "step": 2358
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010254169326687538,
+      "loss": 2.1055,
+      "step": 2359
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010247207222333697,
+      "loss": 2.1592,
+      "step": 2360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010240244998080945,
+      "loss": 2.4023,
+      "step": 2361
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010233282657306061,
+      "loss": 2.2529,
+      "step": 2362
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010226320203385877,
+      "loss": 2.2422,
+      "step": 2363
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010219357639697274,
+      "loss": 2.1162,
+      "step": 2364
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010212394969617196,
+      "loss": 2.3271,
+      "step": 2365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001020543219652263,
+      "loss": 2.2969,
+      "step": 2366
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001019846932379062,
+      "loss": 2.1875,
+      "step": 2367
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010191506354798254,
+      "loss": 2.1953,
+      "step": 2368
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010184543292922669,
+      "loss": 2.2852,
+      "step": 2369
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001017758014154104,
+      "loss": 2.375,
+      "step": 2370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010170616904030601,
+      "loss": 2.1191,
+      "step": 2371
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010163653583768614,
+      "loss": 2.3711,
+      "step": 2372
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010156690184132382,
+      "loss": 2.2285,
+      "step": 2373
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001014972670849925,
+      "loss": 2.2305,
+      "step": 2374
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010142763160246604,
+      "loss": 2.2773,
+      "step": 2375
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001013579954275186,
+      "loss": 2.1865,
+      "step": 2376
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010128835859392464,
+      "loss": 2.1543,
+      "step": 2377
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010121872113545902,
+      "loss": 2.0625,
+      "step": 2378
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010114908308589692,
+      "loss": 2.1523,
+      "step": 2379
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010107944447901363,
+      "loss": 2.3438,
+      "step": 2380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010100980534858493,
+      "loss": 2.2676,
+      "step": 2381
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010094016572838671,
+      "loss": 2.292,
+      "step": 2382
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010087052565219514,
+      "loss": 2.2227,
+      "step": 2383
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010080088515378666,
+      "loss": 2.2051,
+      "step": 2384
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010073124426693786,
+      "loss": 2.2461,
+      "step": 2385
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010066160302542552,
+      "loss": 2.2002,
+      "step": 2386
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010059196146302658,
+      "loss": 2.1533,
+      "step": 2387
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010052231961351816,
+      "loss": 2.1807,
+      "step": 2388
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010045267751067758,
+      "loss": 2.2578,
+      "step": 2389
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010038303518828213,
+      "loss": 2.1035,
+      "step": 2390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010031339268010934,
+      "loss": 2.1641,
+      "step": 2391
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001002437500199368,
+      "loss": 2.1152,
+      "step": 2392
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010017410724154215,
+      "loss": 2.3184,
+      "step": 2393
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010010446437870304,
+      "loss": 2.2793,
+      "step": 2394
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010003482146519731,
+      "loss": 2.2744,
+      "step": 2395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009996517853480271,
+      "loss": 2.1582,
+      "step": 2396
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009989553562129694,
+      "loss": 2.3066,
+      "step": 2397
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009982589275845785,
+      "loss": 2.3301,
+      "step": 2398
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000997562499800632,
+      "loss": 2.1602,
+      "step": 2399
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000996866073198907,
+      "loss": 2.2305,
+      "step": 2400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000996169648117179,
+      "loss": 2.0068,
+      "step": 2401
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009954732248932242,
+      "loss": 2.1201,
+      "step": 2402
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009947768038648184,
+      "loss": 2.0645,
+      "step": 2403
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009940803853697347,
+      "loss": 2.1221,
+      "step": 2404
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000993383969745745,
+      "loss": 2.1807,
+      "step": 2405
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009926875573306217,
+      "loss": 2.1191,
+      "step": 2406
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009919911484621334,
+      "loss": 2.1543,
+      "step": 2407
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009912947434780484,
+      "loss": 2.1582,
+      "step": 2408
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009905983427161333,
+      "loss": 2.0635,
+      "step": 2409
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009899019465141512,
+      "loss": 2.1543,
+      "step": 2410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000989205555209864,
+      "loss": 2.2656,
+      "step": 2411
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000988509169141031,
+      "loss": 2.2637,
+      "step": 2412
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009878127886454096,
+      "loss": 2.3291,
+      "step": 2413
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009871164140607535,
+      "loss": 2.4062,
+      "step": 2414
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009864200457248143,
+      "loss": 2.3086,
+      "step": 2415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009857236839753397,
+      "loss": 2.1455,
+      "step": 2416
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009850273291500751,
+      "loss": 2.1357,
+      "step": 2417
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000984330981586762,
+      "loss": 2.1299,
+      "step": 2418
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000983634641623139,
+      "loss": 2.0908,
+      "step": 2419
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009829383095969401,
+      "loss": 2.2324,
+      "step": 2420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000982241985845896,
+      "loss": 2.2812,
+      "step": 2421
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009815456707077334,
+      "loss": 2.2559,
+      "step": 2422
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000980849364520175,
+      "loss": 2.1924,
+      "step": 2423
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000980153067620938,
+      "loss": 2.2705,
+      "step": 2424
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009794567803477372,
+      "loss": 2.2568,
+      "step": 2425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009787605030382807,
+      "loss": 2.0811,
+      "step": 2426
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009780642360302729,
+      "loss": 2.0098,
+      "step": 2427
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009773679796614124,
+      "loss": 2.1807,
+      "step": 2428
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009766717342693937,
+      "loss": 2.1943,
+      "step": 2429
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009759755001919057,
+      "loss": 2.1035,
+      "step": 2430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009752792777666309,
+      "loss": 2.166,
+      "step": 2431
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009745830673312465,
+      "loss": 2.2812,
+      "step": 2432
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009738868692234247,
+      "loss": 2.2627,
+      "step": 2433
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009731906837808312,
+      "loss": 2.3018,
+      "step": 2434
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009724945113411252,
+      "loss": 2.1582,
+      "step": 2435
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009717983522419606,
+      "loss": 2.1855,
+      "step": 2436
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009711022068209832,
+      "loss": 2.1797,
+      "step": 2437
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009704060754158336,
+      "loss": 2.2842,
+      "step": 2438
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000969709958364145,
+      "loss": 2.126,
+      "step": 2439
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009690138560035441,
+      "loss": 2.3408,
+      "step": 2440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00096831776867165,
+      "loss": 2.1582,
+      "step": 2441
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009676216967060742,
+      "loss": 2.2012,
+      "step": 2442
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009669256404444214,
+      "loss": 2.2363,
+      "step": 2443
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000966229600224289,
+      "loss": 2.3105,
+      "step": 2444
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009655335763832653,
+      "loss": 2.208,
+      "step": 2445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009648375692589324,
+      "loss": 2.1494,
+      "step": 2446
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009641415791888624,
+      "loss": 2.1963,
+      "step": 2447
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009634456065106205,
+      "loss": 2.249,
+      "step": 2448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009627496515617628,
+      "loss": 2.1895,
+      "step": 2449
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009620537146798374,
+      "loss": 2.1719,
+      "step": 2450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009613577962023831,
+      "loss": 2.2773,
+      "step": 2451
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009606618964669303,
+      "loss": 2.1387,
+      "step": 2452
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009599660158109991,
+      "loss": 2.2129,
+      "step": 2453
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009592701545721021,
+      "loss": 2.125,
+      "step": 2454
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009585743130877409,
+      "loss": 2.2061,
+      "step": 2455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009578784916954085,
+      "loss": 2.1211,
+      "step": 2456
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009571826907325882,
+      "loss": 2.1006,
+      "step": 2457
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009564869105367526,
+      "loss": 2.2617,
+      "step": 2458
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009557911514453646,
+      "loss": 2.2812,
+      "step": 2459
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000955095413795877,
+      "loss": 2.1084,
+      "step": 2460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009543996979257327,
+      "loss": 2.2217,
+      "step": 2461
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009537040041723633,
+      "loss": 2.2207,
+      "step": 2462
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000953008332873189,
+      "loss": 2.1719,
+      "step": 2463
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009523126843656207,
+      "loss": 2.1484,
+      "step": 2464
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009516170589870575,
+      "loss": 2.2402,
+      "step": 2465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009509214570748871,
+      "loss": 2.2051,
+      "step": 2466
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009502258789664865,
+      "loss": 2.2539,
+      "step": 2467
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009495303249992199,
+      "loss": 2.1533,
+      "step": 2468
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009488347955104409,
+      "loss": 2.1416,
+      "step": 2469
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009481392908374909,
+      "loss": 2.1924,
+      "step": 2470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009474438113176994,
+      "loss": 2.2598,
+      "step": 2471
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009467483572883832,
+      "loss": 2.2002,
+      "step": 2472
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009460529290868476,
+      "loss": 2.2119,
+      "step": 2473
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009453575270503842,
+      "loss": 2.1338,
+      "step": 2474
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009446621515162731,
+      "loss": 2.2227,
+      "step": 2475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009439668028217806,
+      "loss": 1.9668,
+      "step": 2476
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009432714813041607,
+      "loss": 2.2607,
+      "step": 2477
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009425761873006541,
+      "loss": 2.2207,
+      "step": 2478
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009418809211484874,
+      "loss": 2.1797,
+      "step": 2479
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009411856831848745,
+      "loss": 2.3789,
+      "step": 2480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009404904737470155,
+      "loss": 2.1914,
+      "step": 2481
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009397952931720966,
+      "loss": 2.249,
+      "step": 2482
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009391001417972902,
+      "loss": 2.1621,
+      "step": 2483
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009384050199597534,
+      "loss": 2.1709,
+      "step": 2484
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009377099279966307,
+      "loss": 2.2246,
+      "step": 2485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009370148662450507,
+      "loss": 2.3018,
+      "step": 2486
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009363198350421282,
+      "loss": 2.1641,
+      "step": 2487
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009356248347249632,
+      "loss": 2.1465,
+      "step": 2488
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009349298656306404,
+      "loss": 2.2959,
+      "step": 2489
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009342349280962287,
+      "loss": 2.2676,
+      "step": 2490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009335400224587826,
+      "loss": 2.2207,
+      "step": 2491
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009328451490553417,
+      "loss": 2.127,
+      "step": 2492
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009321503082229282,
+      "loss": 2.2188,
+      "step": 2493
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009314555002985502,
+      "loss": 2.2051,
+      "step": 2494
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009307607256191984,
+      "loss": 2.1025,
+      "step": 2495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009300659845218488,
+      "loss": 2.292,
+      "step": 2496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009293712773434598,
+      "loss": 2.248,
+      "step": 2497
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009286766044209742,
+      "loss": 2.4043,
+      "step": 2498
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009279819660913184,
+      "loss": 2.127,
+      "step": 2499
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000927287362691401,
+      "loss": 2.2158,
+      "step": 2500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009265927945581139,
+      "loss": 2.1992,
+      "step": 2501
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000925898262028333,
+      "loss": 2.1777,
+      "step": 2502
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009252037654389157,
+      "loss": 2.1406,
+      "step": 2503
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009245093051267023,
+      "loss": 2.2705,
+      "step": 2504
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009238148814285165,
+      "loss": 2.1162,
+      "step": 2505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009231204946811623,
+      "loss": 2.2246,
+      "step": 2506
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000922426145221427,
+      "loss": 2.2266,
+      "step": 2507
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009217318333860799,
+      "loss": 2.0479,
+      "step": 2508
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000921037559511872,
+      "loss": 2.1777,
+      "step": 2509
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009203433239355357,
+      "loss": 2.25,
+      "step": 2510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009196491269937842,
+      "loss": 2.3672,
+      "step": 2511
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000918954969023313,
+      "loss": 2.1455,
+      "step": 2512
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009182608503607984,
+      "loss": 2.1367,
+      "step": 2513
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009175667713428971,
+      "loss": 2.2461,
+      "step": 2514
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009168727323062478,
+      "loss": 2.2246,
+      "step": 2515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009161787335874679,
+      "loss": 2.3232,
+      "step": 2516
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009154847755231572,
+      "loss": 2.2373,
+      "step": 2517
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009147908584498942,
+      "loss": 2.2227,
+      "step": 2518
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009140969827042391,
+      "loss": 2.2383,
+      "step": 2519
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009134031486227309,
+      "loss": 2.3291,
+      "step": 2520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009127093565418881,
+      "loss": 2.167,
+      "step": 2521
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009120156067982097,
+      "loss": 2.2598,
+      "step": 2522
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009113218997281743,
+      "loss": 2.1152,
+      "step": 2523
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000910628235668239,
+      "loss": 2.1943,
+      "step": 2524
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009099346149548405,
+      "loss": 2.1143,
+      "step": 2525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009092410379243947,
+      "loss": 2.3496,
+      "step": 2526
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009085475049132956,
+      "loss": 2.2334,
+      "step": 2527
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009078540162579162,
+      "loss": 2.2754,
+      "step": 2528
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009071605722946084,
+      "loss": 2.2715,
+      "step": 2529
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000906467173359702,
+      "loss": 2.2139,
+      "step": 2530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009057738197895053,
+      "loss": 2.1953,
+      "step": 2531
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009050805119203034,
+      "loss": 2.1533,
+      "step": 2532
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009043872500883611,
+      "loss": 2.1436,
+      "step": 2533
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009036940346299194,
+      "loss": 2.1436,
+      "step": 2534
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009030008658811975,
+      "loss": 2.2998,
+      "step": 2535
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009023077441783925,
+      "loss": 2.0488,
+      "step": 2536
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009016146698576768,
+      "loss": 2.1602,
+      "step": 2537
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009009216432552014,
+      "loss": 2.207,
+      "step": 2538
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009002286647070939,
+      "loss": 2.25,
+      "step": 2539
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008995357345494587,
+      "loss": 2.167,
+      "step": 2540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008988428531183759,
+      "loss": 2.1357,
+      "step": 2541
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008981500207499031,
+      "loss": 2.2168,
+      "step": 2542
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000897457237780073,
+      "loss": 2.2148,
+      "step": 2543
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008967645045448953,
+      "loss": 2.1572,
+      "step": 2544
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008960718213803549,
+      "loss": 2.2402,
+      "step": 2545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008953791886224128,
+      "loss": 2.2725,
+      "step": 2546
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008946866066070056,
+      "loss": 2.2207,
+      "step": 2547
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008939940756700448,
+      "loss": 2.3379,
+      "step": 2548
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008933015961474174,
+      "loss": 2.1357,
+      "step": 2549
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000892609168374986,
+      "loss": 2.2754,
+      "step": 2550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000891916792688587,
+      "loss": 2.1279,
+      "step": 2551
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008912244694240328,
+      "loss": 2.2139,
+      "step": 2552
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000890532198917109,
+      "loss": 2.0605,
+      "step": 2553
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008898399815035769,
+      "loss": 2.1699,
+      "step": 2554
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008891478175191709,
+      "loss": 2.0469,
+      "step": 2555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008884557072996006,
+      "loss": 2.1885,
+      "step": 2556
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008877636511805492,
+      "loss": 2.1553,
+      "step": 2557
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008870716494976729,
+      "loss": 2.2363,
+      "step": 2558
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000886379702586602,
+      "loss": 2.1797,
+      "step": 2559
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008856878107829405,
+      "loss": 2.2227,
+      "step": 2560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008849959744222657,
+      "loss": 2.2031,
+      "step": 2561
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008843041938401273,
+      "loss": 2.0439,
+      "step": 2562
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008836124693720491,
+      "loss": 2.2207,
+      "step": 2563
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000882920801353526,
+      "loss": 2.252,
+      "step": 2564
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008822291901200275,
+      "loss": 2.2197,
+      "step": 2565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008815376360069937,
+      "loss": 2.3242,
+      "step": 2566
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008808461393498385,
+      "loss": 2.1514,
+      "step": 2567
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008801547004839475,
+      "loss": 2.0732,
+      "step": 2568
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008794633197446771,
+      "loss": 2.2549,
+      "step": 2569
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008787719974673572,
+      "loss": 1.9951,
+      "step": 2570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008780807339872886,
+      "loss": 2.0156,
+      "step": 2571
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008773895296397433,
+      "loss": 2.1621,
+      "step": 2572
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008766983847599655,
+      "loss": 2.1221,
+      "step": 2573
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008760072996831694,
+      "loss": 2.2227,
+      "step": 2574
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000875316274744541,
+      "loss": 2.2236,
+      "step": 2575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008746253102792367,
+      "loss": 2.2129,
+      "step": 2576
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008739344066223842,
+      "loss": 2.1367,
+      "step": 2577
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008732435641090813,
+      "loss": 2.0664,
+      "step": 2578
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000872552783074396,
+      "loss": 2.1611,
+      "step": 2579
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008718620638533661,
+      "loss": 2.043,
+      "step": 2580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008711714067810006,
+      "loss": 2.2422,
+      "step": 2581
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008704808121922774,
+      "loss": 2.3525,
+      "step": 2582
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008697902804221442,
+      "loss": 2.2207,
+      "step": 2583
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008690998118055193,
+      "loss": 2.2236,
+      "step": 2584
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008684094066772887,
+      "loss": 2.1201,
+      "step": 2585
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008677190653723083,
+      "loss": 1.9961,
+      "step": 2586
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008670287882254035,
+      "loss": 2.2969,
+      "step": 2587
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008663385755713683,
+      "loss": 2.124,
+      "step": 2588
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008656484277449653,
+      "loss": 2.251,
+      "step": 2589
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008649583450809253,
+      "loss": 2.1553,
+      "step": 2590
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008642683279139481,
+      "loss": 2.0195,
+      "step": 2591
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008635783765787019,
+      "loss": 2.0303,
+      "step": 2592
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000862888491409822,
+      "loss": 2.0596,
+      "step": 2593
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008621986727419126,
+      "loss": 2.1104,
+      "step": 2594
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008615089209095456,
+      "loss": 2.1533,
+      "step": 2595
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008608192362472597,
+      "loss": 2.1211,
+      "step": 2596
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000860129619089561,
+      "loss": 2.1201,
+      "step": 2597
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008594400697709241,
+      "loss": 2.248,
+      "step": 2598
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008587505886257895,
+      "loss": 2.2256,
+      "step": 2599
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008580611759885655,
+      "loss": 2.2266,
+      "step": 2600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008573718321936262,
+      "loss": 2.1729,
+      "step": 2601
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000856682557575313,
+      "loss": 2.1816,
+      "step": 2602
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008559933524679335,
+      "loss": 2.3438,
+      "step": 2603
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008553042172057619,
+      "loss": 2.084,
+      "step": 2604
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008546151521230385,
+      "loss": 2.209,
+      "step": 2605
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000853926157553969,
+      "loss": 2.2148,
+      "step": 2606
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008532372338327248,
+      "loss": 2.2012,
+      "step": 2607
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008525483812934441,
+      "loss": 2.1299,
+      "step": 2608
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00085185960027023,
+      "loss": 2.2002,
+      "step": 2609
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008511708910971505,
+      "loss": 2.2686,
+      "step": 2610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008504822541082386,
+      "loss": 2.0137,
+      "step": 2611
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008497936896374931,
+      "loss": 2.041,
+      "step": 2612
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008491051980188776,
+      "loss": 2.2793,
+      "step": 2613
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008484167795863195,
+      "loss": 2.125,
+      "step": 2614
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008477284346737115,
+      "loss": 2.0576,
+      "step": 2615
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008470401636149105,
+      "loss": 2.168,
+      "step": 2616
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008463519667437369,
+      "loss": 2.1895,
+      "step": 2617
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008456638443939759,
+      "loss": 2.1094,
+      "step": 2618
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008449757968993768,
+      "loss": 2.1348,
+      "step": 2619
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008442878245936513,
+      "loss": 2.1562,
+      "step": 2620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008435999278104763,
+      "loss": 2.2607,
+      "step": 2621
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008429121068834905,
+      "loss": 2.1338,
+      "step": 2622
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008422243621462969,
+      "loss": 2.0215,
+      "step": 2623
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000841536693932461,
+      "loss": 2.2305,
+      "step": 2624
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008408491025755114,
+      "loss": 2.126,
+      "step": 2625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008401615884089397,
+      "loss": 2.2412,
+      "step": 2626
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008394741517661994,
+      "loss": 2.2266,
+      "step": 2627
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008387867929807065,
+      "loss": 2.2119,
+      "step": 2628
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008380995123858401,
+      "loss": 2.0791,
+      "step": 2629
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008374123103149401,
+      "loss": 2.334,
+      "step": 2630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008367251871013095,
+      "loss": 2.0693,
+      "step": 2631
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008360381430782125,
+      "loss": 2.2617,
+      "step": 2632
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008353511785788747,
+      "loss": 2.208,
+      "step": 2633
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000834664293936483,
+      "loss": 2.1904,
+      "step": 2634
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008339774894841865,
+      "loss": 2.2383,
+      "step": 2635
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008332907655550948,
+      "loss": 2.1348,
+      "step": 2636
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008326041224822785,
+      "loss": 2.25,
+      "step": 2637
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008319175605987683,
+      "loss": 2.3828,
+      "step": 2638
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008312310802375566,
+      "loss": 2.0928,
+      "step": 2639
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000830544681731596,
+      "loss": 2.2734,
+      "step": 2640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000829858365413799,
+      "loss": 2.1367,
+      "step": 2641
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008291721316170389,
+      "loss": 2.1025,
+      "step": 2642
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008284859806741477,
+      "loss": 2.2598,
+      "step": 2643
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008277999129179188,
+      "loss": 2.3848,
+      "step": 2644
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000827113928681104,
+      "loss": 2.3984,
+      "step": 2645
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008264280282964157,
+      "loss": 2.2803,
+      "step": 2646
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008257422120965247,
+      "loss": 2.0361,
+      "step": 2647
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008250564804140612,
+      "loss": 1.9707,
+      "step": 2648
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008243708335816145,
+      "loss": 2.4473,
+      "step": 2649
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008236852719317329,
+      "loss": 2.1494,
+      "step": 2650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008229997957969232,
+      "loss": 2.209,
+      "step": 2651
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008223144055096507,
+      "loss": 2.0928,
+      "step": 2652
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008216291014023396,
+      "loss": 2.2695,
+      "step": 2653
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008209438838073712,
+      "loss": 2.1875,
+      "step": 2654
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008202587530570858,
+      "loss": 2.1699,
+      "step": 2655
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000819573709483781,
+      "loss": 2.1406,
+      "step": 2656
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008188887534197127,
+      "loss": 2.2334,
+      "step": 2657
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008182038851970941,
+      "loss": 2.2568,
+      "step": 2658
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000817519105148095,
+      "loss": 2.1973,
+      "step": 2659
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008168344136048437,
+      "loss": 2.1729,
+      "step": 2660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008161498108994253,
+      "loss": 2.1963,
+      "step": 2661
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008154652973638809,
+      "loss": 2.2539,
+      "step": 2662
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008147808733302098,
+      "loss": 2.1553,
+      "step": 2663
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008140965391303664,
+      "loss": 2.2754,
+      "step": 2664
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008134122950962621,
+      "loss": 2.1475,
+      "step": 2665
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008127281415597652,
+      "loss": 2.2578,
+      "step": 2666
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008120440788526995,
+      "loss": 2.1846,
+      "step": 2667
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008113601073068444,
+      "loss": 2.0996,
+      "step": 2668
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008106762272539363,
+      "loss": 2.1562,
+      "step": 2669
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008099924390256657,
+      "loss": 2.3613,
+      "step": 2670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008093087429536797,
+      "loss": 2.1865,
+      "step": 2671
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00080862513936958,
+      "loss": 2.3584,
+      "step": 2672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000807941628604924,
+      "loss": 2.2695,
+      "step": 2673
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008072582109912243,
+      "loss": 2.1445,
+      "step": 2674
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000806574886859947,
+      "loss": 2.1367,
+      "step": 2675
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000805891656542514,
+      "loss": 2.1475,
+      "step": 2676
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008052085203703017,
+      "loss": 2.1182,
+      "step": 2677
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008045254786746401,
+      "loss": 2.252,
+      "step": 2678
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008038425317868146,
+      "loss": 2.0254,
+      "step": 2679
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008031596800380628,
+      "loss": 2.2119,
+      "step": 2680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008024769237595779,
+      "loss": 2.2646,
+      "step": 2681
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008017942632825057,
+      "loss": 2.2412,
+      "step": 2682
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008011116989379463,
+      "loss": 2.0859,
+      "step": 2683
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008004292310569528,
+      "loss": 2.334,
+      "step": 2684
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007997468599705318,
+      "loss": 2.2793,
+      "step": 2685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007990645860096416,
+      "loss": 2.208,
+      "step": 2686
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007983824095051952,
+      "loss": 2.1406,
+      "step": 2687
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007977003307880579,
+      "loss": 2.2393,
+      "step": 2688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007970183501890466,
+      "loss": 2.2002,
+      "step": 2689
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007963364680389322,
+      "loss": 2.1836,
+      "step": 2690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007956546846684358,
+      "loss": 2.125,
+      "step": 2691
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007949730004082325,
+      "loss": 2.2852,
+      "step": 2692
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000794291415588948,
+      "loss": 2.1943,
+      "step": 2693
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007936099305411611,
+      "loss": 2.167,
+      "step": 2694
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007929285455954007,
+      "loss": 2.1689,
+      "step": 2695
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007922472610821485,
+      "loss": 2.2529,
+      "step": 2696
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007915660773318361,
+      "loss": 2.1924,
+      "step": 2697
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007908849946748477,
+      "loss": 2.1436,
+      "step": 2698
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007902040134415174,
+      "loss": 2.2178,
+      "step": 2699
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007895231339621311,
+      "loss": 2.248,
+      "step": 2700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007888423565669236,
+      "loss": 2.1221,
+      "step": 2701
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007881616815860823,
+      "loss": 2.1914,
+      "step": 2702
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007874811093497433,
+      "loss": 2.2715,
+      "step": 2703
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007868006401879938,
+      "loss": 2.2344,
+      "step": 2704
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000786120274430871,
+      "loss": 2.1572,
+      "step": 2705
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007854400124083615,
+      "loss": 2.0664,
+      "step": 2706
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007847598544504013,
+      "loss": 2.0674,
+      "step": 2707
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007840798008868766,
+      "loss": 1.9004,
+      "step": 2708
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000783399852047623,
+      "loss": 2.2012,
+      "step": 2709
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007827200082624249,
+      "loss": 2.2266,
+      "step": 2710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007820402698610163,
+      "loss": 2.1729,
+      "step": 2711
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000781360637173079,
+      "loss": 2.1455,
+      "step": 2712
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007806811105282444,
+      "loss": 2.2461,
+      "step": 2713
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007800016902560923,
+      "loss": 2.3125,
+      "step": 2714
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007793223766861513,
+      "loss": 2.1006,
+      "step": 2715
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007786431701478977,
+      "loss": 2.2949,
+      "step": 2716
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007779640709707553,
+      "loss": 2.3203,
+      "step": 2717
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007772850794840973,
+      "loss": 2.207,
+      "step": 2718
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007766061960172439,
+      "loss": 2.2207,
+      "step": 2719
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007759274208994626,
+      "loss": 2.1406,
+      "step": 2720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007752487544599686,
+      "loss": 2.1855,
+      "step": 2721
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007745701970279255,
+      "loss": 2.2266,
+      "step": 2722
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007738917489324419,
+      "loss": 2.248,
+      "step": 2723
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007732134105025744,
+      "loss": 2.0176,
+      "step": 2724
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007725351820673273,
+      "loss": 2.2529,
+      "step": 2725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007718570639556501,
+      "loss": 2.2188,
+      "step": 2726
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007711790564964399,
+      "loss": 2.3027,
+      "step": 2727
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000770501160018539,
+      "loss": 2.2148,
+      "step": 2728
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007698233748507372,
+      "loss": 2.3311,
+      "step": 2729
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007691457013217693,
+      "loss": 2.1211,
+      "step": 2730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007684681397603162,
+      "loss": 2.2637,
+      "step": 2731
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007677906904950053,
+      "loss": 2.1973,
+      "step": 2732
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000767113353854408,
+      "loss": 2.1104,
+      "step": 2733
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000766436130167042,
+      "loss": 2.1943,
+      "step": 2734
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007657590197613701,
+      "loss": 2.0645,
+      "step": 2735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007650820229658009,
+      "loss": 2.1797,
+      "step": 2736
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007644051401086863,
+      "loss": 2.166,
+      "step": 2737
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007637283715183245,
+      "loss": 2.1602,
+      "step": 2738
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000763051717522957,
+      "loss": 2.2012,
+      "step": 2739
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007623751784507707,
+      "loss": 2.0264,
+      "step": 2740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007616987546298959,
+      "loss": 2.1924,
+      "step": 2741
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007610224463884081,
+      "loss": 2.3682,
+      "step": 2742
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007603462540543258,
+      "loss": 2.083,
+      "step": 2743
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007596701779556117,
+      "loss": 2.1953,
+      "step": 2744
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007589942184201716,
+      "loss": 2.1172,
+      "step": 2745
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007583183757758559,
+      "loss": 2.1621,
+      "step": 2746
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007576426503504568,
+      "loss": 2.1416,
+      "step": 2747
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007569670424717113,
+      "loss": 2.1943,
+      "step": 2748
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007562915524672977,
+      "loss": 2.0566,
+      "step": 2749
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007556161806648385,
+      "loss": 2.1348,
+      "step": 2750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000754940927391898,
+      "loss": 2.1914,
+      "step": 2751
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007542657929759835,
+      "loss": 2.2139,
+      "step": 2752
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007535907777445449,
+      "loss": 2.2021,
+      "step": 2753
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007529158820249733,
+      "loss": 1.9805,
+      "step": 2754
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007522411061446023,
+      "loss": 2.1865,
+      "step": 2755
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007515664504307078,
+      "loss": 2.0938,
+      "step": 2756
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007508919152105074,
+      "loss": 2.252,
+      "step": 2757
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007502175008111594,
+      "loss": 2.3271,
+      "step": 2758
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007495432075597649,
+      "loss": 2.0381,
+      "step": 2759
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007488690357833646,
+      "loss": 2.1768,
+      "step": 2760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007481949858089412,
+      "loss": 2.2129,
+      "step": 2761
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007475210579634183,
+      "loss": 2.0693,
+      "step": 2762
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007468472525736606,
+      "loss": 2.126,
+      "step": 2763
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007461735699664729,
+      "loss": 2.1445,
+      "step": 2764
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007455000104685998,
+      "loss": 2.1846,
+      "step": 2765
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007448265744067276,
+      "loss": 2.3008,
+      "step": 2766
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007441532621074819,
+      "loss": 2.3105,
+      "step": 2767
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007434800738974284,
+      "loss": 2.2305,
+      "step": 2768
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007428070101030732,
+      "loss": 2.1758,
+      "step": 2769
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007421340710508605,
+      "loss": 2.166,
+      "step": 2770
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007414612570671758,
+      "loss": 2.2188,
+      "step": 2771
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007407885684783428,
+      "loss": 2.2441,
+      "step": 2772
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007401160056106249,
+      "loss": 2.1934,
+      "step": 2773
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007394435687902241,
+      "loss": 2.1514,
+      "step": 2774
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007387712583432821,
+      "loss": 2.3203,
+      "step": 2775
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000738099074595878,
+      "loss": 2.2676,
+      "step": 2776
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007374270178740306,
+      "loss": 2.1143,
+      "step": 2777
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007367550885036964,
+      "loss": 2.2334,
+      "step": 2778
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007360832868107708,
+      "loss": 2.1445,
+      "step": 2779
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007354116131210868,
+      "loss": 2.1777,
+      "step": 2780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007347400677604151,
+      "loss": 2.2051,
+      "step": 2781
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007340686510544644,
+      "loss": 2.3838,
+      "step": 2782
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007333973633288813,
+      "loss": 2.2539,
+      "step": 2783
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007327262049092497,
+      "loss": 2.1865,
+      "step": 2784
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007320551761210907,
+      "loss": 2.1885,
+      "step": 2785
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000731384277289862,
+      "loss": 2.1943,
+      "step": 2786
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007307135087409591,
+      "loss": 2.2139,
+      "step": 2787
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007300428707997141,
+      "loss": 2.1494,
+      "step": 2788
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007293723637913954,
+      "loss": 2.2246,
+      "step": 2789
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007287019880412086,
+      "loss": 2.1514,
+      "step": 2790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007280317438742945,
+      "loss": 2.2129,
+      "step": 2791
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007273616316157312,
+      "loss": 2.1602,
+      "step": 2792
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007266916515905322,
+      "loss": 2.3555,
+      "step": 2793
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007260218041236473,
+      "loss": 2.1709,
+      "step": 2794
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007253520895399612,
+      "loss": 2.0947,
+      "step": 2795
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007246825081642956,
+      "loss": 2.2412,
+      "step": 2796
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007240130603214057,
+      "loss": 2.208,
+      "step": 2797
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007233437463359836,
+      "loss": 2.0371,
+      "step": 2798
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007226745665326553,
+      "loss": 2.0811,
+      "step": 2799
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007220055212359826,
+      "loss": 1.9941,
+      "step": 2800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007213366107704619,
+      "loss": 2.2451,
+      "step": 2801
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007206678354605236,
+      "loss": 2.1221,
+      "step": 2802
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007199991956305327,
+      "loss": 2.168,
+      "step": 2803
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007193306916047891,
+      "loss": 2.3066,
+      "step": 2804
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007186623237075265,
+      "loss": 2.2539,
+      "step": 2805
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007179940922629127,
+      "loss": 2.2295,
+      "step": 2806
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007173259975950486,
+      "loss": 2.1797,
+      "step": 2807
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007166580400279697,
+      "loss": 2.1777,
+      "step": 2808
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007159902198856443,
+      "loss": 2.2598,
+      "step": 2809
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007153225374919747,
+      "loss": 2.2383,
+      "step": 2810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007146549931707959,
+      "loss": 2.0098,
+      "step": 2811
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007139875872458767,
+      "loss": 2.1816,
+      "step": 2812
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007133203200409171,
+      "loss": 2.1885,
+      "step": 2813
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007126531918795514,
+      "loss": 2.082,
+      "step": 2814
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007119862030853461,
+      "loss": 2.167,
+      "step": 2815
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007113193539817997,
+      "loss": 2.2441,
+      "step": 2816
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007106526448923436,
+      "loss": 2.0977,
+      "step": 2817
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007099860761403403,
+      "loss": 2.207,
+      "step": 2818
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007093196480490853,
+      "loss": 2.2578,
+      "step": 2819
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007086533609418052,
+      "loss": 2.2422,
+      "step": 2820
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007079872151416583,
+      "loss": 2.2656,
+      "step": 2821
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007073212109717349,
+      "loss": 2.0752,
+      "step": 2822
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007066553487550558,
+      "loss": 2.1709,
+      "step": 2823
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007059896288145734,
+      "loss": 2.2207,
+      "step": 2824
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007053240514731713,
+      "loss": 2.1416,
+      "step": 2825
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007046586170536633,
+      "loss": 2.0566,
+      "step": 2826
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007039933258787945,
+      "loss": 2.2383,
+      "step": 2827
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007033281782712407,
+      "loss": 2.1553,
+      "step": 2828
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000702663174553607,
+      "loss": 2.0977,
+      "step": 2829
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007019983150484293,
+      "loss": 2.1201,
+      "step": 2830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007013336000781738,
+      "loss": 2.1436,
+      "step": 2831
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000700669029965237,
+      "loss": 2.1562,
+      "step": 2832
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000700004605031944,
+      "loss": 2.1494,
+      "step": 2833
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006993403256005498,
+      "loss": 2.1709,
+      "step": 2834
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006986761919932394,
+      "loss": 2.2275,
+      "step": 2835
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006980122045321268,
+      "loss": 2.0605,
+      "step": 2836
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000697348363539255,
+      "loss": 2.0859,
+      "step": 2837
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006966846693365965,
+      "loss": 2.2051,
+      "step": 2838
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006960211222460511,
+      "loss": 2.1143,
+      "step": 2839
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006953577225894491,
+      "loss": 2.1494,
+      "step": 2840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000694694470688548,
+      "loss": 2.2324,
+      "step": 2841
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006940313668650345,
+      "loss": 2.1406,
+      "step": 2842
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006933684114405231,
+      "loss": 2.2441,
+      "step": 2843
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006927056047365557,
+      "loss": 2.1123,
+      "step": 2844
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006920429470746032,
+      "loss": 2.2285,
+      "step": 2845
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006913804387760637,
+      "loss": 2.2285,
+      "step": 2846
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006907180801622626,
+      "loss": 2.209,
+      "step": 2847
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006900558715544531,
+      "loss": 2.248,
+      "step": 2848
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006893938132738159,
+      "loss": 2.2959,
+      "step": 2849
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006887319056414578,
+      "loss": 2.0996,
+      "step": 2850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006880701489784131,
+      "loss": 2.165,
+      "step": 2851
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000687408543605643,
+      "loss": 2.2305,
+      "step": 2852
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006867470898440357,
+      "loss": 2.1885,
+      "step": 2853
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006860857880144052,
+      "loss": 1.9727,
+      "step": 2854
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006854246384374911,
+      "loss": 2.165,
+      "step": 2855
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006847636414339613,
+      "loss": 2.0176,
+      "step": 2856
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006841027973244076,
+      "loss": 2.2412,
+      "step": 2857
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006834421064293488,
+      "loss": 2.2012,
+      "step": 2858
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006827815690692294,
+      "loss": 2.1611,
+      "step": 2859
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006821211855644187,
+      "loss": 2.1924,
+      "step": 2860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006814609562352117,
+      "loss": 2.1631,
+      "step": 2861
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000680800881401829,
+      "loss": 2.125,
+      "step": 2862
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006801409613844161,
+      "loss": 2.2129,
+      "step": 2863
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006794811965030428,
+      "loss": 2.1738,
+      "step": 2864
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000678821587077705,
+      "loss": 2.1221,
+      "step": 2865
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006781621334283212,
+      "loss": 2.0898,
+      "step": 2866
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006775028358747366,
+      "loss": 2.3184,
+      "step": 2867
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006768436947367188,
+      "loss": 2.2109,
+      "step": 2868
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006761847103339608,
+      "loss": 2.1934,
+      "step": 2869
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006755258829860791,
+      "loss": 2.0957,
+      "step": 2870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006748672130126138,
+      "loss": 2.1875,
+      "step": 2871
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006742087007330288,
+      "loss": 2.1592,
+      "step": 2872
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006735503464667123,
+      "loss": 2.2559,
+      "step": 2873
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006728921505329743,
+      "loss": 2.2393,
+      "step": 2874
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006722341132510499,
+      "loss": 2.3076,
+      "step": 2875
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006715762349400952,
+      "loss": 2.082,
+      "step": 2876
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006709185159191909,
+      "loss": 2.2422,
+      "step": 2877
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006702609565073394,
+      "loss": 2.1504,
+      "step": 2878
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006696035570234661,
+      "loss": 2.1758,
+      "step": 2879
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006689463177864197,
+      "loss": 2.1758,
+      "step": 2880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006682892391149692,
+      "loss": 2.1934,
+      "step": 2881
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006676323213278071,
+      "loss": 2.1875,
+      "step": 2882
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006669755647435474,
+      "loss": 2.0273,
+      "step": 2883
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006663189696807268,
+      "loss": 2.0723,
+      "step": 2884
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006656625364578021,
+      "loss": 2.209,
+      "step": 2885
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006650062653931534,
+      "loss": 2.1182,
+      "step": 2886
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006643501568050802,
+      "loss": 2.1367,
+      "step": 2887
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006636942110118049,
+      "loss": 2.1104,
+      "step": 2888
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006630384283314697,
+      "loss": 2.126,
+      "step": 2889
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006623828090821388,
+      "loss": 2.2197,
+      "step": 2890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006617273535817965,
+      "loss": 2.0674,
+      "step": 2891
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006610720621483473,
+      "loss": 2.1436,
+      "step": 2892
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006604169350996165,
+      "loss": 2.252,
+      "step": 2893
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006597619727533504,
+      "loss": 2.2148,
+      "step": 2894
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006591071754272145,
+      "loss": 2.0625,
+      "step": 2895
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006584525434387944,
+      "loss": 2.2871,
+      "step": 2896
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006577980771055956,
+      "loss": 2.2402,
+      "step": 2897
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006571437767450434,
+      "loss": 2.1826,
+      "step": 2898
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006564896426744824,
+      "loss": 2.2588,
+      "step": 2899
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006558356752111767,
+      "loss": 2.3145,
+      "step": 2900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006551818746723096,
+      "loss": 2.0371,
+      "step": 2901
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006545282413749836,
+      "loss": 2.1357,
+      "step": 2902
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006538747756362191,
+      "loss": 2.084,
+      "step": 2903
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006532214777729566,
+      "loss": 2.1504,
+      "step": 2904
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006525683481020541,
+      "loss": 1.9795,
+      "step": 2905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006519153869402889,
+      "loss": 2.2832,
+      "step": 2906
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006512625946043565,
+      "loss": 2.3154,
+      "step": 2907
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006506099714108695,
+      "loss": 2.0752,
+      "step": 2908
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000649957517676359,
+      "loss": 2.2861,
+      "step": 2909
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006493052337172745,
+      "loss": 2.2373,
+      "step": 2910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006486531198499828,
+      "loss": 2.0752,
+      "step": 2911
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006480011763907681,
+      "loss": 2.2207,
+      "step": 2912
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006473494036558314,
+      "loss": 2.3936,
+      "step": 2913
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006466978019612919,
+      "loss": 2.1494,
+      "step": 2914
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006460463716231854,
+      "loss": 2.1553,
+      "step": 2915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006453951129574643,
+      "loss": 2.3555,
+      "step": 2916
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006447440262799985,
+      "loss": 2.2109,
+      "step": 2917
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006440931119065738,
+      "loss": 2.2041,
+      "step": 2918
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006434423701528924,
+      "loss": 1.9502,
+      "step": 2919
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006427918013345732,
+      "loss": 2.126,
+      "step": 2920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000642141405767151,
+      "loss": 2.251,
+      "step": 2921
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006414911837660767,
+      "loss": 2.1992,
+      "step": 2922
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006408411356467172,
+      "loss": 2.1318,
+      "step": 2923
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000640191261724354,
+      "loss": 2.207,
+      "step": 2924
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006395415623141856,
+      "loss": 2.0762,
+      "step": 2925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006388920377313247,
+      "loss": 2.1445,
+      "step": 2926
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006382426882908001,
+      "loss": 2.0957,
+      "step": 2927
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006375935143075553,
+      "loss": 2.375,
+      "step": 2928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006369445160964481,
+      "loss": 2.1846,
+      "step": 2929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006362956939722518,
+      "loss": 2.1895,
+      "step": 2930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006356470482496538,
+      "loss": 2.165,
+      "step": 2931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006349985792432569,
+      "loss": 2.0957,
+      "step": 2932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006343502872675772,
+      "loss": 2.2266,
+      "step": 2933
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006337021726370448,
+      "loss": 2.0947,
+      "step": 2934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006330542356660046,
+      "loss": 2.2168,
+      "step": 2935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006324064766687152,
+      "loss": 2.2148,
+      "step": 2936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000631758895959348,
+      "loss": 1.9297,
+      "step": 2937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006311114938519892,
+      "loss": 2.1855,
+      "step": 2938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006304642706606376,
+      "loss": 2.0996,
+      "step": 2939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006298172266992049,
+      "loss": 2.0527,
+      "step": 2940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006291703622815167,
+      "loss": 2.2275,
+      "step": 2941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006285236777213112,
+      "loss": 2.21,
+      "step": 2942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006278771733322392,
+      "loss": 2.1367,
+      "step": 2943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006272308494278645,
+      "loss": 2.1514,
+      "step": 2944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006265847063216625,
+      "loss": 2.2021,
+      "step": 2945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006259387443270223,
+      "loss": 1.998,
+      "step": 2946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006252929637572436,
+      "loss": 2.167,
+      "step": 2947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006246473649255391,
+      "loss": 2.1064,
+      "step": 2948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006240019481450337,
+      "loss": 2.2236,
+      "step": 2949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006233567137287628,
+      "loss": 2.0479,
+      "step": 2950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000622711661989674,
+      "loss": 2.208,
+      "step": 2951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006220667932406265,
+      "loss": 2.2422,
+      "step": 2952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006214221077943904,
+      "loss": 2.1689,
+      "step": 2953
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000620777605963647,
+      "loss": 2.1465,
+      "step": 2954
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006201332880609892,
+      "loss": 2.208,
+      "step": 2955
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006194891543989191,
+      "loss": 2.1611,
+      "step": 2956
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006188452052898507,
+      "loss": 2.2021,
+      "step": 2957
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006182014410461082,
+      "loss": 2.3262,
+      "step": 2958
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006175578619799265,
+      "loss": 2.2383,
+      "step": 2959
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006169144684034502,
+      "loss": 2.0908,
+      "step": 2960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006162712606287335,
+      "loss": 2.3242,
+      "step": 2961
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006156282389677414,
+      "loss": 2.1611,
+      "step": 2962
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006149854037323485,
+      "loss": 2.1133,
+      "step": 2963
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006143427552343383,
+      "loss": 2.0322,
+      "step": 2964
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006137002937854049,
+      "loss": 2.1973,
+      "step": 2965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006130580196971498,
+      "loss": 2.165,
+      "step": 2966
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006124159332810858,
+      "loss": 2.1572,
+      "step": 2967
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000611774034848633,
+      "loss": 2.1025,
+      "step": 2968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006111323247111216,
+      "loss": 2.0957,
+      "step": 2969
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006104908031797896,
+      "loss": 2.4004,
+      "step": 2970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006098494705657834,
+      "loss": 2.1104,
+      "step": 2971
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006092083271801583,
+      "loss": 2.1689,
+      "step": 2972
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006085673733338781,
+      "loss": 2.127,
+      "step": 2973
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006079266093378138,
+      "loss": 2.0684,
+      "step": 2974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006072860355027449,
+      "loss": 2.1797,
+      "step": 2975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000606645652139359,
+      "loss": 2.1025,
+      "step": 2976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006060054595582502,
+      "loss": 2.0234,
+      "step": 2977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006053654580699207,
+      "loss": 2.0996,
+      "step": 2978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006047256479847804,
+      "loss": 2.2695,
+      "step": 2979
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000604086029613146,
+      "loss": 2.2041,
+      "step": 2980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006034466032652414,
+      "loss": 2.127,
+      "step": 2981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006028073692511964,
+      "loss": 2.2441,
+      "step": 2982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006021683278810488,
+      "loss": 2.2207,
+      "step": 2983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006015294794647428,
+      "loss": 2.166,
+      "step": 2984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000600890824312128,
+      "loss": 2.2451,
+      "step": 2985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006002523627329617,
+      "loss": 2.1904,
+      "step": 2986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000599614095036906,
+      "loss": 2.0684,
+      "step": 2987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005989760215335295,
+      "loss": 2.2051,
+      "step": 2988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005983381425323068,
+      "loss": 2.0029,
+      "step": 2989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005977004583426183,
+      "loss": 2.2451,
+      "step": 2990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000597062969273749,
+      "loss": 2.0332,
+      "step": 2991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005964256756348907,
+      "loss": 2.1357,
+      "step": 2992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005957885777351386,
+      "loss": 2.1465,
+      "step": 2993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005951516758834948,
+      "loss": 1.9863,
+      "step": 2994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005945149703888649,
+      "loss": 1.9854,
+      "step": 2995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005938784615600601,
+      "loss": 1.9688,
+      "step": 2996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005932421497057963,
+      "loss": 2.2246,
+      "step": 2997
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005926060351346929,
+      "loss": 1.918,
+      "step": 2998
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005919701181552743,
+      "loss": 2.3027,
+      "step": 2999
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005913343990759694,
+      "loss": 2.2412,
+      "step": 3000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005906988782051104,
+      "loss": 2.1455,
+      "step": 3001
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000590063555850934,
+      "loss": 2.1895,
+      "step": 3002
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005894284323215797,
+      "loss": 2.1943,
+      "step": 3003
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005887935079250916,
+      "loss": 2.1641,
+      "step": 3004
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005881587829694166,
+      "loss": 2.1514,
+      "step": 3005
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005875242577624051,
+      "loss": 2.0293,
+      "step": 3006
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005868899326118109,
+      "loss": 2.2822,
+      "step": 3007
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005862558078252902,
+      "loss": 2.2158,
+      "step": 3008
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005856218837104019,
+      "loss": 2.3076,
+      "step": 3009
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005849881605746081,
+      "loss": 2.1475,
+      "step": 3010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005843546387252737,
+      "loss": 2.2002,
+      "step": 3011
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005837213184696649,
+      "loss": 2.1191,
+      "step": 3012
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005830882001149517,
+      "loss": 2.2412,
+      "step": 3013
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005824552839682038,
+      "loss": 2.127,
+      "step": 3014
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005818225703363953,
+      "loss": 2.1123,
+      "step": 3015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005811900595264008,
+      "loss": 2.0957,
+      "step": 3016
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005805577518449963,
+      "loss": 2.1914,
+      "step": 3017
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000579925647598861,
+      "loss": 2.1846,
+      "step": 3018
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005792937470945726,
+      "loss": 2.0732,
+      "step": 3019
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005786620506386124,
+      "loss": 2.1494,
+      "step": 3020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005780305585373618,
+      "loss": 2.2334,
+      "step": 3021
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005773992710971028,
+      "loss": 2.1494,
+      "step": 3022
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005767681886240195,
+      "loss": 2.1602,
+      "step": 3023
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005761373114241946,
+      "loss": 2.1006,
+      "step": 3024
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005755066398036122,
+      "loss": 2.2178,
+      "step": 3025
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005748761740681573,
+      "loss": 2.2354,
+      "step": 3026
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005742459145236143,
+      "loss": 2.2109,
+      "step": 3027
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005736158614756673,
+      "loss": 2.1211,
+      "step": 3028
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005729860152299018,
+      "loss": 2.0469,
+      "step": 3029
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005723563760918009,
+      "loss": 2.0859,
+      "step": 3030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005717269443667482,
+      "loss": 2.2627,
+      "step": 3031
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005710977203600274,
+      "loss": 2.2441,
+      "step": 3032
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005704687043768205,
+      "loss": 2.3301,
+      "step": 3033
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000569839896722209,
+      "loss": 2.1162,
+      "step": 3034
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005692112977011732,
+      "loss": 2.0557,
+      "step": 3035
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000568582907618592,
+      "loss": 2.0947,
+      "step": 3036
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005679547267792437,
+      "loss": 2.2227,
+      "step": 3037
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005673267554878047,
+      "loss": 2.1699,
+      "step": 3038
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005666989940488497,
+      "loss": 2.1582,
+      "step": 3039
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005660714427668512,
+      "loss": 2.2178,
+      "step": 3040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005654441019461806,
+      "loss": 2.1367,
+      "step": 3041
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005648169718911062,
+      "loss": 2.2451,
+      "step": 3042
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005641900529057959,
+      "loss": 2.1055,
+      "step": 3043
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005635633452943132,
+      "loss": 2.2217,
+      "step": 3044
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005629368493606201,
+      "loss": 2.2178,
+      "step": 3045
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000562310565408576,
+      "loss": 2.3096,
+      "step": 3046
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005616844937419368,
+      "loss": 2.2129,
+      "step": 3047
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005610586346643558,
+      "loss": 2.0225,
+      "step": 3048
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000560432988479384,
+      "loss": 2.1768,
+      "step": 3049
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005598075554904679,
+      "loss": 2.0713,
+      "step": 3050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005591823360009513,
+      "loss": 2.0869,
+      "step": 3051
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000558557330314074,
+      "loss": 2.1602,
+      "step": 3052
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005579325387329724,
+      "loss": 2.2773,
+      "step": 3053
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005573079615606795,
+      "loss": 2.1855,
+      "step": 3054
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005566835991001238,
+      "loss": 2.0742,
+      "step": 3055
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005560594516541286,
+      "loss": 2.1113,
+      "step": 3056
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005554355195254154,
+      "loss": 2.2412,
+      "step": 3057
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005548118030165992,
+      "loss": 2.1318,
+      "step": 3058
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005541883024301905,
+      "loss": 2.209,
+      "step": 3059
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005535650180685975,
+      "loss": 2.1475,
+      "step": 3060
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00055294195023412,
+      "loss": 2.2441,
+      "step": 3061
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005523190992289546,
+      "loss": 2.1787,
+      "step": 3062
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005516964653551937,
+      "loss": 2.1025,
+      "step": 3063
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005510740489148225,
+      "loss": 2.1328,
+      "step": 3064
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005504518502097212,
+      "loss": 2.2881,
+      "step": 3065
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005498298695416662,
+      "loss": 2.1592,
+      "step": 3066
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005492081072123249,
+      "loss": 2.249,
+      "step": 3067
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005485865635232618,
+      "loss": 2.1982,
+      "step": 3068
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005479652387759338,
+      "loss": 2.207,
+      "step": 3069
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005473441332716915,
+      "loss": 2.1631,
+      "step": 3070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005467232473117809,
+      "loss": 2.2256,
+      "step": 3071
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005461025811973391,
+      "loss": 2.0488,
+      "step": 3072
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005454821352293977,
+      "loss": 2.3809,
+      "step": 3073
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000544861909708882,
+      "loss": 2.2969,
+      "step": 3074
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005442419049366101,
+      "loss": 2.4414,
+      "step": 3075
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005436221212132923,
+      "loss": 2.2832,
+      "step": 3076
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005430025588395327,
+      "loss": 2.1484,
+      "step": 3077
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005423832181158274,
+      "loss": 2.1543,
+      "step": 3078
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005417640993425647,
+      "loss": 2.1943,
+      "step": 3079
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005411452028200265,
+      "loss": 2.1367,
+      "step": 3080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005405265288483859,
+      "loss": 2.1494,
+      "step": 3081
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005399080777277084,
+      "loss": 2.0928,
+      "step": 3082
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005392898497579508,
+      "loss": 2.2539,
+      "step": 3083
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005386718452389624,
+      "loss": 2.1543,
+      "step": 3084
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005380540644704843,
+      "loss": 2.1709,
+      "step": 3085
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005374365077521486,
+      "loss": 2.0986,
+      "step": 3086
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005368191753834786,
+      "loss": 2.1865,
+      "step": 3087
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005362020676638888,
+      "loss": 2.1484,
+      "step": 3088
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005355851848926854,
+      "loss": 2.1084,
+      "step": 3089
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005349685273690644,
+      "loss": 2.1738,
+      "step": 3090
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005343520953921139,
+      "loss": 2.042,
+      "step": 3091
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005337358892608121,
+      "loss": 2.2578,
+      "step": 3092
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005331199092740262,
+      "loss": 2.1738,
+      "step": 3093
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005325041557305161,
+      "loss": 2.0391,
+      "step": 3094
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005318886289289305,
+      "loss": 2.0566,
+      "step": 3095
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005312733291678079,
+      "loss": 2.0859,
+      "step": 3096
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005306582567455779,
+      "loss": 2.1436,
+      "step": 3097
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005300434119605588,
+      "loss": 2.2832,
+      "step": 3098
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005294287951109586,
+      "loss": 2.1016,
+      "step": 3099
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005288144064948753,
+      "loss": 2.1406,
+      "step": 3100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005282002464102949,
+      "loss": 2.2188,
+      "step": 3101
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005275863151550948,
+      "loss": 2.0244,
+      "step": 3102
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00052697261302704,
+      "loss": 2.0967,
+      "step": 3103
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005263591403237831,
+      "loss": 2.2695,
+      "step": 3104
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005257458973428682,
+      "loss": 2.085,
+      "step": 3105
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000525132884381726,
+      "loss": 2.1514,
+      "step": 3106
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005245201017376761,
+      "loss": 2.1514,
+      "step": 3107
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005239075497079275,
+      "loss": 2.0029,
+      "step": 3108
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005232952285895753,
+      "loss": 2.1777,
+      "step": 3109
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005226831386796037,
+      "loss": 2.1816,
+      "step": 3110
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005220712802748854,
+      "loss": 2.0986,
+      "step": 3111
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005214596536721798,
+      "loss": 2.2188,
+      "step": 3112
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005208482591681343,
+      "loss": 2.0234,
+      "step": 3113
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005202370970592836,
+      "loss": 2.1855,
+      "step": 3114
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005196261676420493,
+      "loss": 2.1299,
+      "step": 3115
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005190154712127415,
+      "loss": 2.083,
+      "step": 3116
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005184050080675557,
+      "loss": 2.0518,
+      "step": 3117
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000517794778502575,
+      "loss": 2.1143,
+      "step": 3118
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005171847828137701,
+      "loss": 2.1533,
+      "step": 3119
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000516575021296996,
+      "loss": 2.1543,
+      "step": 3120
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005159654942479956,
+      "loss": 2.126,
+      "step": 3121
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005153562019623985,
+      "loss": 2.2041,
+      "step": 3122
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005147471447357197,
+      "loss": 2.1816,
+      "step": 3123
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005141383228633604,
+      "loss": 2.2129,
+      "step": 3124
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005135297366406073,
+      "loss": 2.2725,
+      "step": 3125
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005129213863626332,
+      "loss": 2.1836,
+      "step": 3126
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005123132723244961,
+      "loss": 2.1582,
+      "step": 3127
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005117053948211405,
+      "loss": 2.2324,
+      "step": 3128
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005110977541473947,
+      "loss": 2.2441,
+      "step": 3129
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000510490350597973,
+      "loss": 2.2275,
+      "step": 3130
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005098831844674745,
+      "loss": 2.2285,
+      "step": 3131
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005092762560503826,
+      "loss": 2.1719,
+      "step": 3132
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000508669565641067,
+      "loss": 2.2061,
+      "step": 3133
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005080631135337801,
+      "loss": 2.1377,
+      "step": 3134
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005074569000226599,
+      "loss": 2.1963,
+      "step": 3135
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005068509254017282,
+      "loss": 2.2393,
+      "step": 3136
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005062451899648909,
+      "loss": 2.0469,
+      "step": 3137
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005056396940059378,
+      "loss": 2.1816,
+      "step": 3138
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005050344378185434,
+      "loss": 1.9502,
+      "step": 3139
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005044294216962655,
+      "loss": 2.0596,
+      "step": 3140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005038246459325439,
+      "loss": 2.1943,
+      "step": 3141
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005032201108207046,
+      "loss": 2.291,
+      "step": 3142
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005026158166539547,
+      "loss": 2.167,
+      "step": 3143
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005020117637253851,
+      "loss": 2.1377,
+      "step": 3144
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005014079523279709,
+      "loss": 2.1631,
+      "step": 3145
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005008043827545672,
+      "loss": 2.3145,
+      "step": 3146
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000500201055297915,
+      "loss": 2.168,
+      "step": 3147
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004995979702506359,
+      "loss": 2.1338,
+      "step": 3148
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004989951279052341,
+      "loss": 2.1875,
+      "step": 3149
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004983925285540975,
+      "loss": 1.96,
+      "step": 3150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004977901724894938,
+      "loss": 2.25,
+      "step": 3151
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004971880600035746,
+      "loss": 2.1855,
+      "step": 3152
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000496586191388373,
+      "loss": 2.2207,
+      "step": 3153
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004959845669358031,
+      "loss": 2.1572,
+      "step": 3154
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000495383186937661,
+      "loss": 2.0801,
+      "step": 3155
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004947820516856253,
+      "loss": 2.2139,
+      "step": 3156
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004941811614712537,
+      "loss": 2.1904,
+      "step": 3157
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004935805165859862,
+      "loss": 2.1475,
+      "step": 3158
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004929801173211448,
+      "loss": 2.1475,
+      "step": 3159
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004923799639679308,
+      "loss": 2.1436,
+      "step": 3160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004917800568174271,
+      "loss": 2.0801,
+      "step": 3161
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004911803961605966,
+      "loss": 2.0684,
+      "step": 3162
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004905809822882828,
+      "loss": 2.1211,
+      "step": 3163
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004899818154912105,
+      "loss": 2.1094,
+      "step": 3164
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004893828960599833,
+      "loss": 2.127,
+      "step": 3165
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004887842242850855,
+      "loss": 2.0938,
+      "step": 3166
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004881858004568812,
+      "loss": 2.1846,
+      "step": 3167
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00048758762486561405,
+      "loss": 2.0293,
+      "step": 3168
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004869896978014071,
+      "loss": 2.1729,
+      "step": 3169
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00048639201955426407,
+      "loss": 2.2217,
+      "step": 3170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004857945904140667,
+      "loss": 2.2314,
+      "step": 3171
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004851974106705763,
+      "loss": 2.1924,
+      "step": 3172
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004846004806134334,
+      "loss": 2.082,
+      "step": 3173
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004840038005321572,
+      "loss": 2.2129,
+      "step": 3174
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004834073707161455,
+      "loss": 2.1416,
+      "step": 3175
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00048281119145467555,
+      "loss": 2.209,
+      "step": 3176
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00048221526303690224,
+      "loss": 2.1631,
+      "step": 3177
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00048161958575185904,
+      "loss": 2.1963,
+      "step": 3178
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004810241598884577,
+      "loss": 1.9805,
+      "step": 3179
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00048042898573548757,
+      "loss": 2.1807,
+      "step": 3180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00047983406358161697,
+      "loss": 2.1416,
+      "step": 3181
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00047923939371539135,
+      "loss": 2.0186,
+      "step": 3182
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00047864497642523274,
+      "loss": 2.0615,
+      "step": 3183
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004780508119994429,
+      "loss": 2.0967,
+      "step": 3184
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004774569007261992,
+      "loss": 2.207,
+      "step": 3185
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00047686324289355656,
+      "loss": 2.0117,
+      "step": 3186
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047626983878944787,
+      "loss": 2.1377,
+      "step": 3187
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047567668870168234,
+      "loss": 2.127,
+      "step": 3188
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004750837929179448,
+      "loss": 2.2676,
+      "step": 3189
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004744911517257989,
+      "loss": 2.0586,
+      "step": 3190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047389876541268375,
+      "loss": 2.2422,
+      "step": 3191
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047330663426591404,
+      "loss": 2.3184,
+      "step": 3192
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047271475857268297,
+      "loss": 2.2139,
+      "step": 3193
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047212313862005675,
+      "loss": 2.0811,
+      "step": 3194
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047153177469498044,
+      "loss": 2.1543,
+      "step": 3195
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047094066708427317,
+      "loss": 2.0596,
+      "step": 3196
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047034981607463013,
+      "loss": 2.1641,
+      "step": 3197
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004697592219526231,
+      "loss": 2.0684,
+      "step": 3198
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046916888500469735,
+      "loss": 2.2607,
+      "step": 3199
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004685788055171744,
+      "loss": 2.1533,
+      "step": 3200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046798898377625167,
+      "loss": 2.2773,
+      "step": 3201
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004673994200680004,
+      "loss": 2.2236,
+      "step": 3202
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004668101146783673,
+      "loss": 2.2207,
+      "step": 3203
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004662210678931734,
+      "loss": 2.0312,
+      "step": 3204
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046563227999811465,
+      "loss": 2.1377,
+      "step": 3205
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004650437512787609,
+      "loss": 2.292,
+      "step": 3206
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004644554820205572,
+      "loss": 2.1289,
+      "step": 3207
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046386747250882224,
+      "loss": 2.1172,
+      "step": 3208
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046327972302874845,
+      "loss": 2.0635,
+      "step": 3209
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004626922338654026,
+      "loss": 2.1182,
+      "step": 3210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046210500530372445,
+      "loss": 2.2168,
+      "step": 3211
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046151803762852894,
+      "loss": 2.1318,
+      "step": 3212
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004609313311245029,
+      "loss": 2.1455,
+      "step": 3213
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046034488607620726,
+      "loss": 2.0146,
+      "step": 3214
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004597587027680757,
+      "loss": 2.0527,
+      "step": 3215
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004591727814844154,
+      "loss": 2.2012,
+      "step": 3216
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045858712250940593,
+      "loss": 2.1484,
+      "step": 3217
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045800172612710043,
+      "loss": 2.2188,
+      "step": 3218
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004574165926214241,
+      "loss": 2.167,
+      "step": 3219
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004568317222761745,
+      "loss": 2.1406,
+      "step": 3220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045624711537502205,
+      "loss": 2.124,
+      "step": 3221
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004556627722015091,
+      "loss": 2.1465,
+      "step": 3222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004550786930390497,
+      "loss": 2.3193,
+      "step": 3223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045449487817093105,
+      "loss": 2.207,
+      "step": 3224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000453911327880311,
+      "loss": 2.1514,
+      "step": 3225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004533280424502195,
+      "loss": 2.124,
+      "step": 3226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045274502216355805,
+      "loss": 2.3594,
+      "step": 3227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004521622673030992,
+      "loss": 2.3027,
+      "step": 3228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045157977815148777,
+      "loss": 2.1357,
+      "step": 3229
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045099755499123886,
+      "loss": 2.0781,
+      "step": 3230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045041559810473785,
+      "loss": 2.2373,
+      "step": 3231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00044983390777424294,
+      "loss": 2.0986,
+      "step": 3232
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004492524842818815,
+      "loss": 2.1953,
+      "step": 3233
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044867132790965146,
+      "loss": 2.1416,
+      "step": 3234
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000448090438939423,
+      "loss": 2.1934,
+      "step": 3235
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000447509817652934,
+      "loss": 2.0586,
+      "step": 3236
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004469294643317938,
+      "loss": 2.0908,
+      "step": 3237
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044634937925748264,
+      "loss": 2.1543,
+      "step": 3238
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004457695627113493,
+      "loss": 2.1201,
+      "step": 3239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044519001497461273,
+      "loss": 2.1309,
+      "step": 3240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004446107363283616,
+      "loss": 2.2285,
+      "step": 3241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044403172705355376,
+      "loss": 2.291,
+      "step": 3242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044345298743101726,
+      "loss": 2.166,
+      "step": 3243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004428745177414486,
+      "loss": 2.1895,
+      "step": 3244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044229631826541306,
+      "loss": 2.0674,
+      "step": 3245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004417183892833465,
+      "loss": 2.2324,
+      "step": 3246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004411407310755513,
+      "loss": 2.1807,
+      "step": 3247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044056334392219963,
+      "loss": 2.124,
+      "step": 3248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000439986228103333,
+      "loss": 2.0898,
+      "step": 3249
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004394093838988602,
+      "loss": 2.1973,
+      "step": 3250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043883281158855857,
+      "loss": 2.1777,
+      "step": 3251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043825651145207355,
+      "loss": 2.0605,
+      "step": 3252
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004376804837689188,
+      "loss": 2.0811,
+      "step": 3253
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004371047288184752,
+      "loss": 2.1953,
+      "step": 3254
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043652924687999253,
+      "loss": 2.1787,
+      "step": 3255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004359540382325872,
+      "loss": 2.168,
+      "step": 3256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004353791031552433,
+      "loss": 2.3574,
+      "step": 3257
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004348044419268122,
+      "loss": 2.1533,
+      "step": 3258
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004342300548260122,
+      "loss": 2.0576,
+      "step": 3259
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004336559421314298,
+      "loss": 2.1875,
+      "step": 3260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043308210412151696,
+      "loss": 2.2168,
+      "step": 3261
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043250854107459313,
+      "loss": 2.2158,
+      "step": 3262
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004319352532688443,
+      "loss": 2.042,
+      "step": 3263
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043136224098232304,
+      "loss": 2.21,
+      "step": 3264
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043078950449294763,
+      "loss": 2.1016,
+      "step": 3265
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043021704407850413,
+      "loss": 2.1611,
+      "step": 3266
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00042964486001664315,
+      "loss": 2.1006,
+      "step": 3267
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000429072952584882,
+      "loss": 2.125,
+      "step": 3268
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004285013220606034,
+      "loss": 2.1865,
+      "step": 3269
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004279299687210565,
+      "loss": 2.0986,
+      "step": 3270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004273588928433547,
+      "loss": 2.0312,
+      "step": 3271
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000426788094704479,
+      "loss": 2.0732,
+      "step": 3272
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00042621757458127286,
+      "loss": 2.1768,
+      "step": 3273
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00042564733275044756,
+      "loss": 2.2314,
+      "step": 3274
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004250773694885778,
+      "loss": 2.1914,
+      "step": 3275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004245076850721031,
+      "loss": 2.2012,
+      "step": 3276
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00042393827977732935,
+      "loss": 2.1699,
+      "step": 3277
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00042336915388042565,
+      "loss": 2.2188,
+      "step": 3278
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00042280030765742496,
+      "loss": 2.3262,
+      "step": 3279
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004222317413842265,
+      "loss": 2.2334,
+      "step": 3280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00042166345533659245,
+      "loss": 2.0811,
+      "step": 3281
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00042109544979014905,
+      "loss": 2.1445,
+      "step": 3282
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004205277250203878,
+      "loss": 2.1016,
+      "step": 3283
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004199602813026622,
+      "loss": 2.2207,
+      "step": 3284
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004193931189121899,
+      "loss": 2.252,
+      "step": 3285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004188262381240534,
+      "loss": 2.2002,
+      "step": 3286
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041825963921319744,
+      "loss": 2.1738,
+      "step": 3287
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041769332245443003,
+      "loss": 2.1738,
+      "step": 3288
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004171272881224227,
+      "loss": 2.1855,
+      "step": 3289
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004165615364917097,
+      "loss": 2.1602,
+      "step": 3290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041599606783668887,
+      "loss": 2.252,
+      "step": 3291
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041543088243162007,
+      "loss": 2.1631,
+      "step": 3292
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000414865980550626,
+      "loss": 2.1543,
+      "step": 3293
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041430136246769177,
+      "loss": 2.2217,
+      "step": 3294
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041373702845666493,
+      "loss": 2.1465,
+      "step": 3295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041317297879125503,
+      "loss": 2.1113,
+      "step": 3296
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041260921374503445,
+      "loss": 2.1357,
+      "step": 3297
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004120457335914368,
+      "loss": 2.2979,
+      "step": 3298
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041148253860375764,
+      "loss": 2.0811,
+      "step": 3299
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041091962905515436,
+      "loss": 2.0518,
+      "step": 3300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004103570052186457,
+      "loss": 2.1084,
+      "step": 3301
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040979466736711177,
+      "loss": 2.1416,
+      "step": 3302
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040923261577329486,
+      "loss": 2.082,
+      "step": 3303
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004086708507097974,
+      "loss": 2.1104,
+      "step": 3304
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004081093724490831,
+      "loss": 2.1475,
+      "step": 3305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004075481812634766,
+      "loss": 2.2598,
+      "step": 3306
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004069872774251632,
+      "loss": 2.0918,
+      "step": 3307
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040642666120618956,
+      "loss": 1.9844,
+      "step": 3308
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004058663328784622,
+      "loss": 2.167,
+      "step": 3309
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004053062927137472,
+      "loss": 2.1914,
+      "step": 3310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004047465409836728,
+      "loss": 2.1406,
+      "step": 3311
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004041870779597258,
+      "loss": 2.1074,
+      "step": 3312
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004036279039132531,
+      "loss": 2.2383,
+      "step": 3313
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004030690191154628,
+      "loss": 2.2529,
+      "step": 3314
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040251042383742107,
+      "loss": 2.1465,
+      "step": 3315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004019521183500544,
+      "loss": 2.0977,
+      "step": 3316
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004013941029241488,
+      "loss": 2.043,
+      "step": 3317
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004008363778303492,
+      "loss": 2.1582,
+      "step": 3318
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040027894333915983,
+      "loss": 2.0215,
+      "step": 3319
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003997217997209449,
+      "loss": 2.1748,
+      "step": 3320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00039916494724592555,
+      "loss": 2.1445,
+      "step": 3321
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003986083861841839,
+      "loss": 2.0479,
+      "step": 3322
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00039805211680565936,
+      "loss": 2.1289,
+      "step": 3323
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00039749613938014984,
+      "loss": 2.2656,
+      "step": 3324
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003969404541773132,
+      "loss": 1.9541,
+      "step": 3325
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003963850614666633,
+      "loss": 2.1436,
+      "step": 3326
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00039582996151757324,
+      "loss": 2.0254,
+      "step": 3327
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003952751545992749,
+      "loss": 2.125,
+      "step": 3328
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003947206409808569,
+      "loss": 2.0488,
+      "step": 3329
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00039416642093126596,
+      "loss": 2.2354,
+      "step": 3330
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003936124947193065,
+      "loss": 2.1475,
+      "step": 3331
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003930588626136403,
+      "loss": 2.0459,
+      "step": 3332
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00039250552488278635,
+      "loss": 2.3086,
+      "step": 3333
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00039195248179512153,
+      "loss": 2.0596,
+      "step": 3334
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00039139973361887914,
+      "loss": 2.0146,
+      "step": 3335
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003908472806221497,
+      "loss": 2.292,
+      "step": 3336
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00039029512307288053,
+      "loss": 2.2754,
+      "step": 3337
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038974326123887516,
+      "loss": 2.1211,
+      "step": 3338
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000389191695387795,
+      "loss": 2.0947,
+      "step": 3339
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038864042578715666,
+      "loss": 2.043,
+      "step": 3340
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003880894527043335,
+      "loss": 2.1152,
+      "step": 3341
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003875387764065552,
+      "loss": 2.1641,
+      "step": 3342
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003869883971609072,
+      "loss": 2.2578,
+      "step": 3343
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003864383152343307,
+      "loss": 2.2539,
+      "step": 3344
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038588853089362363,
+      "loss": 2.1709,
+      "step": 3345
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038533904440543875,
+      "loss": 2.2881,
+      "step": 3346
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003847898560362846,
+      "loss": 2.1279,
+      "step": 3347
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000384240966052525,
+      "loss": 2.1641,
+      "step": 3348
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003836923747203792,
+      "loss": 2.2041,
+      "step": 3349
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003831440823059212,
+      "loss": 2.4551,
+      "step": 3350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038259608907508105,
+      "loss": 2.166,
+      "step": 3351
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038204839529364267,
+      "loss": 2.2988,
+      "step": 3352
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038150100122724505,
+      "loss": 2.0195,
+      "step": 3353
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000380953907141382,
+      "loss": 2.1006,
+      "step": 3354
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003804071133014012,
+      "loss": 2.0352,
+      "step": 3355
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003798606199725059,
+      "loss": 2.0059,
+      "step": 3356
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037931442741975297,
+      "loss": 2.0225,
+      "step": 3357
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003787685359080523,
+      "loss": 2.0918,
+      "step": 3358
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003782229457021696,
+      "loss": 2.249,
+      "step": 3359
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037767765706672366,
+      "loss": 2.085,
+      "step": 3360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037713267026618627,
+      "loss": 2.1895,
+      "step": 3361
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003765879855648848,
+      "loss": 2.2256,
+      "step": 3362
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003760436032269974,
+      "loss": 2.3545,
+      "step": 3363
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003754995235165579,
+      "loss": 2.1738,
+      "step": 3364
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037495574669745235,
+      "loss": 2.2812,
+      "step": 3365
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037441227303341995,
+      "loss": 2.1357,
+      "step": 3366
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003738691027880524,
+      "loss": 2.125,
+      "step": 3367
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003733262362247959,
+      "loss": 2.1191,
+      "step": 3368
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037278367360694687,
+      "loss": 2.1191,
+      "step": 3369
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003722414151976565,
+      "loss": 2.1211,
+      "step": 3370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037169946125992757,
+      "loss": 2.166,
+      "step": 3371
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003711578120566146,
+      "loss": 2.1816,
+      "step": 3372
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00037061646785042603,
+      "loss": 2.0469,
+      "step": 3373
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003700754289039201,
+      "loss": 2.1855,
+      "step": 3374
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003695346954795084,
+      "loss": 2.165,
+      "step": 3375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036899426783945465,
+      "loss": 2.2656,
+      "step": 3376
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036845414624587326,
+      "loss": 2.1719,
+      "step": 3377
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003679143309607308,
+      "loss": 2.2227,
+      "step": 3378
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036737482224584504,
+      "loss": 2.2158,
+      "step": 3379
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036683562036288507,
+      "loss": 2.1904,
+      "step": 3380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036629672557337114,
+      "loss": 2.2314,
+      "step": 3381
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003657581381386751,
+      "loss": 2.0869,
+      "step": 3382
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003652198583200189,
+      "loss": 2.1709,
+      "step": 3383
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003646818863784759,
+      "loss": 2.1953,
+      "step": 3384
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003641442225749696,
+      "loss": 2.0059,
+      "step": 3385
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036360686717027427,
+      "loss": 2.1709,
+      "step": 3386
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036306982042501524,
+      "loss": 2.0312,
+      "step": 3387
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036253308259966735,
+      "loss": 2.2393,
+      "step": 3388
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003619966539545557,
+      "loss": 2.1035,
+      "step": 3389
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036146053474985564,
+      "loss": 2.1934,
+      "step": 3390
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003609247252455923,
+      "loss": 2.2129,
+      "step": 3391
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036038922570164045,
+      "loss": 2.2051,
+      "step": 3392
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003598540363777254,
+      "loss": 2.2012,
+      "step": 3393
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003593191575334209,
+      "loss": 2.0918,
+      "step": 3394
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003587845894281506,
+      "loss": 2.1426,
+      "step": 3395
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035825033232118756,
+      "loss": 2.1689,
+      "step": 3396
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003577163864716536,
+      "loss": 2.1758,
+      "step": 3397
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003571827521385196,
+      "loss": 2.1396,
+      "step": 3398
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035664942958060655,
+      "loss": 2.1064,
+      "step": 3399
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003561164190565819,
+      "loss": 2.2295,
+      "step": 3400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035558372082496404,
+      "loss": 2.2324,
+      "step": 3401
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003550513351441187,
+      "loss": 2.0527,
+      "step": 3402
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003545192622722599,
+      "loss": 2.1758,
+      "step": 3403
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035398750246745105,
+      "loss": 2.2305,
+      "step": 3404
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003534560559876028,
+      "loss": 2.1514,
+      "step": 3405
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003529249230904732,
+      "loss": 2.4102,
+      "step": 3406
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003523941040336699,
+      "loss": 2.0879,
+      "step": 3407
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035186359907464726,
+      "loss": 2.0332,
+      "step": 3408
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003513334084707069,
+      "loss": 2.2549,
+      "step": 3409
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000350803532479,
+      "loss": 2.1719,
+      "step": 3410
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003502739713565219,
+      "loss": 2.1494,
+      "step": 3411
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00034974472536011836,
+      "loss": 2.0215,
+      "step": 3412
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003492157947464805,
+      "loss": 2.085,
+      "step": 3413
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000348687179772147,
+      "loss": 2.0615,
+      "step": 3414
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003481588806935034,
+      "loss": 2.1123,
+      "step": 3415
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000347630897766782,
+      "loss": 2.1084,
+      "step": 3416
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003471032312480615,
+      "loss": 2.0947,
+      "step": 3417
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003465758813932682,
+      "loss": 2.2285,
+      "step": 3418
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003460488484581735,
+      "loss": 2.0869,
+      "step": 3419
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003455221326983955,
+      "loss": 2.0146,
+      "step": 3420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00034499573436939934,
+      "loss": 2.0986,
+      "step": 3421
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00034446965372649473,
+      "loss": 2.2559,
+      "step": 3422
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00034394389102483793,
+      "loss": 2.0449,
+      "step": 3423
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003434184465194317,
+      "loss": 2.2041,
+      "step": 3424
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00034289332046512355,
+      "loss": 2.2617,
+      "step": 3425
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003423685131166069,
+      "loss": 2.1475,
+      "step": 3426
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003418440247284206,
+      "loss": 2.1406,
+      "step": 3427
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003413198555549486,
+      "loss": 2.3066,
+      "step": 3428
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003407960058504199,
+      "loss": 2.2129,
+      "step": 3429
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003402724758689094,
+      "loss": 2.166,
+      "step": 3430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033974926586433627,
+      "loss": 2.0527,
+      "step": 3431
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033922637609046436,
+      "loss": 2.0029,
+      "step": 3432
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003387038068009024,
+      "loss": 2.1504,
+      "step": 3433
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003381815582491036,
+      "loss": 2.1387,
+      "step": 3434
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033765963068836603,
+      "loss": 2.1084,
+      "step": 3435
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033713802437183193,
+      "loss": 2.1064,
+      "step": 3436
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003366167395524865,
+      "loss": 2.2207,
+      "step": 3437
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033609577648316115,
+      "loss": 2.1377,
+      "step": 3438
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003355751354165297,
+      "loss": 2.0322,
+      "step": 3439
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003350548166051101,
+      "loss": 2.1982,
+      "step": 3440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033453482030126446,
+      "loss": 2.1504,
+      "step": 3441
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033401514675719814,
+      "loss": 2.2676,
+      "step": 3442
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033349579622495983,
+      "loss": 2.1465,
+      "step": 3443
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033297676895644167,
+      "loss": 2.3213,
+      "step": 3444
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003324580652033791,
+      "loss": 2.1387,
+      "step": 3445
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003319396852173503,
+      "loss": 2.2363,
+      "step": 3446
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003314216292497775,
+      "loss": 2.1475,
+      "step": 3447
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033090389755192386,
+      "loss": 2.2344,
+      "step": 3448
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003303864903748973,
+      "loss": 2.2812,
+      "step": 3449
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032986940796964705,
+      "loss": 2.1504,
+      "step": 3450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032935265058696493,
+      "loss": 2.3252,
+      "step": 3451
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032883621847748637,
+      "loss": 2.0537,
+      "step": 3452
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003283201118916871,
+      "loss": 2.1309,
+      "step": 3453
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032780433107988585,
+      "loss": 2.2314,
+      "step": 3454
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032728887629224413,
+      "loss": 2.0908,
+      "step": 3455
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032677374777876435,
+      "loss": 2.2168,
+      "step": 3456
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032625894578929037,
+      "loss": 2.1631,
+      "step": 3457
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003257444705735095,
+      "loss": 2.1016,
+      "step": 3458
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032523032238094787,
+      "loss": 2.3467,
+      "step": 3459
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003247165014609754,
+      "loss": 2.3242,
+      "step": 3460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000324203008062802,
+      "loss": 2.2939,
+      "step": 3461
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003236898424354792,
+      "loss": 2.1055,
+      "step": 3462
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032317700482789916,
+      "loss": 2.1035,
+      "step": 3463
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003226644954887952,
+      "loss": 1.9746,
+      "step": 3464
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003221523146667412,
+      "loss": 2.3193,
+      "step": 3465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00032164046261015254,
+      "loss": 2.2021,
+      "step": 3466
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003211289395672843,
+      "loss": 2.1006,
+      "step": 3467
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000320617745786232,
+      "loss": 2.3301,
+      "step": 3468
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003201068815149317,
+      "loss": 2.0225,
+      "step": 3469
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003195963470011598,
+      "loss": 2.1074,
+      "step": 3470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000319086142492532,
+      "loss": 1.9834,
+      "step": 3471
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003185762682365053,
+      "loss": 2.1172,
+      "step": 3472
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003180667244803754,
+      "loss": 2.1367,
+      "step": 3473
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000317557511471278,
+      "loss": 2.0625,
+      "step": 3474
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003170486294561885,
+      "loss": 2.0693,
+      "step": 3475
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003165400786819216,
+      "loss": 2.0928,
+      "step": 3476
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003160318593951311,
+      "loss": 2.0996,
+      "step": 3477
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031552397184231106,
+      "loss": 2.1172,
+      "step": 3478
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003150164162697937,
+      "loss": 2.0967,
+      "step": 3479
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031450919292375036,
+      "loss": 2.167,
+      "step": 3480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003140023020501912,
+      "loss": 2.3271,
+      "step": 3481
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003134957438949653,
+      "loss": 1.9746,
+      "step": 3482
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031298951870376067,
+      "loss": 2.2148,
+      "step": 3483
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031248362672210375,
+      "loss": 2.0762,
+      "step": 3484
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003119780681953579,
+      "loss": 2.1104,
+      "step": 3485
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031147284336872715,
+      "loss": 2.0977,
+      "step": 3486
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003109679524872521,
+      "loss": 2.1543,
+      "step": 3487
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003104633957958115,
+      "loss": 2.1299,
+      "step": 3488
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003099591735391232,
+      "loss": 2.1182,
+      "step": 3489
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030945528596174054,
+      "loss": 1.998,
+      "step": 3490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030895173330805714,
+      "loss": 2.0693,
+      "step": 3491
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003084485158223026,
+      "loss": 2.1953,
+      "step": 3492
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003079456337485442,
+      "loss": 2.0703,
+      "step": 3493
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003074430873306865,
+      "loss": 2.0439,
+      "step": 3494
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003069408768124723,
+      "loss": 2.168,
+      "step": 3495
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030643900243747935,
+      "loss": 2.0146,
+      "step": 3496
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003059374644491246,
+      "loss": 2.043,
+      "step": 3497
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030543626309066066,
+      "loss": 2.2617,
+      "step": 3498
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003049353986051766,
+      "loss": 2.123,
+      "step": 3499
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030443487123559974,
+      "loss": 2.1133,
+      "step": 3500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003039346812246916,
+      "loss": 2.1904,
+      "step": 3501
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030343482881505136,
+      "loss": 2.1289,
+      "step": 3502
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003029353142491149,
+      "loss": 2.127,
+      "step": 3503
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030243613776915346,
+      "loss": 1.998,
+      "step": 3504
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030193729961727436,
+      "loss": 2.2031,
+      "step": 3505
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000301438800035421,
+      "loss": 2.3281,
+      "step": 3506
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003009406392653723,
+      "loss": 2.2695,
+      "step": 3507
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003004428175487437,
+      "loss": 2.1816,
+      "step": 3508
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002999453351269853,
+      "loss": 2.0391,
+      "step": 3509
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029944819224138276,
+      "loss": 2.1709,
+      "step": 3510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029895138913305753,
+      "loss": 2.085,
+      "step": 3511
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029845492604296564,
+      "loss": 2.0762,
+      "step": 3512
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002979588032118985,
+      "loss": 2.2217,
+      "step": 3513
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002974630208804829,
+      "loss": 2.1045,
+      "step": 3514
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029696757928917997,
+      "loss": 2.1104,
+      "step": 3515
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029647247867828553,
+      "loss": 2.1543,
+      "step": 3516
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002959777192879304,
+      "loss": 2.2002,
+      "step": 3517
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029548330135807963,
+      "loss": 2.1406,
+      "step": 3518
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002949892251285323,
+      "loss": 2.1455,
+      "step": 3519
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002944954908389229,
+      "loss": 2.1318,
+      "step": 3520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029400209872871907,
+      "loss": 2.0,
+      "step": 3521
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029350904903722275,
+      "loss": 2.2812,
+      "step": 3522
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029301634200356983,
+      "loss": 2.1924,
+      "step": 3523
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002925239778667301,
+      "loss": 2.0557,
+      "step": 3524
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029203195686550653,
+      "loss": 2.1299,
+      "step": 3525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000291540279238537,
+      "loss": 2.21,
+      "step": 3526
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002910489452242907,
+      "loss": 2.168,
+      "step": 3527
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002905579550610724,
+      "loss": 2.2051,
+      "step": 3528
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002900673089870187,
+      "loss": 2.2412,
+      "step": 3529
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002895770072400995,
+      "loss": 2.1562,
+      "step": 3530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002890870500581184,
+      "loss": 2.2139,
+      "step": 3531
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002885974376787115,
+      "loss": 2.251,
+      "step": 3532
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028810817033934655,
+      "loss": 2.2061,
+      "step": 3533
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028761924827732575,
+      "loss": 2.1406,
+      "step": 3534
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028713067172978304,
+      "loss": 2.1143,
+      "step": 3535
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002866424409336842,
+      "loss": 2.085,
+      "step": 3536
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000286154556125829,
+      "loss": 2.1396,
+      "step": 3537
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028566701754284687,
+      "loss": 2.1602,
+      "step": 3538
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002851798254212017,
+      "loss": 2.1494,
+      "step": 3539
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028469297999718824,
+      "loss": 2.165,
+      "step": 3540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002842064815069331,
+      "loss": 2.1152,
+      "step": 3541
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002837203301863948,
+      "loss": 2.0801,
+      "step": 3542
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028323452627136323,
+      "loss": 2.1943,
+      "step": 3543
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028274906999746,
+      "loss": 2.1533,
+      "step": 3544
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028226396160013854,
+      "loss": 2.2188,
+      "step": 3545
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002817792013146827,
+      "loss": 2.0947,
+      "step": 3546
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028129478937620776,
+      "loss": 2.1201,
+      "step": 3547
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028081072601966105,
+      "loss": 2.0996,
+      "step": 3548
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028032701147981897,
+      "loss": 2.2275,
+      "step": 3549
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00027984364599128986,
+      "loss": 2.1406,
+      "step": 3550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00027936062978851315,
+      "loss": 2.2607,
+      "step": 3551
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002788779631057581,
+      "loss": 2.042,
+      "step": 3552
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00027839564617712463,
+      "loss": 2.1416,
+      "step": 3553
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00027791367923654333,
+      "loss": 2.1289,
+      "step": 3554
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002774320625177741,
+      "loss": 2.1045,
+      "step": 3555
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002769507962544087,
+      "loss": 2.1973,
+      "step": 3556
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00027646988067986725,
+      "loss": 2.123,
+      "step": 3557
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002759893160274006,
+      "loss": 2.2529,
+      "step": 3558
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002755091025300893,
+      "loss": 2.1611,
+      "step": 3559
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002750292404208433,
+      "loss": 2.3574,
+      "step": 3560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00027454972993240224,
+      "loss": 2.2275,
+      "step": 3561
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00027407057129733584,
+      "loss": 2.1709,
+      "step": 3562
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002735917647480426,
+      "loss": 2.2441,
+      "step": 3563
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002731133105167495,
+      "loss": 2.2715,
+      "step": 3564
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00027263520883551406,
+      "loss": 2.0811,
+      "step": 3565
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002721574599362222,
+      "loss": 2.1885,
+      "step": 3566
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002716800640505883,
+      "loss": 2.1133,
+      "step": 3567
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002712030214101565,
+      "loss": 2.1816,
+      "step": 3568
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002707263322462987,
+      "loss": 2.1611,
+      "step": 3569
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002702499967902159,
+      "loss": 2.002,
+      "step": 3570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026977401527293723,
+      "loss": 2.1465,
+      "step": 3571
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026929838792532035,
+      "loss": 2.1211,
+      "step": 3572
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002688231149780507,
+      "loss": 2.0098,
+      "step": 3573
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002683481966616431,
+      "loss": 2.1797,
+      "step": 3574
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002678736332064383,
+      "loss": 2.083,
+      "step": 3575
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026739942484260704,
+      "loss": 2.2227,
+      "step": 3576
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026692557180014667,
+      "loss": 2.0664,
+      "step": 3577
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026645207430888194,
+      "loss": 2.1094,
+      "step": 3578
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002659789325984667,
+      "loss": 2.1533,
+      "step": 3579
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002655061468983804,
+      "loss": 2.1875,
+      "step": 3580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002650337174379304,
+      "loss": 2.0293,
+      "step": 3581
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002645616444462522,
+      "loss": 1.9512,
+      "step": 3582
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026408992815230746,
+      "loss": 2.2529,
+      "step": 3583
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026361856878488467,
+      "loss": 2.2529,
+      "step": 3584
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002631475665726005,
+      "loss": 2.1621,
+      "step": 3585
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002626769217438965,
+      "loss": 2.1758,
+      "step": 3586
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026220663452704263,
+      "loss": 2.2471,
+      "step": 3587
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026173670515013434,
+      "loss": 1.9492,
+      "step": 3588
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000261267133841094,
+      "loss": 2.0947,
+      "step": 3589
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002607979208276701,
+      "loss": 2.083,
+      "step": 3590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026032906633743734,
+      "loss": 2.124,
+      "step": 3591
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00025986057059779633,
+      "loss": 2.1377,
+      "step": 3592
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002593924338359747,
+      "loss": 2.1484,
+      "step": 3593
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002589246562790247,
+      "loss": 2.1396,
+      "step": 3594
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00025845723815382513,
+      "loss": 2.1133,
+      "step": 3595
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00025799017968708004,
+      "loss": 2.1572,
+      "step": 3596
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00025752348110531933,
+      "loss": 1.958,
+      "step": 3597
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00025705714263489775,
+      "loss": 2.1836,
+      "step": 3598
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002565911645019967,
+      "loss": 2.1055,
+      "step": 3599
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002561255469326217,
+      "loss": 2.1143,
+      "step": 3600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002556602901526035,
+      "loss": 2.251,
+      "step": 3601
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002551953943875982,
+      "loss": 2.1836,
+      "step": 3602
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00025473085986308633,
+      "loss": 2.1113,
+      "step": 3603
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000254266686804374,
+      "loss": 2.1602,
+      "step": 3604
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002538028754365914,
+      "loss": 2.1475,
+      "step": 3605
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00025333942598469337,
+      "loss": 2.252,
+      "step": 3606
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00025287633867345936,
+      "loss": 2.1289,
+      "step": 3607
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002524136137274929,
+      "loss": 2.0527,
+      "step": 3608
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002519512513712219,
+      "loss": 2.1035,
+      "step": 3609
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002514892518288988,
+      "loss": 2.1318,
+      "step": 3610
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00025102761532460005,
+      "loss": 2.0332,
+      "step": 3611
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002505663420822245,
+      "loss": 2.2588,
+      "step": 3612
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002501054323254972,
+      "loss": 1.959,
+      "step": 3613
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024964488627796533,
+      "loss": 2.3613,
+      "step": 3614
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024918470416299967,
+      "loss": 2.001,
+      "step": 3615
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000248724886203796,
+      "loss": 2.1221,
+      "step": 3616
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002482654326233709,
+      "loss": 2.1758,
+      "step": 3617
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002478063436445667,
+      "loss": 2.0762,
+      "step": 3618
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024734761949004767,
+      "loss": 2.0957,
+      "step": 3619
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002468892603823012,
+      "loss": 2.2334,
+      "step": 3620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024643126654363767,
+      "loss": 2.1572,
+      "step": 3621
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000245973638196191,
+      "loss": 2.1406,
+      "step": 3622
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002455163755619164,
+      "loss": 2.2061,
+      "step": 3623
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002450594788625932,
+      "loss": 2.2168,
+      "step": 3624
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002446029483198223,
+      "loss": 2.0654,
+      "step": 3625
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002441467841550271,
+      "loss": 2.1582,
+      "step": 3626
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024369098658945433,
+      "loss": 2.1104,
+      "step": 3627
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024323555584417124,
+      "loss": 2.1152,
+      "step": 3628
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024278049214006804,
+      "loss": 2.2168,
+      "step": 3629
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002423257956978574,
+      "loss": 2.334,
+      "step": 3630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024187146673807302,
+      "loss": 2.0137,
+      "step": 3631
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000241417505481071,
+      "loss": 2.084,
+      "step": 3632
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002409639121470284,
+      "loss": 2.0254,
+      "step": 3633
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024051068695594424,
+      "loss": 2.0449,
+      "step": 3634
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002400578301276395,
+      "loss": 2.1055,
+      "step": 3635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023960534188175554,
+      "loss": 1.9268,
+      "step": 3636
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002391532224377556,
+      "loss": 2.248,
+      "step": 3637
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023870147201492364,
+      "loss": 2.083,
+      "step": 3638
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023825009083236503,
+      "loss": 2.1514,
+      "step": 3639
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023779907910900544,
+      "loss": 2.1846,
+      "step": 3640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023734843706359245,
+      "loss": 2.0469,
+      "step": 3641
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023689816491469318,
+      "loss": 2.126,
+      "step": 3642
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002364482628806961,
+      "loss": 2.207,
+      "step": 3643
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023599873117980964,
+      "loss": 2.0059,
+      "step": 3644
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023554957003006305,
+      "loss": 2.1035,
+      "step": 3645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023510077964930532,
+      "loss": 2.2051,
+      "step": 3646
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023465236025520654,
+      "loss": 2.0811,
+      "step": 3647
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023420431206525594,
+      "loss": 2.1504,
+      "step": 3648
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023375663529676317,
+      "loss": 2.1533,
+      "step": 3649
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023330933016685752,
+      "loss": 2.1553,
+      "step": 3650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000232862396892488,
+      "loss": 2.2314,
+      "step": 3651
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002324158356904239,
+      "loss": 2.1387,
+      "step": 3652
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00023196964677725353,
+      "loss": 2.1533,
+      "step": 3653
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00023152383036938385,
+      "loss": 2.2236,
+      "step": 3654
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002310783866830427,
+      "loss": 2.0664,
+      "step": 3655
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00023063331593427627,
+      "loss": 2.0195,
+      "step": 3656
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00023018861833894966,
+      "loss": 2.0293,
+      "step": 3657
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002297442941127479,
+      "loss": 2.1182,
+      "step": 3658
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022930034347117434,
+      "loss": 2.1719,
+      "step": 3659
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022885676662955025,
+      "loss": 2.2412,
+      "step": 3660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022841356380301757,
+      "loss": 2.0176,
+      "step": 3661
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022797073520653522,
+      "loss": 2.0898,
+      "step": 3662
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002275282810548811,
+      "loss": 2.2412,
+      "step": 3663
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022708620156265235,
+      "loss": 2.1182,
+      "step": 3664
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022664449694426248,
+      "loss": 2.1367,
+      "step": 3665
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022620316741394508,
+      "loss": 1.9453,
+      "step": 3666
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022576221318575086,
+      "loss": 2.1143,
+      "step": 3667
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022532163447354868,
+      "loss": 2.1143,
+      "step": 3668
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022488143149102535,
+      "loss": 2.1943,
+      "step": 3669
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022444160445168527,
+      "loss": 2.0537,
+      "step": 3670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022400215356885035,
+      "loss": 2.1348,
+      "step": 3671
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022356307905566086,
+      "loss": 2.166,
+      "step": 3672
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022312438112507383,
+      "loss": 2.1035,
+      "step": 3673
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022268605998986346,
+      "loss": 2.1484,
+      "step": 3674
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002222481158626223,
+      "loss": 2.1426,
+      "step": 3675
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022181054895575847,
+      "loss": 2.1611,
+      "step": 3676
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022137335948149806,
+      "loss": 2.1084,
+      "step": 3677
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022093654765188454,
+      "loss": 2.0332,
+      "step": 3678
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002205001136787772,
+      "loss": 2.1719,
+      "step": 3679
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022006405777385264,
+      "loss": 2.0635,
+      "step": 3680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021962838014860398,
+      "loss": 2.1182,
+      "step": 3681
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021919308101434065,
+      "loss": 2.1904,
+      "step": 3682
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021875816058218922,
+      "loss": 2.1152,
+      "step": 3683
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021832361906309184,
+      "loss": 2.2686,
+      "step": 3684
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021788945666780714,
+      "loss": 2.1768,
+      "step": 3685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021745567360690978,
+      "loss": 2.123,
+      "step": 3686
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021702227009079056,
+      "loss": 2.1367,
+      "step": 3687
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021658924632965594,
+      "loss": 2.0605,
+      "step": 3688
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021615660253352886,
+      "loss": 1.9941,
+      "step": 3689
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002157243389122474,
+      "loss": 2.2188,
+      "step": 3690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021529245567546517,
+      "loss": 2.1514,
+      "step": 3691
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021486095303265175,
+      "loss": 2.2695,
+      "step": 3692
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021442983119309157,
+      "loss": 2.2266,
+      "step": 3693
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002139990903658845,
+      "loss": 2.0664,
+      "step": 3694
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021356873075994642,
+      "loss": 2.1748,
+      "step": 3695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021313875258400738,
+      "loss": 2.083,
+      "step": 3696
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021270915604661256,
+      "loss": 2.1816,
+      "step": 3697
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002122799413561224,
+      "loss": 1.9805,
+      "step": 3698
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00021185110872071146,
+      "loss": 2.1953,
+      "step": 3699
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00021142265834837016,
+      "loss": 1.9941,
+      "step": 3700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00021099459044690284,
+      "loss": 2.0352,
+      "step": 3701
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002105669052239274,
+      "loss": 2.1396,
+      "step": 3702
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00021013960288687795,
+      "loss": 2.1133,
+      "step": 3703
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020971268364300177,
+      "loss": 2.1973,
+      "step": 3704
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020928614769936038,
+      "loss": 2.1035,
+      "step": 3705
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020885999526283017,
+      "loss": 2.124,
+      "step": 3706
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020843422654010025,
+      "loss": 2.0479,
+      "step": 3707
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002080088417376743,
+      "loss": 2.2168,
+      "step": 3708
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020758384106187044,
+      "loss": 2.1152,
+      "step": 3709
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020715922471881953,
+      "loss": 2.1582,
+      "step": 3710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020673499291446596,
+      "loss": 2.25,
+      "step": 3711
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000206311145854569,
+      "loss": 2.0527,
+      "step": 3712
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020588768374469902,
+      "loss": 2.0039,
+      "step": 3713
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020546460679024203,
+      "loss": 2.1709,
+      "step": 3714
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002050419151963957,
+      "loss": 2.2021,
+      "step": 3715
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020461960916817146,
+      "loss": 2.2822,
+      "step": 3716
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020419768891039337,
+      "loss": 2.085,
+      "step": 3717
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002037761546276986,
+      "loss": 2.1143,
+      "step": 3718
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020335500652453697,
+      "loss": 2.0654,
+      "step": 3719
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020293424480517154,
+      "loss": 2.1152,
+      "step": 3720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020251386967367723,
+      "loss": 2.1064,
+      "step": 3721
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020209388133394179,
+      "loss": 2.2734,
+      "step": 3722
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020167427998966537,
+      "loss": 2.0801,
+      "step": 3723
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002012550658443605,
+      "loss": 2.2422,
+      "step": 3724
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002008362391013514,
+      "loss": 2.0537,
+      "step": 3725
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020041779996377541,
+      "loss": 1.959,
+      "step": 3726
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019999974863458102,
+      "loss": 2.2051,
+      "step": 3727
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019958208531652876,
+      "loss": 2.1309,
+      "step": 3728
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019916481021219136,
+      "loss": 2.1084,
+      "step": 3729
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019874792352395242,
+      "loss": 2.2832,
+      "step": 3730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001983314254540084,
+      "loss": 2.3115,
+      "step": 3731
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019791531620436643,
+      "loss": 2.1201,
+      "step": 3732
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001974995959768451,
+      "loss": 2.2158,
+      "step": 3733
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019708426497307442,
+      "loss": 2.1367,
+      "step": 3734
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019666932339449584,
+      "loss": 2.1309,
+      "step": 3735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019625477144236125,
+      "loss": 2.2119,
+      "step": 3736
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001958406093177346,
+      "loss": 2.0967,
+      "step": 3737
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019542683722149047,
+      "loss": 2.1826,
+      "step": 3738
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019501345535431302,
+      "loss": 1.998,
+      "step": 3739
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019460046391669893,
+      "loss": 2.1699,
+      "step": 3740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019418786310895463,
+      "loss": 2.0996,
+      "step": 3741
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019377565313119684,
+      "loss": 2.2266,
+      "step": 3742
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001933638341833538,
+      "loss": 2.0361,
+      "step": 3743
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019295240646516242,
+      "loss": 2.2256,
+      "step": 3744
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019254137017617157,
+      "loss": 1.9248,
+      "step": 3745
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001921307255157393,
+      "loss": 1.9844,
+      "step": 3746
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001917204726830335,
+      "loss": 2.0352,
+      "step": 3747
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019131061187703313,
+      "loss": 2.0547,
+      "step": 3748
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001909011432965263,
+      "loss": 2.0977,
+      "step": 3749
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019049206714011002,
+      "loss": 1.9824,
+      "step": 3750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001900833836061928,
+      "loss": 2.2773,
+      "step": 3751
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018967509289299147,
+      "loss": 2.0908,
+      "step": 3752
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001892671951985323,
+      "loss": 2.1094,
+      "step": 3753
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018885969072065224,
+      "loss": 2.1777,
+      "step": 3754
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001884525796569957,
+      "loss": 2.0225,
+      "step": 3755
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018804586220501706,
+      "loss": 2.1211,
+      "step": 3756
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018763953856198057,
+      "loss": 2.167,
+      "step": 3757
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018723360892495844,
+      "loss": 2.0195,
+      "step": 3758
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001868280734908321,
+      "loss": 2.1104,
+      "step": 3759
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018642293245629195,
+      "loss": 2.1641,
+      "step": 3760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001860181860178366,
+      "loss": 2.0898,
+      "step": 3761
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018561383437177425,
+      "loss": 2.334,
+      "step": 3762
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001852098777142207,
+      "loss": 2.0645,
+      "step": 3763
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018480631624110056,
+      "loss": 2.1348,
+      "step": 3764
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018440315014814669,
+      "loss": 2.252,
+      "step": 3765
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018400037963089999,
+      "loss": 2.1416,
+      "step": 3766
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018359800488470978,
+      "loss": 2.1162,
+      "step": 3767
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018319602610473374,
+      "loss": 2.1348,
+      "step": 3768
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018279444348593676,
+      "loss": 2.1895,
+      "step": 3769
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018239325722309196,
+      "loss": 2.1982,
+      "step": 3770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018199246751078035,
+      "loss": 1.959,
+      "step": 3771
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018159207454339043,
+      "loss": 2.1016,
+      "step": 3772
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018119207851511788,
+      "loss": 1.9873,
+      "step": 3773
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001807924796199669,
+      "loss": 2.1621,
+      "step": 3774
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001803932780517483,
+      "loss": 2.1094,
+      "step": 3775
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001799944740040802,
+      "loss": 2.0596,
+      "step": 3776
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017959606767038817,
+      "loss": 2.124,
+      "step": 3777
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017919805924390442,
+      "loss": 2.1895,
+      "step": 3778
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000178800448917669,
+      "loss": 2.1133,
+      "step": 3779
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017840323688452832,
+      "loss": 2.1914,
+      "step": 3780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017800642333713545,
+      "loss": 2.1436,
+      "step": 3781
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001776100084679506,
+      "loss": 2.1934,
+      "step": 3782
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001772139924692404,
+      "loss": 2.1807,
+      "step": 3783
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017681837553307778,
+      "loss": 2.2178,
+      "step": 3784
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000176423157851343,
+      "loss": 2.1211,
+      "step": 3785
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017602833961572197,
+      "loss": 2.2529,
+      "step": 3786
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017563392101770626,
+      "loss": 2.2578,
+      "step": 3787
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017523990224859498,
+      "loss": 2.0156,
+      "step": 3788
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001748462834994926,
+      "loss": 2.1758,
+      "step": 3789
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001744530649613093,
+      "loss": 2.1025,
+      "step": 3790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017406024682476218,
+      "loss": 2.1699,
+      "step": 3791
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001736678292803725,
+      "loss": 2.0986,
+      "step": 3792
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017327581251846903,
+      "loss": 2.0264,
+      "step": 3793
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017288419672918488,
+      "loss": 2.2539,
+      "step": 3794
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001724929821024588,
+      "loss": 2.1797,
+      "step": 3795
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001721021688280362,
+      "loss": 2.0938,
+      "step": 3796
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017171175709546616,
+      "loss": 2.0801,
+      "step": 3797
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017132174709410364,
+      "loss": 2.0439,
+      "step": 3798
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017093213901310934,
+      "loss": 1.999,
+      "step": 3799
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017054293304144842,
+      "loss": 2.2793,
+      "step": 3800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017015412936789088,
+      "loss": 2.1748,
+      "step": 3801
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016976572818101245,
+      "loss": 2.127,
+      "step": 3802
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001693777296691924,
+      "loss": 2.1074,
+      "step": 3803
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016899013402061548,
+      "loss": 2.208,
+      "step": 3804
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016860294142327116,
+      "loss": 2.1777,
+      "step": 3805
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001682161520649531,
+      "loss": 2.1191,
+      "step": 3806
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016782976613325952,
+      "loss": 2.0146,
+      "step": 3807
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016744378381559278,
+      "loss": 2.1406,
+      "step": 3808
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016705820529915939,
+      "loss": 2.0557,
+      "step": 3809
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016667303077097084,
+      "loss": 2.1934,
+      "step": 3810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016628826041784173,
+      "loss": 2.1709,
+      "step": 3811
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001659038944263911,
+      "loss": 2.0771,
+      "step": 3812
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001655199329830417,
+      "loss": 2.1006,
+      "step": 3813
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016513637627402,
+      "loss": 2.123,
+      "step": 3814
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016475322448535613,
+      "loss": 2.2246,
+      "step": 3815
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001643704778028845,
+      "loss": 2.1318,
+      "step": 3816
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016398813641224232,
+      "loss": 2.2158,
+      "step": 3817
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001636062004988703,
+      "loss": 2.127,
+      "step": 3818
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001632246702480128,
+      "loss": 2.2314,
+      "step": 3819
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016284354584471705,
+      "loss": 1.9561,
+      "step": 3820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016246282747383356,
+      "loss": 2.1084,
+      "step": 3821
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016208251532001628,
+      "loss": 2.2363,
+      "step": 3822
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016170260956772177,
+      "loss": 2.3164,
+      "step": 3823
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001613231104012095,
+      "loss": 2.2012,
+      "step": 3824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016094401800454184,
+      "loss": 2.2109,
+      "step": 3825
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016056533256158346,
+      "loss": 1.9971,
+      "step": 3826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016018705425600255,
+      "loss": 2.0371,
+      "step": 3827
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015980918327126949,
+      "loss": 2.0176,
+      "step": 3828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015943171979065584,
+      "loss": 2.208,
+      "step": 3829
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015905466399723756,
+      "loss": 2.0205,
+      "step": 3830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015867801607389153,
+      "loss": 2.0283,
+      "step": 3831
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001583017762032971,
+      "loss": 2.0498,
+      "step": 3832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001579259445679364,
+      "loss": 2.0576,
+      "step": 3833
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015755052135009228,
+      "loss": 1.9111,
+      "step": 3834
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001571755067318501,
+      "loss": 2.1689,
+      "step": 3835
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015680090089509758,
+      "loss": 2.1611,
+      "step": 3836
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001564267040215236,
+      "loss": 2.0859,
+      "step": 3837
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015605291629261842,
+      "loss": 2.2783,
+      "step": 3838
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015567953788967503,
+      "loss": 2.1455,
+      "step": 3839
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015530656899378624,
+      "loss": 2.0801,
+      "step": 3840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001549340097858476,
+      "loss": 2.1875,
+      "step": 3841
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001545618604465554,
+      "loss": 2.1211,
+      "step": 3842
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015419012115640686,
+      "loss": 2.1465,
+      "step": 3843
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015381879209570138,
+      "loss": 2.0889,
+      "step": 3844
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015344787344453802,
+      "loss": 2.0127,
+      "step": 3845
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015307736538281737,
+      "loss": 2.2285,
+      "step": 3846
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015270726809024139,
+      "loss": 1.9482,
+      "step": 3847
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015233758174631228,
+      "loss": 2.0684,
+      "step": 3848
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015196830653033277,
+      "loss": 2.3066,
+      "step": 3849
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015159944262140669,
+      "loss": 2.1553,
+      "step": 3850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000151230990198438,
+      "loss": 2.0547,
+      "step": 3851
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001508629494401309,
+      "loss": 2.0947,
+      "step": 3852
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015049532052499072,
+      "loss": 2.1094,
+      "step": 3853
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015012810363132246,
+      "loss": 2.1924,
+      "step": 3854
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001497612989372311,
+      "loss": 2.2012,
+      "step": 3855
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014939490662062229,
+      "loss": 2.1436,
+      "step": 3856
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014902892685920088,
+      "loss": 2.1699,
+      "step": 3857
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014866335983047262,
+      "loss": 2.0879,
+      "step": 3858
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014829820571174234,
+      "loss": 2.1973,
+      "step": 3859
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001479334646801148,
+      "loss": 2.2148,
+      "step": 3860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001475691369124945,
+      "loss": 2.0898,
+      "step": 3861
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014720522258558543,
+      "loss": 2.0146,
+      "step": 3862
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014684172187589062,
+      "loss": 2.042,
+      "step": 3863
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001464786349597136,
+      "loss": 2.0898,
+      "step": 3864
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014611596201315647,
+      "loss": 2.1514,
+      "step": 3865
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001457537032121199,
+      "loss": 2.0352,
+      "step": 3866
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000145391858732305,
+      "loss": 2.084,
+      "step": 3867
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014503042874921125,
+      "loss": 2.1055,
+      "step": 3868
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014466941343813688,
+      "loss": 2.0947,
+      "step": 3869
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014430881297417965,
+      "loss": 2.0801,
+      "step": 3870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001439486275322357,
+      "loss": 2.0957,
+      "step": 3871
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014358885728699977,
+      "loss": 2.0215,
+      "step": 3872
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001432295024129655,
+      "loss": 2.2178,
+      "step": 3873
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001428705630844247,
+      "loss": 2.1719,
+      "step": 3874
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014251203947546832,
+      "loss": 2.2148,
+      "step": 3875
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014215393175998525,
+      "loss": 2.1475,
+      "step": 3876
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001417962401116619,
+      "loss": 2.0596,
+      "step": 3877
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014143896470398455,
+      "loss": 2.208,
+      "step": 3878
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014108210571023628,
+      "loss": 2.1143,
+      "step": 3879
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014072566330349846,
+      "loss": 2.1543,
+      "step": 3880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014036963765665122,
+      "loss": 2.1807,
+      "step": 3881
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014001402894237125,
+      "loss": 2.1797,
+      "step": 3882
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013965883733313368,
+      "loss": 2.1699,
+      "step": 3883
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013930406300121178,
+      "loss": 2.1357,
+      "step": 3884
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013894970611867574,
+      "loss": 2.0986,
+      "step": 3885
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013859576685739351,
+      "loss": 2.1416,
+      "step": 3886
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013824224538903053,
+      "loss": 2.1357,
+      "step": 3887
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013788914188504943,
+      "loss": 2.2393,
+      "step": 3888
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013753645651671053,
+      "loss": 2.0625,
+      "step": 3889
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013718418945507117,
+      "loss": 2.1543,
+      "step": 3890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013683234087098517,
+      "loss": 2.0176,
+      "step": 3891
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013648091093510485,
+      "loss": 2.0371,
+      "step": 3892
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001361298998178778,
+      "loss": 2.377,
+      "step": 3893
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013577930768954926,
+      "loss": 2.1914,
+      "step": 3894
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001354291347201617,
+      "loss": 2.1885,
+      "step": 3895
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013507938107955365,
+      "loss": 2.002,
+      "step": 3896
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013473004693736036,
+      "loss": 2.1621,
+      "step": 3897
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013438113246301374,
+      "loss": 2.0781,
+      "step": 3898
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013403263782574217,
+      "loss": 2.2373,
+      "step": 3899
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013368456319457002,
+      "loss": 2.0713,
+      "step": 3900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013333690873831882,
+      "loss": 2.1582,
+      "step": 3901
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001329896746256054,
+      "loss": 2.1387,
+      "step": 3902
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013264286102484336,
+      "loss": 2.1885,
+      "step": 3903
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001322964681042418,
+      "loss": 1.9805,
+      "step": 3904
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013195049603180597,
+      "loss": 2.1201,
+      "step": 3905
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001316049449753375,
+      "loss": 2.0137,
+      "step": 3906
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013125981510243322,
+      "loss": 2.0693,
+      "step": 3907
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001309151065804859,
+      "loss": 2.0059,
+      "step": 3908
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013057081957668383,
+      "loss": 2.1768,
+      "step": 3909
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001302269542580109,
+      "loss": 2.0957,
+      "step": 3910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001298835107912465,
+      "loss": 2.2227,
+      "step": 3911
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012954048934296582,
+      "loss": 2.0791,
+      "step": 3912
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012919789007953886,
+      "loss": 2.207,
+      "step": 3913
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001288557131671305,
+      "loss": 2.2314,
+      "step": 3914
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001285139587717018,
+      "loss": 2.1006,
+      "step": 3915
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012817262705900812,
+      "loss": 2.3008,
+      "step": 3916
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012783171819459995,
+      "loss": 2.1816,
+      "step": 3917
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012749123234382333,
+      "loss": 2.1943,
+      "step": 3918
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012715116967181774,
+      "loss": 2.1611,
+      "step": 3919
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012681153034351878,
+      "loss": 2.0664,
+      "step": 3920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012647231452365626,
+      "loss": 2.1504,
+      "step": 3921
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012613352237675414,
+      "loss": 2.0928,
+      "step": 3922
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012579515406713193,
+      "loss": 2.0176,
+      "step": 3923
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001254572097589024,
+      "loss": 2.168,
+      "step": 3924
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012511968961597297,
+      "loss": 2.0215,
+      "step": 3925
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012478259380204615,
+      "loss": 2.1064,
+      "step": 3926
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012444592248061782,
+      "loss": 2.0898,
+      "step": 3927
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000124109675814978,
+      "loss": 2.208,
+      "step": 3928
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001237738539682115,
+      "loss": 2.1953,
+      "step": 3929
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000123438457103196,
+      "loss": 2.1289,
+      "step": 3930
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012310348538260363,
+      "loss": 2.0488,
+      "step": 3931
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001227689389689006,
+      "loss": 2.1045,
+      "step": 3932
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001224348180243464,
+      "loss": 2.0566,
+      "step": 3933
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012210112271099428,
+      "loss": 2.1875,
+      "step": 3934
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012176785319069095,
+      "loss": 2.1338,
+      "step": 3935
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001214350096250767,
+      "loss": 2.2373,
+      "step": 3936
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001211025921755855,
+      "loss": 2.1162,
+      "step": 3937
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001207706010034444,
+      "loss": 2.2227,
+      "step": 3938
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012043903626967345,
+      "loss": 2.2949,
+      "step": 3939
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012010789813508615,
+      "loss": 2.2695,
+      "step": 3940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011977718676028915,
+      "loss": 2.1719,
+      "step": 3941
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001194469023056819,
+      "loss": 2.0527,
+      "step": 3942
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011911704493145702,
+      "loss": 2.1953,
+      "step": 3943
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011878761479760004,
+      "loss": 2.1777,
+      "step": 3944
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011845861206388874,
+      "loss": 2.1699,
+      "step": 3945
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011813003688989432,
+      "loss": 2.1602,
+      "step": 3946
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011780188943498005,
+      "loss": 2.0156,
+      "step": 3947
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011747416985830184,
+      "loss": 2.1484,
+      "step": 3948
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011714687831880865,
+      "loss": 2.0801,
+      "step": 3949
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011682001497524108,
+      "loss": 2.0527,
+      "step": 3950
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011649357998613241,
+      "loss": 2.1201,
+      "step": 3951
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000116167573509808,
+      "loss": 2.0811,
+      "step": 3952
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011584199570438547,
+      "loss": 2.123,
+      "step": 3953
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011551684672777485,
+      "loss": 2.1494,
+      "step": 3954
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011519212673767787,
+      "loss": 2.167,
+      "step": 3955
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011486783589158755,
+      "loss": 2.0596,
+      "step": 3956
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001145439743467902,
+      "loss": 2.0283,
+      "step": 3957
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011422054226036271,
+      "loss": 2.126,
+      "step": 3958
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011389753978917394,
+      "loss": 2.0879,
+      "step": 3959
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011357496708988501,
+      "loss": 2.1309,
+      "step": 3960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011325282431894812,
+      "loss": 2.1533,
+      "step": 3961
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011293111163260639,
+      "loss": 2.1641,
+      "step": 3962
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011260982918689533,
+      "loss": 1.9941,
+      "step": 3963
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001122889771376413,
+      "loss": 2.1387,
+      "step": 3964
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011196855564046171,
+      "loss": 2.0967,
+      "step": 3965
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011164856485076614,
+      "loss": 2.2061,
+      "step": 3966
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011132900492375341,
+      "loss": 2.1377,
+      "step": 3967
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011100987601441547,
+      "loss": 2.2305,
+      "step": 3968
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011069117827753373,
+      "loss": 2.2334,
+      "step": 3969
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011037291186768095,
+      "loss": 2.1113,
+      "step": 3970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011005507693922134,
+      "loss": 2.0537,
+      "step": 3971
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010973767364630849,
+      "loss": 2.0713,
+      "step": 3972
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010942070214288746,
+      "loss": 2.0361,
+      "step": 3973
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010910416258269407,
+      "loss": 2.1533,
+      "step": 3974
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010878805511925438,
+      "loss": 2.1631,
+      "step": 3975
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010847237990588476,
+      "loss": 1.9551,
+      "step": 3976
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010815713709569196,
+      "loss": 2.0918,
+      "step": 3977
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010784232684157324,
+      "loss": 1.9863,
+      "step": 3978
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010752794929621557,
+      "loss": 2.2256,
+      "step": 3979
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010721400461209684,
+      "loss": 2.1465,
+      "step": 3980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010690049294148441,
+      "loss": 2.0869,
+      "step": 3981
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010658741443643561,
+      "loss": 1.8467,
+      "step": 3982
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010627476924879798,
+      "loss": 2.0518,
+      "step": 3983
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010596255753020856,
+      "loss": 2.0654,
+      "step": 3984
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010565077943209456,
+      "loss": 2.0498,
+      "step": 3985
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010533943510567256,
+      "loss": 2.0459,
+      "step": 3986
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010502852470194878,
+      "loss": 1.8906,
+      "step": 3987
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010471804837171917,
+      "loss": 1.9941,
+      "step": 3988
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010440800626556902,
+      "loss": 2.1729,
+      "step": 3989
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010409839853387271,
+      "loss": 2.1416,
+      "step": 3990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010378922532679469,
+      "loss": 2.1836,
+      "step": 3991
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000103480486794288,
+      "loss": 2.2324,
+      "step": 3992
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010317218308609522,
+      "loss": 2.1377,
+      "step": 3993
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001028643143517477,
+      "loss": 2.291,
+      "step": 3994
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010255688074056624,
+      "loss": 2.167,
+      "step": 3995
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010224988240166,
+      "loss": 1.9561,
+      "step": 3996
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010194331948392777,
+      "loss": 2.1875,
+      "step": 3997
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010163719213605682,
+      "loss": 2.0615,
+      "step": 3998
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010133150050652296,
+      "loss": 2.3359,
+      "step": 3999
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001010262447435908,
+      "loss": 2.0859,
+      "step": 4000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010072142499531345,
+      "loss": 2.0342,
+      "step": 4001
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010041704140953312,
+      "loss": 2.2441,
+      "step": 4002
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010011309413388003,
+      "loss": 1.9004,
+      "step": 4003
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.980958331577217e-05,
+      "loss": 2.1006,
+      "step": 4004
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.95065091024171e-05,
+      "loss": 2.082,
+      "step": 4005
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.92038716408098e-05,
+      "loss": 2.1553,
+      "step": 4006
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.89016710777334e-05,
+      "loss": 2.0771,
+      "step": 4007
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.859990755975979e-05,
+      "loss": 1.9355,
+      "step": 4008
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.829858123324797e-05,
+      "loss": 2.1016,
+      "step": 4009
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.79976922443454e-05,
+      "loss": 2.4072,
+      "step": 4010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.769724073898778e-05,
+      "loss": 2.0977,
+      "step": 4011
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.739722686289787e-05,
+      "loss": 2.0752,
+      "step": 4012
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.709765076158661e-05,
+      "loss": 2.0762,
+      "step": 4013
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.679851258035277e-05,
+      "loss": 2.1074,
+      "step": 4014
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.649981246428197e-05,
+      "loss": 2.0879,
+      "step": 4015
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.620155055824841e-05,
+      "loss": 2.0947,
+      "step": 4016
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.590372700691297e-05,
+      "loss": 2.0381,
+      "step": 4017
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.560634195472406e-05,
+      "loss": 2.2217,
+      "step": 4018
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.530939554591811e-05,
+      "loss": 2.2617,
+      "step": 4019
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.501288792451746e-05,
+      "loss": 2.1025,
+      "step": 4020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.471681923433261e-05,
+      "loss": 2.2021,
+      "step": 4021
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.442118961896129e-05,
+      "loss": 2.1328,
+      "step": 4022
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.412599922178767e-05,
+      "loss": 1.9834,
+      "step": 4023
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.383124818598322e-05,
+      "loss": 1.915,
+      "step": 4024
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.353693665450624e-05,
+      "loss": 2.0029,
+      "step": 4025
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.32430647701018e-05,
+      "loss": 2.0488,
+      "step": 4026
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.294963267530176e-05,
+      "loss": 1.9854,
+      "step": 4027
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.265664051242506e-05,
+      "loss": 2.166,
+      "step": 4028
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.236408842357669e-05,
+      "loss": 2.1299,
+      "step": 4029
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.207197655064847e-05,
+      "loss": 2.084,
+      "step": 4030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.178030503531865e-05,
+      "loss": 2.2227,
+      "step": 4031
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.148907401905182e-05,
+      "loss": 2.0898,
+      "step": 4032
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.119828364309934e-05,
+      "loss": 2.2109,
+      "step": 4033
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.090793404849852e-05,
+      "loss": 2.1982,
+      "step": 4034
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.061802537607267e-05,
+      "loss": 2.1426,
+      "step": 4035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.032855776643167e-05,
+      "loss": 2.1895,
+      "step": 4036
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.003953135997122e-05,
+      "loss": 2.1885,
+      "step": 4037
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.975094629687287e-05,
+      "loss": 1.9912,
+      "step": 4038
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.946280271710494e-05,
+      "loss": 2.041,
+      "step": 4039
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.917510076042057e-05,
+      "loss": 2.2129,
+      "step": 4040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.888784056635935e-05,
+      "loss": 2.2949,
+      "step": 4041
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.860102227424637e-05,
+      "loss": 2.1729,
+      "step": 4042
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.83146460231925e-05,
+      "loss": 2.1211,
+      "step": 4043
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.802871195209394e-05,
+      "loss": 2.0352,
+      "step": 4044
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.774322019963322e-05,
+      "loss": 2.2266,
+      "step": 4045
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.745817090427699e-05,
+      "loss": 2.1191,
+      "step": 4046
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.717356420427869e-05,
+      "loss": 2.1963,
+      "step": 4047
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.688940023767634e-05,
+      "loss": 2.1006,
+      "step": 4048
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.660567914229312e-05,
+      "loss": 2.1484,
+      "step": 4049
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.632240105573808e-05,
+      "loss": 2.0771,
+      "step": 4050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.603956611540498e-05,
+      "loss": 1.9883,
+      "step": 4051
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.575717445847208e-05,
+      "loss": 2.1104,
+      "step": 4052
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.547522622190384e-05,
+      "loss": 2.0205,
+      "step": 4053
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.519372154244886e-05,
+      "loss": 2.1338,
+      "step": 4054
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.491266055664049e-05,
+      "loss": 2.0264,
+      "step": 4055
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.463204340079789e-05,
+      "loss": 1.998,
+      "step": 4056
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.435187021102353e-05,
+      "loss": 2.1025,
+      "step": 4057
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.407214112320538e-05,
+      "loss": 2.0684,
+      "step": 4058
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.379285627301625e-05,
+      "loss": 2.0547,
+      "step": 4059
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.3514015795913e-05,
+      "loss": 2.1436,
+      "step": 4060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.32356198271369e-05,
+      "loss": 2.1982,
+      "step": 4061
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.295766850171404e-05,
+      "loss": 2.1953,
+      "step": 4062
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.268016195445449e-05,
+      "loss": 2.043,
+      "step": 4063
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.240310031995291e-05,
+      "loss": 2.1172,
+      "step": 4064
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.212648373258791e-05,
+      "loss": 2.1484,
+      "step": 4065
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.185031232652251e-05,
+      "loss": 2.1846,
+      "step": 4066
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.157458623570335e-05,
+      "loss": 2.0254,
+      "step": 4067
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.129930559386167e-05,
+      "loss": 2.1406,
+      "step": 4068
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.1024470534512e-05,
+      "loss": 2.1562,
+      "step": 4069
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.075008119095351e-05,
+      "loss": 2.0977,
+      "step": 4070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.047613769626871e-05,
+      "loss": 2.1328,
+      "step": 4071
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.02026401833239e-05,
+      "loss": 2.04,
+      "step": 4072
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.99295887847693e-05,
+      "loss": 2.1895,
+      "step": 4073
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.96569836330383e-05,
+      "loss": 2.0977,
+      "step": 4074
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.938482486034826e-05,
+      "loss": 2.2412,
+      "step": 4075
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.911311259870014e-05,
+      "loss": 2.1143,
+      "step": 4076
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.884184697987806e-05,
+      "loss": 2.1104,
+      "step": 4077
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.857102813544936e-05,
+      "loss": 2.1865,
+      "step": 4078
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.830065619676518e-05,
+      "loss": 2.1348,
+      "step": 4079
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.80307312949593e-05,
+      "loss": 2.2148,
+      "step": 4080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.776125356094943e-05,
+      "loss": 2.3047,
+      "step": 4081
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.749222312543602e-05,
+      "loss": 2.2422,
+      "step": 4082
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.722364011890182e-05,
+      "loss": 2.1875,
+      "step": 4083
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.695550467161405e-05,
+      "loss": 1.9355,
+      "step": 4084
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.668781691362181e-05,
+      "loss": 2.165,
+      "step": 4085
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.642057697475713e-05,
+      "loss": 2.2158,
+      "step": 4086
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.615378498463543e-05,
+      "loss": 2.2188,
+      "step": 4087
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.588744107265443e-05,
+      "loss": 2.1104,
+      "step": 4088
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.562154536799448e-05,
+      "loss": 2.0146,
+      "step": 4089
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.535609799961873e-05,
+      "loss": 2.1553,
+      "step": 4090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.509109909627287e-05,
+      "loss": 2.0176,
+      "step": 4091
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.482654878648464e-05,
+      "loss": 2.0342,
+      "step": 4092
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.456244719856531e-05,
+      "loss": 2.1738,
+      "step": 4093
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.429879446060695e-05,
+      "loss": 2.125,
+      "step": 4094
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.403559070048538e-05,
+      "loss": 2.1992,
+      "step": 4095
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.377283604585783e-05,
+      "loss": 2.0684,
+      "step": 4096
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.35105306241638e-05,
+      "loss": 2.1943,
+      "step": 4097
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.32486745626254e-05,
+      "loss": 2.1562,
+      "step": 4098
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.298726798824618e-05,
+      "loss": 2.1152,
+      "step": 4099
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.272631102781168e-05,
+      "loss": 2.0791,
+      "step": 4100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.246580380789014e-05,
+      "loss": 2.0957,
+      "step": 4101
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.22057464548308e-05,
+      "loss": 2.1514,
+      "step": 4102
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.19461390947651e-05,
+      "loss": 2.0781,
+      "step": 4103
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.168698185360656e-05,
+      "loss": 2.2197,
+      "step": 4104
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.142827485704951e-05,
+      "loss": 2.083,
+      "step": 4105
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.117001823057045e-05,
+      "loss": 2.1582,
+      "step": 4106
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.091221209942766e-05,
+      "loss": 2.2002,
+      "step": 4107
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.065485658866067e-05,
+      "loss": 2.0312,
+      "step": 4108
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.039795182309027e-05,
+      "loss": 2.1768,
+      "step": 4109
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.014149792731883e-05,
+      "loss": 2.1631,
+      "step": 4110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.988549502572993e-05,
+      "loss": 2.0742,
+      "step": 4111
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.962994324248883e-05,
+      "loss": 2.2168,
+      "step": 4112
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.937484270154138e-05,
+      "loss": 2.1895,
+      "step": 4113
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.912019352661502e-05,
+      "loss": 2.0244,
+      "step": 4114
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.8865995841218e-05,
+      "loss": 2.1445,
+      "step": 4115
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.861224976863978e-05,
+      "loss": 2.0654,
+      "step": 4116
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.835895543195047e-05,
+      "loss": 2.1436,
+      "step": 4117
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.81061129540017e-05,
+      "loss": 2.1279,
+      "step": 4118
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.785372245742527e-05,
+      "loss": 2.0225,
+      "step": 4119
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.760178406463424e-05,
+      "loss": 2.208,
+      "step": 4120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.7350297897822e-05,
+      "loss": 2.2617,
+      "step": 4121
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.709926407896294e-05,
+      "loss": 2.293,
+      "step": 4122
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.68486827298117e-05,
+      "loss": 2.1172,
+      "step": 4123
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.659855397190406e-05,
+      "loss": 2.2832,
+      "step": 4124
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.634887792655575e-05,
+      "loss": 2.334,
+      "step": 4125
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.6099654714863e-05,
+      "loss": 2.1289,
+      "step": 4126
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.585088445770249e-05,
+      "loss": 2.04,
+      "step": 4127
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.560256727573122e-05,
+      "loss": 2.1621,
+      "step": 4128
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.535470328938665e-05,
+      "loss": 2.2676,
+      "step": 4129
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.510729261888626e-05,
+      "loss": 2.0283,
+      "step": 4130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.48603353842272e-05,
+      "loss": 2.1084,
+      "step": 4131
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.461383170518743e-05,
+      "loss": 2.1396,
+      "step": 4132
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.436778170132474e-05,
+      "loss": 2.2168,
+      "step": 4133
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.412218549197635e-05,
+      "loss": 2.1279,
+      "step": 4134
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.387704319626042e-05,
+      "loss": 2.0781,
+      "step": 4135
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.363235493307374e-05,
+      "loss": 2.1953,
+      "step": 4136
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.33881208210939e-05,
+      "loss": 2.0273,
+      "step": 4137
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.314434097877763e-05,
+      "loss": 2.0791,
+      "step": 4138
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.290101552436155e-05,
+      "loss": 2.2607,
+      "step": 4139
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.265814457586172e-05,
+      "loss": 2.1641,
+      "step": 4140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.241572825107433e-05,
+      "loss": 2.2158,
+      "step": 4141
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.217376666757402e-05,
+      "loss": 2.168,
+      "step": 4142
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.193225994271601e-05,
+      "loss": 2.1045,
+      "step": 4143
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.169120819363405e-05,
+      "loss": 2.0771,
+      "step": 4144
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.145061153724163e-05,
+      "loss": 1.9971,
+      "step": 4145
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.121047009023173e-05,
+      "loss": 1.998,
+      "step": 4146
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.097078396907596e-05,
+      "loss": 2.1963,
+      "step": 4147
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.073155329002533e-05,
+      "loss": 2.167,
+      "step": 4148
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.0492778169110254e-05,
+      "loss": 2.041,
+      "step": 4149
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.025445872213986e-05,
+      "loss": 2.1953,
+      "step": 4150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.0016595064702364e-05,
+      "loss": 2.2275,
+      "step": 4151
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.977918731216481e-05,
+      "loss": 2.248,
+      "step": 4152
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.954223557967342e-05,
+      "loss": 2.126,
+      "step": 4153
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.9305739982152826e-05,
+      "loss": 2.1914,
+      "step": 4154
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.906970063430683e-05,
+      "loss": 1.9863,
+      "step": 4155
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.883411765061775e-05,
+      "loss": 2.1104,
+      "step": 4156
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.859899114534661e-05,
+      "loss": 2.0596,
+      "step": 4157
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.836432123253288e-05,
+      "loss": 2.2158,
+      "step": 4158
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.813010802599461e-05,
+      "loss": 2.002,
+      "step": 4159
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.7896351639328715e-05,
+      "loss": 2.0791,
+      "step": 4160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.766305218591006e-05,
+      "loss": 2.1387,
+      "step": 4161
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.743020977889224e-05,
+      "loss": 2.2383,
+      "step": 4162
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.719782453120692e-05,
+      "loss": 2.0908,
+      "step": 4163
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.696589655556428e-05,
+      "loss": 2.084,
+      "step": 4164
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.673442596445222e-05,
+      "loss": 2.1172,
+      "step": 4165
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.6503412870137605e-05,
+      "loss": 2.1074,
+      "step": 4166
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.62728573846647e-05,
+      "loss": 2.1875,
+      "step": 4167
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.604275961985628e-05,
+      "loss": 2.1895,
+      "step": 4168
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.581311968731284e-05,
+      "loss": 2.1943,
+      "step": 4169
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5583937698412854e-05,
+      "loss": 2.1367,
+      "step": 4170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.535521376431263e-05,
+      "loss": 2.1924,
+      "step": 4171
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5126947995946866e-05,
+      "loss": 2.208,
+      "step": 4172
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.489914050402711e-05,
+      "loss": 2.0264,
+      "step": 4173
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.467179139904344e-05,
+      "loss": 2.0625,
+      "step": 4174
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4444900791263184e-05,
+      "loss": 2.0586,
+      "step": 4175
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.421846879073133e-05,
+      "loss": 2.1338,
+      "step": 4176
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.399249550727081e-05,
+      "loss": 2.2002,
+      "step": 4177
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.376698105048161e-05,
+      "loss": 2.1748,
+      "step": 4178
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.354192552974124e-05,
+      "loss": 2.0713,
+      "step": 4179
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3317329054204835e-05,
+      "loss": 1.9639,
+      "step": 4180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3093191732804805e-05,
+      "loss": 2.1279,
+      "step": 4181
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2869513674250635e-05,
+      "loss": 2.1973,
+      "step": 4182
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2646294987029666e-05,
+      "loss": 2.1855,
+      "step": 4183
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.242353577940562e-05,
+      "loss": 2.2031,
+      "step": 4184
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2201236159420074e-05,
+      "loss": 2.1592,
+      "step": 4185
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.197939623489123e-05,
+      "loss": 2.0645,
+      "step": 4186
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.175801611341457e-05,
+      "loss": 2.1475,
+      "step": 4187
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1537095902362553e-05,
+      "loss": 2.2754,
+      "step": 4188
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.131663570888456e-05,
+      "loss": 2.0322,
+      "step": 4189
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1096635639906406e-05,
+      "loss": 2.1465,
+      "step": 4190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.087709580213173e-05,
+      "loss": 2.2031,
+      "step": 4191
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0658016302040144e-05,
+      "loss": 2.1816,
+      "step": 4192
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.043939724588808e-05,
+      "loss": 2.0645,
+      "step": 4193
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.022123873970919e-05,
+      "loss": 2.0605,
+      "step": 4194
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0003540889312915e-05,
+      "loss": 2.2529,
+      "step": 4195
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.978630380028581e-05,
+      "loss": 2.3008,
+      "step": 4196
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9569527577991045e-05,
+      "loss": 1.9551,
+      "step": 4197
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9353212327567956e-05,
+      "loss": 2.1553,
+      "step": 4198
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.913735815393239e-05,
+      "loss": 2.1387,
+      "step": 4199
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8921965161776606e-05,
+      "loss": 2.2354,
+      "step": 4200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.870703345556926e-05,
+      "loss": 2.0342,
+      "step": 4201
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.849256313955497e-05,
+      "loss": 2.1348,
+      "step": 4202
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8278554317755076e-05,
+      "loss": 2.0859,
+      "step": 4203
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8065007093966664e-05,
+      "loss": 2.0098,
+      "step": 4204
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.78519215717631e-05,
+      "loss": 2.0635,
+      "step": 4205
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.763929785449383e-05,
+      "loss": 2.1172,
+      "step": 4206
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.742713604528404e-05,
+      "loss": 2.0264,
+      "step": 4207
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7215436247035506e-05,
+      "loss": 2.1143,
+      "step": 4208
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.700419856242555e-05,
+      "loss": 2.0225,
+      "step": 4209
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6793423093906885e-05,
+      "loss": 2.3135,
+      "step": 4210
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6583109943708954e-05,
+      "loss": 2.2646,
+      "step": 4211
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6373259213836395e-05,
+      "loss": 2.0889,
+      "step": 4212
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6163871006069554e-05,
+      "loss": 2.1631,
+      "step": 4213
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.595494542196488e-05,
+      "loss": 2.0098,
+      "step": 4214
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5746482562854075e-05,
+      "loss": 2.1787,
+      "step": 4215
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5538482529844386e-05,
+      "loss": 2.1367,
+      "step": 4216
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.533094542381877e-05,
+      "loss": 2.0918,
+      "step": 4217
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.51238713454355e-05,
+      "loss": 2.0127,
+      "step": 4218
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4917260395128444e-05,
+      "loss": 2.2344,
+      "step": 4219
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.47111126731069e-05,
+      "loss": 2.2734,
+      "step": 4220
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.450542827935489e-05,
+      "loss": 2.0273,
+      "step": 4221
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.430020731363271e-05,
+      "loss": 2.2715,
+      "step": 4222
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.409544987547509e-05,
+      "loss": 2.0703,
+      "step": 4223
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.389115606419203e-05,
+      "loss": 2.3086,
+      "step": 4224
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3687325978869376e-05,
+      "loss": 2.1396,
+      "step": 4225
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.348395971836716e-05,
+      "loss": 2.2051,
+      "step": 4226
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3281057381320596e-05,
+      "loss": 2.0869,
+      "step": 4227
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.307861906614063e-05,
+      "loss": 2.2246,
+      "step": 4228
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.287664487101239e-05,
+      "loss": 2.0957,
+      "step": 4229
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.267513489389596e-05,
+      "loss": 2.0791,
+      "step": 4230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2474089232526824e-05,
+      "loss": 2.1699,
+      "step": 4231
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2273507984414536e-05,
+      "loss": 2.0811,
+      "step": 4232
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.207339124684406e-05,
+      "loss": 2.1514,
+      "step": 4233
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.187373911687464e-05,
+      "loss": 2.1572,
+      "step": 4234
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.167455169134027e-05,
+      "loss": 2.0469,
+      "step": 4235
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.147582906684977e-05,
+      "loss": 2.0752,
+      "step": 4236
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.127757133978605e-05,
+      "loss": 2.1279,
+      "step": 4237
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.107977860630696e-05,
+      "loss": 2.0342,
+      "step": 4238
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0882450962344886e-05,
+      "loss": 2.1533,
+      "step": 4239
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0685588503606376e-05,
+      "loss": 2.1475,
+      "step": 4240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.048919132557227e-05,
+      "loss": 2.1328,
+      "step": 4241
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.029325952349816e-05,
+      "loss": 1.9814,
+      "step": 4242
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0097793192413355e-05,
+      "loss": 2.2236,
+      "step": 4243
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.99027924271218e-05,
+      "loss": 2.0479,
+      "step": 4244
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.970825732220185e-05,
+      "loss": 1.9717,
+      "step": 4245
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9514187972005455e-05,
+      "loss": 2.0342,
+      "step": 4246
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.932058447065889e-05,
+      "loss": 2.0527,
+      "step": 4247
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.91274469120626e-05,
+      "loss": 2.0811,
+      "step": 4248
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.89347753898911e-05,
+      "loss": 2.1543,
+      "step": 4249
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8742569997592294e-05,
+      "loss": 2.1963,
+      "step": 4250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.855083082838895e-05,
+      "loss": 2.0361,
+      "step": 4251
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8359557975277015e-05,
+      "loss": 2.1465,
+      "step": 4252
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8168751531026394e-05,
+      "loss": 2.1172,
+      "step": 4253
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.797841158818094e-05,
+      "loss": 2.2031,
+      "step": 4254
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.778853823905792e-05,
+      "loss": 1.9199,
+      "step": 4255
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.759913157574879e-05,
+      "loss": 2.0938,
+      "step": 4256
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.74101916901185e-05,
+      "loss": 2.0654,
+      "step": 4257
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.722171867380497e-05,
+      "loss": 2.1709,
+      "step": 4258
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.703371261822075e-05,
+      "loss": 2.1162,
+      "step": 4259
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6846173614551113e-05,
+      "loss": 2.0928,
+      "step": 4260
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6659101753754974e-05,
+      "loss": 2.1582,
+      "step": 4261
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6472497126565085e-05,
+      "loss": 2.1729,
+      "step": 4262
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.628635982348683e-05,
+      "loss": 2.0693,
+      "step": 4263
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.610068993479976e-05,
+      "loss": 2.1641,
+      "step": 4264
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.591548755055618e-05,
+      "loss": 1.9814,
+      "step": 4265
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5730752760581684e-05,
+      "loss": 2.1699,
+      "step": 4266
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.554648565447527e-05,
+      "loss": 2.1104,
+      "step": 4267
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5362686321609216e-05,
+      "loss": 2.1162,
+      "step": 4268
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5179354851128445e-05,
+      "loss": 2.3027,
+      "step": 4269
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.499649133195138e-05,
+      "loss": 2.1934,
+      "step": 4270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.481409585276929e-05,
+      "loss": 2.1221,
+      "step": 4271
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4632168502046426e-05,
+      "loss": 2.1348,
+      "step": 4272
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4450709368020416e-05,
+      "loss": 1.9092,
+      "step": 4273
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.426971853870109e-05,
+      "loss": 2.3047,
+      "step": 4274
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.408919610187145e-05,
+      "loss": 2.0254,
+      "step": 4275
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3909142145087554e-05,
+      "loss": 2.085,
+      "step": 4276
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.372955675567812e-05,
+      "loss": 2.2598,
+      "step": 4277
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3550440020744365e-05,
+      "loss": 2.0459,
+      "step": 4278
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.337179202716045e-05,
+      "loss": 2.1738,
+      "step": 4279
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.319361286157285e-05,
+      "loss": 2.0918,
+      "step": 4280
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.301590261040144e-05,
+      "loss": 2.2129,
+      "step": 4281
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.283866135983771e-05,
+      "loss": 2.2051,
+      "step": 4282
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.266188919584634e-05,
+      "loss": 2.0537,
+      "step": 4283
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2485586204164086e-05,
+      "loss": 2.2373,
+      "step": 4284
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2309752470300545e-05,
+      "loss": 2.165,
+      "step": 4285
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.213438807953717e-05,
+      "loss": 2.2969,
+      "step": 4286
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.195949311692847e-05,
+      "loss": 2.1221,
+      "step": 4287
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.178506766730071e-05,
+      "loss": 2.125,
+      "step": 4288
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.161111181525267e-05,
+      "loss": 2.1055,
+      "step": 4289
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.14376256451554e-05,
+      "loss": 2.1494,
+      "step": 4290
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.126460924115193e-05,
+      "loss": 2.2744,
+      "step": 4291
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.109206268715758e-05,
+      "loss": 2.3203,
+      "step": 4292
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.091998606686008e-05,
+      "loss": 2.0967,
+      "step": 4293
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.074837946371878e-05,
+      "loss": 2.2354,
+      "step": 4294
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0577242960965314e-05,
+      "loss": 2.25,
+      "step": 4295
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0406576641603177e-05,
+      "loss": 2.207,
+      "step": 4296
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.023638058840783e-05,
+      "loss": 2.1133,
+      "step": 4297
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.006665488392679e-05,
+      "loss": 2.1074,
+      "step": 4298
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9897399610479437e-05,
+      "loss": 2.1162,
+      "step": 4299
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9728614850156656e-05,
+      "loss": 2.0732,
+      "step": 4300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9560300684821627e-05,
+      "loss": 2.252,
+      "step": 4301
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9392457196108947e-05,
+      "loss": 2.2266,
+      "step": 4302
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9225084465425043e-05,
+      "loss": 2.1475,
+      "step": 4303
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.905818257394799e-05,
+      "loss": 2.0674,
+      "step": 4304
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.889175160262758e-05,
+      "loss": 2.0762,
+      "step": 4305
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8725791632184806e-05,
+      "loss": 2.2266,
+      "step": 4306
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8560302743112942e-05,
+      "loss": 2.1904,
+      "step": 4307
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.839528501567623e-05,
+      "loss": 2.1611,
+      "step": 4308
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8230738529910427e-05,
+      "loss": 2.2061,
+      "step": 4309
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8066663365623136e-05,
+      "loss": 2.0547,
+      "step": 4310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7903059602392588e-05,
+      "loss": 2.1797,
+      "step": 4311
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7739927319569314e-05,
+      "loss": 2.0518,
+      "step": 4312
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7577266596274574e-05,
+      "loss": 2.0176,
+      "step": 4313
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7415077511401043e-05,
+      "loss": 2.0273,
+      "step": 4314
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.725336014361268e-05,
+      "loss": 2.1455,
+      "step": 4315
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.709211457134464e-05,
+      "loss": 1.9629,
+      "step": 4316
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6931340872803024e-05,
+      "loss": 2.0068,
+      "step": 4317
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.677103912596568e-05,
+      "loss": 2.25,
+      "step": 4318
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6611209408580862e-05,
+      "loss": 2.1416,
+      "step": 4319
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6451851798168337e-05,
+      "loss": 2.0879,
+      "step": 4320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6292966372018833e-05,
+      "loss": 2.1846,
+      "step": 4321
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6134553207193712e-05,
+      "loss": 2.1123,
+      "step": 4322
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.597661238052551e-05,
+      "loss": 2.0635,
+      "step": 4323
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5819143968618065e-05,
+      "loss": 2.0527,
+      "step": 4324
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.56621480478455e-05,
+      "loss": 1.9307,
+      "step": 4325
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5505624694353023e-05,
+      "loss": 2.0635,
+      "step": 4326
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5349573984056574e-05,
+      "loss": 2.1035,
+      "step": 4327
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5193995992642938e-05,
+      "loss": 2.0107,
+      "step": 4328
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5038890795569537e-05,
+      "loss": 2.2109,
+      "step": 4329
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4884258468064745e-05,
+      "loss": 2.2461,
+      "step": 4330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.473009908512702e-05,
+      "loss": 2.3467,
+      "step": 4331
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.457641272152611e-05,
+      "loss": 1.9746,
+      "step": 4332
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4423199451801825e-05,
+      "loss": 2.1641,
+      "step": 4333
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.427045935026462e-05,
+      "loss": 2.1455,
+      "step": 4334
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4118192490995892e-05,
+      "loss": 2.2686,
+      "step": 4335
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3966398947846913e-05,
+      "loss": 2.1006,
+      "step": 4336
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.381507879443967e-05,
+      "loss": 2.0742,
+      "step": 4337
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3664232104166462e-05,
+      "loss": 2.0635,
+      "step": 4338
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3513858950190204e-05,
+      "loss": 2.2031,
+      "step": 4339
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.336395940544378e-05,
+      "loss": 2.0859,
+      "step": 4340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.321453354263059e-05,
+      "loss": 2.1738,
+      "step": 4341
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3065581434224325e-05,
+      "loss": 1.9902,
+      "step": 4342
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2917103152468644e-05,
+      "loss": 2.2568,
+      "step": 4343
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2769098769377607e-05,
+      "loss": 2.0381,
+      "step": 4344
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2621568356735457e-05,
+      "loss": 2.3291,
+      "step": 4345
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.24745119860964e-05,
+      "loss": 2.1689,
+      "step": 4346
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2327929728784945e-05,
+      "loss": 2.1631,
+      "step": 4347
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.218182165589522e-05,
+      "loss": 2.0146,
+      "step": 4348
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2036187838291976e-05,
+      "loss": 2.0342,
+      "step": 4349
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1891028346609386e-05,
+      "loss": 2.1035,
+      "step": 4350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1746343251251798e-05,
+      "loss": 2.1719,
+      "step": 4351
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1602132622393743e-05,
+      "loss": 2.1895,
+      "step": 4352
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1458396529979164e-05,
+      "loss": 2.2852,
+      "step": 4353
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1315135043721955e-05,
+      "loss": 2.208,
+      "step": 4354
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1172348233105986e-05,
+      "loss": 2.0303,
+      "step": 4355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1030036167384968e-05,
+      "loss": 2.0312,
+      "step": 4356
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0888198915582025e-05,
+      "loss": 2.3027,
+      "step": 4357
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0746836546490456e-05,
+      "loss": 2.2012,
+      "step": 4358
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.060594912867253e-05,
+      "loss": 2.1719,
+      "step": 4359
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.046553673046092e-05,
+      "loss": 2.2119,
+      "step": 4360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0325599419957486e-05,
+      "loss": 2.0537,
+      "step": 4361
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.018613726503371e-05,
+      "loss": 2.1289,
+      "step": 4362
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.004715033333071e-05,
+      "loss": 2.1914,
+      "step": 4363
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9908638692259006e-05,
+      "loss": 2.1416,
+      "step": 4364
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.977060240899864e-05,
+      "loss": 2.1445,
+      "step": 4365
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9633041550499166e-05,
+      "loss": 2.0967,
+      "step": 4366
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9495956183479548e-05,
+      "loss": 2.0332,
+      "step": 4367
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9359346374428043e-05,
+      "loss": 1.959,
+      "step": 4368
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9223212189602103e-05,
+      "loss": 2.1367,
+      "step": 4369
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9087553695029014e-05,
+      "loss": 2.2207,
+      "step": 4370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8952370956504705e-05,
+      "loss": 2.0635,
+      "step": 4371
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.881766403959484e-05,
+      "loss": 1.9912,
+      "step": 4372
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8683433009634043e-05,
+      "loss": 2.1133,
+      "step": 4373
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.854967793172635e-05,
+      "loss": 2.2793,
+      "step": 4374
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8416398870744754e-05,
+      "loss": 2.2051,
+      "step": 4375
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8283595891331217e-05,
+      "loss": 2.1816,
+      "step": 4376
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8151269057897435e-05,
+      "loss": 2.0654,
+      "step": 4377
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.80194184346234e-05,
+      "loss": 2.1641,
+      "step": 4378
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7888044085458634e-05,
+      "loss": 2.1074,
+      "step": 4379
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.775714607412138e-05,
+      "loss": 2.1279,
+      "step": 4380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7626724464099185e-05,
+      "loss": 1.9922,
+      "step": 4381
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7496779318648015e-05,
+      "loss": 2.0703,
+      "step": 4382
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7367310700793447e-05,
+      "loss": 2.0986,
+      "step": 4383
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7238318673329366e-05,
+      "loss": 2.1553,
+      "step": 4384
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7109803298818504e-05,
+      "loss": 2.2988,
+      "step": 4385
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6981764639592778e-05,
+      "loss": 2.292,
+      "step": 4386
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6854202757752735e-05,
+      "loss": 2.0811,
+      "step": 4387
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.672711771516744e-05,
+      "loss": 2.0664,
+      "step": 4388
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6600509573475254e-05,
+      "loss": 2.0342,
+      "step": 4389
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.647437839408239e-05,
+      "loss": 2.1201,
+      "step": 4390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.634872423816458e-05,
+      "loss": 2.0732,
+      "step": 4391
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6223547166665743e-05,
+      "loss": 1.9668,
+      "step": 4392
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6098847240298308e-05,
+      "loss": 2.1992,
+      "step": 4393
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5974624519543677e-05,
+      "loss": 1.999,
+      "step": 4394
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5850879064651545e-05,
+      "loss": 2.0977,
+      "step": 4395
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5727610935640013e-05,
+      "loss": 2.0137,
+      "step": 4396
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.560482019229592e-05,
+      "loss": 1.96,
+      "step": 4397
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.548250689417452e-05,
+      "loss": 2.2432,
+      "step": 4398
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5360671100599356e-05,
+      "loss": 2.1328,
+      "step": 4399
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5239312870662603e-05,
+      "loss": 2.2832,
+      "step": 4400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5118432263224513e-05,
+      "loss": 2.1934,
+      "step": 4401
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4998029336913743e-05,
+      "loss": 2.084,
+      "step": 4402
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4878104150127691e-05,
+      "loss": 2.0791,
+      "step": 4403
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4758656761031608e-05,
+      "loss": 2.0859,
+      "step": 4404
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.463968722755904e-05,
+      "loss": 2.1807,
+      "step": 4405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4521195607411942e-05,
+      "loss": 2.1191,
+      "step": 4406
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.440318195806023e-05,
+      "loss": 2.0361,
+      "step": 4407
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4285646336742341e-05,
+      "loss": 2.1133,
+      "step": 4408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4168588800464566e-05,
+      "loss": 2.126,
+      "step": 4409
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4052009406001375e-05,
+      "loss": 2.2109,
+      "step": 4410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3935908209895542e-05,
+      "loss": 2.1787,
+      "step": 4411
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3820285268457578e-05,
+      "loss": 2.0654,
+      "step": 4412
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.370514063776629e-05,
+      "loss": 2.126,
+      "step": 4413
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3590474373668338e-05,
+      "loss": 1.9688,
+      "step": 4414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.347628653177857e-05,
+      "loss": 2.1777,
+      "step": 4415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3362577167479573e-05,
+      "loss": 2.2266,
+      "step": 4416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3249346335922008e-05,
+      "loss": 2.167,
+      "step": 4417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.313659409202439e-05,
+      "loss": 2.0518,
+      "step": 4418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3024320490472862e-05,
+      "loss": 2.1572,
+      "step": 4419
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2912525585722091e-05,
+      "loss": 2.1113,
+      "step": 4420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.280120943199381e-05,
+      "loss": 2.0967,
+      "step": 4421
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2690372083278056e-05,
+      "loss": 2.0693,
+      "step": 4422
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2580013593332495e-05,
+      "loss": 2.3184,
+      "step": 4423
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2470134015682311e-05,
+      "loss": 2.0781,
+      "step": 4424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2360733403620761e-05,
+      "loss": 2.2383,
+      "step": 4425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2251811810208736e-05,
+      "loss": 1.9893,
+      "step": 4426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2143369288274531e-05,
+      "loss": 2.2207,
+      "step": 4427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2035405890414297e-05,
+      "loss": 2.1934,
+      "step": 4428
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1927921668991815e-05,
+      "loss": 2.0215,
+      "step": 4429
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1820916676138383e-05,
+      "loss": 2.0137,
+      "step": 4430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.171439096375304e-05,
+      "loss": 1.9199,
+      "step": 4431
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1608344583502128e-05,
+      "loss": 2.1309,
+      "step": 4432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1502777586819391e-05,
+      "loss": 2.0439,
+      "step": 4433
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1397690024906648e-05,
+      "loss": 1.9355,
+      "step": 4434
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1293081948732575e-05,
+      "loss": 2.165,
+      "step": 4435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1188953409033476e-05,
+      "loss": 2.0635,
+      "step": 4436
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1085304456313394e-05,
+      "loss": 2.1484,
+      "step": 4437
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0982135140843231e-05,
+      "loss": 2.0957,
+      "step": 4438
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0879445512661624e-05,
+      "loss": 2.0645,
+      "step": 4439
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.077723562157451e-05,
+      "loss": 2.1426,
+      "step": 4440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0675505517155126e-05,
+      "loss": 2.1104,
+      "step": 4441
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0574255248743891e-05,
+      "loss": 2.0654,
+      "step": 4442
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0473484865448525e-05,
+      "loss": 1.9648,
+      "step": 4443
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0373194416144039e-05,
+      "loss": 2.001,
+      "step": 4444
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0273383949472859e-05,
+      "loss": 2.1816,
+      "step": 4445
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0174053513844373e-05,
+      "loss": 2.2441,
+      "step": 4446
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0075203157435042e-05,
+      "loss": 2.1348,
+      "step": 4447
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.976832928188961e-06,
+      "loss": 2.1465,
+      "step": 4448
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.878942873816854e-06,
+      "loss": 2.0117,
+      "step": 4449
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.78153304179663e-06,
+      "loss": 1.9766,
+      "step": 4450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.684603479373611e-06,
+      "loss": 2.0977,
+      "step": 4451
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.588154233559853e-06,
+      "loss": 2.2109,
+      "step": 4452
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.49218535113472e-06,
+      "loss": 2.2188,
+      "step": 4453
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.396696878644195e-06,
+      "loss": 2.1553,
+      "step": 4454
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.30168886240168e-06,
+      "loss": 2.1689,
+      "step": 4455
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.207161348487314e-06,
+      "loss": 2.2188,
+      "step": 4456
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.1131143827482e-06,
+      "loss": 2.3613,
+      "step": 4457
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.019548010798628e-06,
+      "loss": 2.0938,
+      "step": 4458
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.926462278019298e-06,
+      "loss": 2.0303,
+      "step": 4459
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.833857229558207e-06,
+      "loss": 2.1729,
+      "step": 4460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.741732910330092e-06,
+      "loss": 2.2344,
+      "step": 4461
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.650089365016433e-06,
+      "loss": 2.165,
+      "step": 4462
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.558926638065789e-06,
+      "loss": 2.0771,
+      "step": 4463
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.468244773693123e-06,
+      "loss": 2.0605,
+      "step": 4464
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.378043815880476e-06,
+      "loss": 2.251,
+      "step": 4465
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.28832380837663e-06,
+      "loss": 2.1758,
+      "step": 4466
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.199084794696888e-06,
+      "loss": 2.0908,
+      "step": 4467
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.110326818123515e-06,
+      "loss": 2.1533,
+      "step": 4468
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.0220499217053e-06,
+      "loss": 2.2578,
+      "step": 4469
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.93425414825777e-06,
+      "loss": 2.1074,
+      "step": 4470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.846939540363086e-06,
+      "loss": 2.0889,
+      "step": 4471
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.760106140369928e-06,
+      "loss": 2.1758,
+      "step": 4472
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.673753990393828e-06,
+      "loss": 2.1602,
+      "step": 4473
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.587883132316842e-06,
+      "loss": 2.166,
+      "step": 4474
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.50249360778732e-06,
+      "loss": 2.248,
+      "step": 4475
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.417585458220466e-06,
+      "loss": 2.1172,
+      "step": 4476
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.333158724797784e-06,
+      "loss": 2.1338,
+      "step": 4477
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.249213448467518e-06,
+      "loss": 2.0986,
+      "step": 4478
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.165749669944433e-06,
+      "loss": 2.2021,
+      "step": 4479
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.0827674297093694e-06,
+      "loss": 2.1992,
+      "step": 4480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.0002667680098e-06,
+      "loss": 2.1914,
+      "step": 4481
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.918247724859938e-06,
+      "loss": 2.2783,
+      "step": 4482
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.836710340039965e-06,
+      "loss": 2.0273,
+      "step": 4483
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.755654653096688e-06,
+      "loss": 2.1436,
+      "step": 4484
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.675080703343328e-06,
+      "loss": 2.0762,
+      "step": 4485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.594988529859181e-06,
+      "loss": 2.1162,
+      "step": 4486
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.515378171490172e-06,
+      "loss": 2.1729,
+      "step": 4487
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.436249666848415e-06,
+      "loss": 2.2021,
+      "step": 4488
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.357603054312211e-06,
+      "loss": 1.8857,
+      "step": 4489
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.279438372026491e-06,
+      "loss": 2.209,
+      "step": 4490
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.201755657901931e-06,
+      "loss": 2.1162,
+      "step": 4491
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.124554949615835e-06,
+      "loss": 2.0371,
+      "step": 4492
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.0478362846116965e-06,
+      "loss": 2.1172,
+      "step": 4493
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.971599700098973e-06,
+      "loss": 2.3555,
+      "step": 4494
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.895845233053643e-06,
+      "loss": 2.208,
+      "step": 4495
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.820572920217426e-06,
+      "loss": 2.0029,
+      "step": 4496
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.745782798098676e-06,
+      "loss": 2.1895,
+      "step": 4497
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6714749029713745e-06,
+      "loss": 2.083,
+      "step": 4498
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.597649270876138e-06,
+      "loss": 2.2598,
+      "step": 4499
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.524305937619212e-06,
+      "loss": 2.2422,
+      "step": 4500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.451444938773253e-06,
+      "loss": 2.1084,
+      "step": 4501
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.379066309676772e-06,
+      "loss": 2.1318,
+      "step": 4502
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.307170085434354e-06,
+      "loss": 2.2412,
+      "step": 4503
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.235756300916772e-06,
+      "loss": 2.2119,
+      "step": 4504
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.164824990760764e-06,
+      "loss": 2.2285,
+      "step": 4505
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.09437618936881e-06,
+      "loss": 2.1201,
+      "step": 4506
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0244099309095795e-06,
+      "loss": 2.0801,
+      "step": 4507
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.954926249317815e-06,
+      "loss": 2.0859,
+      "step": 4508
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.885925178293782e-06,
+      "loss": 2.0566,
+      "step": 4509
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8174067513042655e-06,
+      "loss": 2.083,
+      "step": 4510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.74937100158157e-06,
+      "loss": 2.1182,
+      "step": 4511
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.681817962123747e-06,
+      "loss": 2.1895,
+      "step": 4512
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.614747665695251e-06,
+      "loss": 2.0488,
+      "step": 4513
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.548160144825841e-06,
+      "loss": 2.1328,
+      "step": 4514
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4820554318115715e-06,
+      "loss": 2.0166,
+      "step": 4515
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.416433558714128e-06,
+      "loss": 2.1377,
+      "step": 4516
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.35129455736083e-06,
+      "loss": 2.1367,
+      "step": 4517
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2866384593451825e-06,
+      "loss": 2.084,
+      "step": 4518
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.222465296026212e-06,
+      "loss": 2.2285,
+      "step": 4519
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1587750985288e-06,
+      "loss": 2.0771,
+      "step": 4520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.09556789774368e-06,
+      "loss": 1.9912,
+      "step": 4521
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.032843724326996e-06,
+      "loss": 2.1289,
+      "step": 4522
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.970602608700969e-06,
+      "loss": 2.0498,
+      "step": 4523
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.90884458105345e-06,
+      "loss": 2.1211,
+      "step": 4524
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.847569671337703e-06,
+      "loss": 2.1338,
+      "step": 4525
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.786777909273176e-06,
+      "loss": 2.0869,
+      "step": 4526
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.726469324344617e-06,
+      "loss": 2.1475,
+      "step": 4527
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.666643945802406e-06,
+      "loss": 2.0957,
+      "step": 4528
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6073018026627767e-06,
+      "loss": 2.2148,
+      "step": 4529
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5484429237075954e-06,
+      "loss": 2.1895,
+      "step": 4530
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4900673374840262e-06,
+      "loss": 2.1025,
+      "step": 4531
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4321750723050883e-06,
+      "loss": 1.9863,
+      "step": 4532
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3747661562493202e-06,
+      "loss": 2.0674,
+      "step": 4533
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3178406171608944e-06,
+      "loss": 1.9746,
+      "step": 4534
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2613984826495023e-06,
+      "loss": 2.1006,
+      "step": 4535
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.205439780090358e-06,
+      "loss": 2.1104,
+      "step": 4536
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1499645366240837e-06,
+      "loss": 2.1104,
+      "step": 4537
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.094972779157046e-06,
+      "loss": 2.0215,
+      "step": 4538
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0404645343610205e-06,
+      "loss": 2.1318,
+      "step": 4539
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.986439828673082e-06,
+      "loss": 2.2529,
+      "step": 4540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9328986882961594e-06,
+      "loss": 2.2021,
+      "step": 4541
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8798411391983694e-06,
+      "loss": 2.082,
+      "step": 4542
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.827267207113349e-06,
+      "loss": 2.2012,
+      "step": 4543
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7751769175401454e-06,
+      "loss": 2.1523,
+      "step": 4544
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7235702957433273e-06,
+      "loss": 2.1846,
+      "step": 4545
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6724473667527617e-06,
+      "loss": 2.2129,
+      "step": 4546
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6218081553638363e-06,
+      "loss": 2.1162,
+      "step": 4547
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.571652686137238e-06,
+      "loss": 2.1729,
+      "step": 4548
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.521980983399064e-06,
+      "loss": 2.123,
+      "step": 4549
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4727930712408198e-06,
+      "loss": 2.1377,
+      "step": 4550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4240889735192006e-06,
+      "loss": 2.3311,
+      "step": 4551
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3758687138564218e-06,
+      "loss": 2.0498,
+      "step": 4552
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.328132315639997e-06,
+      "loss": 2.2031,
+      "step": 4553
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2808798020227393e-06,
+      "loss": 2.0762,
+      "step": 4554
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.23411119592265e-06,
+      "loss": 2.2148,
+      "step": 4555
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1878265200233617e-06,
+      "loss": 2.1406,
+      "step": 4556
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1420257967734723e-06,
+      "loss": 2.1973,
+      "step": 4557
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.09670904838688e-06,
+      "loss": 2.1416,
+      "step": 4558
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0518762968430025e-06,
+      "loss": 2.1562,
+      "step": 4559
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0075275638862243e-06,
+      "loss": 2.1221,
+      "step": 4560
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9636628710263394e-06,
+      "loss": 2.1387,
+      "step": 4561
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.920282239538218e-06,
+      "loss": 2.1445,
+      "step": 4562
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8773856904621412e-06,
+      "loss": 2.2812,
+      "step": 4563
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8349732446036882e-06,
+      "loss": 2.1846,
+      "step": 4564
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7930449225331824e-06,
+      "loss": 2.1484,
+      "step": 4565
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.75160074458669e-06,
+      "loss": 1.9814,
+      "step": 4566
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7106407308650207e-06,
+      "loss": 2.0791,
+      "step": 4567
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6701649012345055e-06,
+      "loss": 2.1992,
+      "step": 4568
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6301732753263298e-06,
+      "loss": 2.2285,
+      "step": 4569
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5906658725370893e-06,
+      "loss": 2.1533,
+      "step": 4570
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.551642712028345e-06,
+      "loss": 2.1514,
+      "step": 4571
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.513103812726957e-06,
+      "loss": 2.1914,
+      "step": 4572
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4750491933247513e-06,
+      "loss": 1.9385,
+      "step": 4573
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4374788722787414e-06,
+      "loss": 2.2188,
+      "step": 4574
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.400392867811129e-06,
+      "loss": 2.1865,
+      "step": 4575
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3637911979090812e-06,
+      "loss": 2.1377,
+      "step": 4576
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3276738803248423e-06,
+      "loss": 2.0254,
+      "step": 4577
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2920409325759552e-06,
+      "loss": 2.2061,
+      "step": 4578
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2568923719447068e-06,
+      "loss": 2.2256,
+      "step": 4579
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2222282154787933e-06,
+      "loss": 2.0332,
+      "step": 4580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1880484799907665e-06,
+      "loss": 2.1914,
+      "step": 4581
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1543531820582542e-06,
+      "loss": 2.1396,
+      "step": 4582
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1211423380239615e-06,
+      "loss": 2.1387,
+      "step": 4583
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0884159639955593e-06,
+      "loss": 2.2324,
+      "step": 4584
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.056174075845795e-06,
+      "loss": 2.1045,
+      "step": 4585
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0244166892124928e-06,
+      "loss": 1.9863,
+      "step": 4586
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.931438194983322e-07,
+      "loss": 2.1221,
+      "step": 4587
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.6235548187118e-07,
+      "loss": 2.166,
+      "step": 4588
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.320516912637577e-07,
+      "loss": 2.2256,
+      "step": 4589
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.022324623737532e-07,
+      "loss": 2.0586,
+      "step": 4590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.728978096640417e-07,
+      "loss": 2.1152,
+      "step": 4591
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.440477473622421e-07,
+      "loss": 2.0625,
+      "step": 4592
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.156822894610505e-07,
+      "loss": 2.043,
+      "step": 4593
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.878014497181285e-07,
+      "loss": 2.1514,
+      "step": 4594
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.604052416559925e-07,
+      "loss": 2.0273,
+      "step": 4595
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.334936785622359e-07,
+      "loss": 2.0605,
+      "step": 4596
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.070667734894176e-07,
+      "loss": 1.9609,
+      "step": 4597
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.811245392548405e-07,
+      "loss": 2.1641,
+      "step": 4598
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.556669884408839e-07,
+      "loss": 2.1582,
+      "step": 4599
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.306941333946715e-07,
+      "loss": 2.2041,
+      "step": 4600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.062059862286251e-07,
+      "loss": 2.0596,
+      "step": 4601
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.822025588196889e-07,
+      "loss": 2.0332,
+      "step": 4602
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.586838628099944e-07,
+      "loss": 2.1494,
+      "step": 4603
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.356499096061951e-07,
+      "loss": 1.9873,
+      "step": 4604
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.131007103802432e-07,
+      "loss": 2.0752,
+      "step": 4605
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.910362760688347e-07,
+      "loss": 2.1221,
+      "step": 4606
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6945661737340937e-07,
+      "loss": 2.0537,
+      "step": 4607
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4836174476037274e-07,
+      "loss": 2.2676,
+      "step": 4608
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.277516684612071e-07,
+      "loss": 2.2852,
+      "step": 4609
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0762639847191643e-07,
+      "loss": 2.2119,
+      "step": 4610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8798594455369265e-07,
+      "loss": 2.1221,
+      "step": 4611
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6883031623224927e-07,
+      "loss": 2.1025,
+      "step": 4612
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.501595227984877e-07,
+      "loss": 2.0645,
+      "step": 4613
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.319735733079421e-07,
+      "loss": 2.0889,
+      "step": 4614
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1427247658100124e-07,
+      "loss": 2.0186,
+      "step": 4615
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9705624120290876e-07,
+      "loss": 2.1221,
+      "step": 4616
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.803248755238741e-07,
+      "loss": 1.9814,
+      "step": 4617
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.640783876588504e-07,
+      "loss": 2.1689,
+      "step": 4618
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4831678548753456e-07,
+      "loss": 2.2119,
+      "step": 4619
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3304007665458927e-07,
+      "loss": 2.2041,
+      "step": 4620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1824826856942092e-07,
+      "loss": 2.2539,
+      "step": 4621
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0394136840617972e-07,
+      "loss": 2.0986,
+      "step": 4622
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9011938310387056e-07,
+      "loss": 2.0967,
+      "step": 4623
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7678231936657518e-07,
+      "loss": 2.082,
+      "step": 4624
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.63930183662786e-07,
+      "loss": 2.0322,
+      "step": 4625
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.515629822259612e-07,
+      "loss": 1.9717,
+      "step": 4626
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3968072105441375e-07,
+      "loss": 2.1797,
+      "step": 4627
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2828340591120035e-07,
+      "loss": 2.1572,
+      "step": 4628
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1737104232412144e-07,
+      "loss": 2.1426,
+      "step": 4629
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0694363558594323e-07,
+      "loss": 2.0801,
+      "step": 4630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.700119075395363e-08,
+      "loss": 2.0889,
+      "step": 4631
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.754371265040639e-08,
+      "loss": 2.3115,
+      "step": 4632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.857120586240996e-08,
+      "loss": 1.9727,
+      "step": 4633
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.008367474170551e-08,
+      "loss": 2.2646,
+      "step": 4634
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.208112340488902e-08,
+      "loss": 2.0127,
+      "step": 4635
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.4563555733189163e-08,
+      "loss": 2.1094,
+      "step": 4636
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7530975372800375e-08,
+      "loss": 2.1504,
+      "step": 4637
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.098338573466087e-08,
+      "loss": 2.0293,
+      "step": 4638
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.492078999434156e-08,
+      "loss": 2.0732,
+      "step": 4639
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9343191092490173e-08,
+      "loss": 2.1631,
+      "step": 4640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4250591734054083e-08,
+      "loss": 2.1201,
+      "step": 4641
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.96429943891685e-08,
+      "loss": 2.2236,
+      "step": 4642
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552040129260135e-08,
+      "loss": 1.9844,
+      "step": 4643
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1882814443864298e-08,
+      "loss": 2.2998,
+      "step": 4644
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.730235607101733e-09,
+      "loss": 2.1611,
+      "step": 4645
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.062666311534848e-09,
+      "loss": 2.1484,
+      "step": 4646
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8801078507955115e-09,
+      "loss": 2.1387,
+      "step": 4647
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.182561283592399e-09,
+      "loss": 2.0029,
+      "step": 4648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.700274331558844e-10,
+      "loss": 2.0723,
+      "step": 4649
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.425068876821257e-10,
+      "loss": 2.0986,
+      "step": 4650
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 2.0771,
+      "step": 4651
+    },
+    {
+      "epoch": 1.0,
+      "step": 4651,
+      "total_flos": 550753257259008.0,
+      "train_loss": 2.269884416657708,
+      "train_runtime": 29841.694,
+      "train_samples_per_second": 19.951,
+      "train_steps_per_second": 0.156
+    }
+  ],
+  "max_steps": 4651,
+  "num_train_epochs": 1,
+  "total_flos": 550753257259008.0,
+  "trial_name": null,
+  "trial_params": null
+}