diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,75312 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9997609561752988,
+  "eval_steps": 500,
+  "global_step": 12547,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6525198938992043e-07,
+      "loss": 1.1438,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.305039787798409e-07,
+      "loss": 1.1849,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.957559681697613e-07,
+      "loss": 78667.75,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0610079575596817e-06,
+      "loss": 116534.0234,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3262599469496022e-06,
+      "loss": 2.016,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5915119363395226e-06,
+      "loss": 1.2084,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.856763925729443e-06,
+      "loss": 131126.9062,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1220159151193635e-06,
+      "loss": 1.1197,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3872679045092837e-06,
+      "loss": 116836.3906,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6525198938992043e-06,
+      "loss": 99199.8047,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.917771883289125e-06,
+      "loss": 2.0361,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.183023872679045e-06,
+      "loss": 1.1692,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.448275862068966e-06,
+      "loss": 1.1501,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.713527851458886e-06,
+      "loss": 1.1418,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.978779840848806e-06,
+      "loss": 96497.7109,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.244031830238727e-06,
+      "loss": 2.0812,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5092838196286476e-06,
+      "loss": 85363.2969,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.774535809018567e-06,
+      "loss": 85148.2891,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.039787798408489e-06,
+      "loss": 79336.5234,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.305039787798409e-06,
+      "loss": 1.1527,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.570291777188329e-06,
+      "loss": 72181.5469,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.83554376657825e-06,
+      "loss": 64859.043,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.10079575596817e-06,
+      "loss": 70854.9766,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.36604774535809e-06,
+      "loss": 2.0746,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.631299734748012e-06,
+      "loss": 64783.7305,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.896551724137932e-06,
+      "loss": 54746.8672,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.161803713527852e-06,
+      "loss": 1.1066,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.427055702917772e-06,
+      "loss": 1.1497,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 26047.8711,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.957559681697613e-06,
+      "loss": 20386.918,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.222811671087533e-06,
+      "loss": 1.1175,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.488063660477454e-06,
+      "loss": 20767.6016,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.753315649867374e-06,
+      "loss": 2.0749,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.018567639257295e-06,
+      "loss": 1.9958,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.283819628647216e-06,
+      "loss": 14124.5654,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.549071618037135e-06,
+      "loss": 11819.8281,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.814323607427055e-06,
+      "loss": 5490.8789,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0079575596816978e-05,
+      "loss": 2.0373,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0344827586206897e-05,
+      "loss": 1.1114,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0610079575596817e-05,
+      "loss": 1.9692,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0875331564986738e-05,
+      "loss": 2.0309,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1140583554376659e-05,
+      "loss": 2.0604,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.140583554376658e-05,
+      "loss": 4333.25,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.16710875331565e-05,
+      "loss": 1.1414,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1936339522546419e-05,
+      "loss": 2375.7837,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.220159151193634e-05,
+      "loss": 2480.5154,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.246684350132626e-05,
+      "loss": 2.0685,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.273209549071618e-05,
+      "loss": 1884.1302,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2997347480106101e-05,
+      "loss": 1.0927,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3262599469496024e-05,
+      "loss": 1301.2297,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3527851458885943e-05,
+      "loss": 1.2443,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3793103448275863e-05,
+      "loss": 1.1608,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4058355437665782e-05,
+      "loss": 1719.1904,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4323607427055705e-05,
+      "loss": 691.9996,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4588859416445624e-05,
+      "loss": 539.2179,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4854111405835544e-05,
+      "loss": 1.1456,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5119363395225463e-05,
+      "loss": 423.6406,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 454.1903,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5649867374005304e-05,
+      "loss": 389.9123,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5915119363395225e-05,
+      "loss": 1.1586,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.618037135278515e-05,
+      "loss": 1.9938,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6445623342175066e-05,
+      "loss": 382.8839,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6710875331564987e-05,
+      "loss": 389.3924,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6976127320954908e-05,
+      "loss": 384.2767,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7241379310344828e-05,
+      "loss": 373.0518,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.750663129973475e-05,
+      "loss": 1.117,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.777188328912467e-05,
+      "loss": 365.0361,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.803713527851459e-05,
+      "loss": 371.5454,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.830238726790451e-05,
+      "loss": 1.1146,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.856763925729443e-05,
+      "loss": 1.133,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.883289124668435e-05,
+      "loss": 1.9581,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.909814323607427e-05,
+      "loss": 329.4828,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9363395225464193e-05,
+      "loss": 360.2592,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.962864721485411e-05,
+      "loss": 323.4049,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.989389920424403e-05,
+      "loss": 318.7509,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0159151193633955e-05,
+      "loss": 299.1027,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0424403183023873e-05,
+      "loss": 293.281,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0689655172413793e-05,
+      "loss": 1.1455,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0954907161803714e-05,
+      "loss": 1.1643,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1220159151193635e-05,
+      "loss": 1.1752,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1485411140583555e-05,
+      "loss": 1.0841,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1750663129973476e-05,
+      "loss": 271.4213,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2015915119363396e-05,
+      "loss": 1.0942,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2281167108753317e-05,
+      "loss": 270.0804,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2546419098143238e-05,
+      "loss": 261.1505,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.281167108753316e-05,
+      "loss": 232.6382,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.307692307692308e-05,
+      "loss": 265.482,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3342175066313e-05,
+      "loss": 1.1028,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.360742705570292e-05,
+      "loss": 1.9207,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3872679045092838e-05,
+      "loss": 1.8309,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.413793103448276e-05,
+      "loss": 1.0432,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.440318302387268e-05,
+      "loss": 1.8596,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.46684350132626e-05,
+      "loss": 1.1064,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.493368700265252e-05,
+      "loss": 223.9126,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.519893899204244e-05,
+      "loss": 1.1257,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.546419098143236e-05,
+      "loss": 222.718,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5729442970822282e-05,
+      "loss": 220.6425,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5994694960212203e-05,
+      "loss": 234.346,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.625994694960212e-05,
+      "loss": 1.1053,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6525198938992047e-05,
+      "loss": 215.3165,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6790450928381965e-05,
+      "loss": 201.4491,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7055702917771885e-05,
+      "loss": 195.7555,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7320954907161806e-05,
+      "loss": 194.3041,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7586206896551727e-05,
+      "loss": 1.7546,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7851458885941644e-05,
+      "loss": 1.0609,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8116710875331565e-05,
+      "loss": 186.4985,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8381962864721485e-05,
+      "loss": 180.6586,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.864721485411141e-05,
+      "loss": 177.7431,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.891246684350133e-05,
+      "loss": 1.8257,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9177718832891247e-05,
+      "loss": 187.3801,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9442970822281168e-05,
+      "loss": 1.0851,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.970822281167109e-05,
+      "loss": 166.7317,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.997347480106101e-05,
+      "loss": 160.6154,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0238726790450926e-05,
+      "loss": 1.0568,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0503978779840854e-05,
+      "loss": 1.0968,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0769230769230774e-05,
+      "loss": 157.903,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.103448275862069e-05,
+      "loss": 159.0765,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.129973474801061e-05,
+      "loss": 1.7854,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.156498673740053e-05,
+      "loss": 1.0523,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.183023872679045e-05,
+      "loss": 169.4536,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2095490716180374e-05,
+      "loss": 1.0641,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.23607427055703e-05,
+      "loss": 151.2007,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2625994694960215e-05,
+      "loss": 145.9515,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.289124668435013e-05,
+      "loss": 149.296,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.315649867374006e-05,
+      "loss": 1.6705,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3421750663129974e-05,
+      "loss": 137.8877,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.368700265251989e-05,
+      "loss": 1.7182,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3952254641909815e-05,
+      "loss": 146.9661,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.421750663129973e-05,
+      "loss": 137.856,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4482758620689657e-05,
+      "loss": 1.0947,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.474801061007958e-05,
+      "loss": 140.0615,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.50132625994695e-05,
+      "loss": 134.6272,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5278514588859415e-05,
+      "loss": 140.5058,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.554376657824934e-05,
+      "loss": 141.6792,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5809018567639256e-05,
+      "loss": 129.391,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.607427055702918e-05,
+      "loss": 126.2296,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6339522546419104e-05,
+      "loss": 1.0623,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.660477453580902e-05,
+      "loss": 155.9792,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.687002652519894e-05,
+      "loss": 1.5996,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.713527851458886e-05,
+      "loss": 1.0305,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.740053050397878e-05,
+      "loss": 209.1221,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.76657824933687e-05,
+      "loss": 147.9033,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.793103448275862e-05,
+      "loss": 118.6487,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.819628647214854e-05,
+      "loss": 124.6899,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.846153846153846e-05,
+      "loss": 1.0433,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.872679045092839e-05,
+      "loss": 1.0257,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8992042440318304e-05,
+      "loss": 118.0657,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.925729442970822e-05,
+      "loss": 1.6625,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9522546419098145e-05,
+      "loss": 129.51,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.978779840848806e-05,
+      "loss": 1.57,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.005305039787799e-05,
+      "loss": 116.8382,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.031830238726791e-05,
+      "loss": 108.7079,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.058355437665783e-05,
+      "loss": 1.015,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0848806366047745e-05,
+      "loss": 118.4852,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.111405835543767e-05,
+      "loss": 109.3836,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1379310344827587e-05,
+      "loss": 1.0522,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.164456233421751e-05,
+      "loss": 1.6135,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.190981432360743e-05,
+      "loss": 114.3172,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.217506631299735e-05,
+      "loss": 109.3478,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.244031830238727e-05,
+      "loss": 111.5312,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.270557029177719e-05,
+      "loss": 124.475,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.297082228116711e-05,
+      "loss": 107.6154,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.323607427055703e-05,
+      "loss": 105.0226,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.350132625994695e-05,
+      "loss": 1.5187,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.376657824933687e-05,
+      "loss": 1.0226,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.403183023872679e-05,
+      "loss": 114.2837,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.429708222811672e-05,
+      "loss": 108.3702,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4562334217506634e-05,
+      "loss": 105.9932,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.482758620689655e-05,
+      "loss": 106.0498,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5092838196286476e-05,
+      "loss": 1.4861,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.535809018567639e-05,
+      "loss": 1.4858,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.562334217506632e-05,
+      "loss": 121.2421,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5888594164456234e-05,
+      "loss": 1.0467,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.615384615384616e-05,
+      "loss": 108.0447,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6419098143236075e-05,
+      "loss": 105.2491,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6684350132626e-05,
+      "loss": 1.0022,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.694960212201592e-05,
+      "loss": 108.6749,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.721485411140584e-05,
+      "loss": 1.0496,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.748010610079576e-05,
+      "loss": 1.06,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7745358090185675e-05,
+      "loss": 101.1895,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.80106100795756e-05,
+      "loss": 103.607,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.827586206896552e-05,
+      "loss": 105.4453,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.854111405835544e-05,
+      "loss": 0.9979,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.880636604774536e-05,
+      "loss": 1.0466,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.907161803713528e-05,
+      "loss": 1.122,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.93368700265252e-05,
+      "loss": 100.7116,
+      "step": 186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.960212201591512e-05,
+      "loss": 100.7483,
+      "step": 187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.986737400530504e-05,
+      "loss": 101.4875,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.013262599469496e-05,
+      "loss": 1.0697,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.039787798408488e-05,
+      "loss": 98.5862,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.06631299734748e-05,
+      "loss": 97.4005,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.092838196286472e-05,
+      "loss": 1.3986,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.119363395225465e-05,
+      "loss": 1.0767,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1458885941644564e-05,
+      "loss": 99.9202,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.172413793103449e-05,
+      "loss": 93.1333,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1989389920424406e-05,
+      "loss": 1.3565,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.225464190981433e-05,
+      "loss": 94.543,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.251989389920424e-05,
+      "loss": 92.4249,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2785145888594164e-05,
+      "loss": 94.0831,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3050397877984095e-05,
+      "loss": 91.9573,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3315649867374005e-05,
+      "loss": 1.0199,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.358090185676393e-05,
+      "loss": 98.9728,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.384615384615385e-05,
+      "loss": 1.2879,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.411140583554377e-05,
+      "loss": 1.2819,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.437665782493369e-05,
+      "loss": 1.0203,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.464190981432361e-05,
+      "loss": 91.1127,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4907161803713536e-05,
+      "loss": 1.271,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.517241379310345e-05,
+      "loss": 88.9557,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.543766578249338e-05,
+      "loss": 1.2665,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.570291777188329e-05,
+      "loss": 101.2674,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.596816976127321e-05,
+      "loss": 1.0232,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.623342175066313e-05,
+      "loss": 1.1991,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.649867374005305e-05,
+      "loss": 91.1892,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.676392572944297e-05,
+      "loss": 89.3168,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.7029177718832894e-05,
+      "loss": 92.7059,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.729442970822282e-05,
+      "loss": 93.0447,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.7559681697612736e-05,
+      "loss": 0.9756,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.782493368700266e-05,
+      "loss": 89.1112,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.809018567639257e-05,
+      "loss": 1.1434,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8355437665782494e-05,
+      "loss": 88.256,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.862068965517241e-05,
+      "loss": 1.0268,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8885941644562335e-05,
+      "loss": 1.0944,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.915119363395226e-05,
+      "loss": 95.5426,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.941644562334218e-05,
+      "loss": 86.5213,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.96816976127321e-05,
+      "loss": 88.0832,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.994694960212202e-05,
+      "loss": 1.12,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.021220159151194e-05,
+      "loss": 1.0794,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.047745358090185e-05,
+      "loss": 1.0194,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.0742705570291783e-05,
+      "loss": 93.3819,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.100795755968171e-05,
+      "loss": 1.0433,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.127320954907162e-05,
+      "loss": 1.0773,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.153846153846155e-05,
+      "loss": 89.1942,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.180371352785146e-05,
+      "loss": 1.0695,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.206896551724138e-05,
+      "loss": 1.0206,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.23342175066313e-05,
+      "loss": 0.99,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.259946949602122e-05,
+      "loss": 93.6153,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.286472148541116e-05,
+      "loss": 1.0306,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.312997347480107e-05,
+      "loss": 86.6046,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.339522546419099e-05,
+      "loss": 88.6429,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.36604774535809e-05,
+      "loss": 1.009,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.392572944297082e-05,
+      "loss": 103.7266,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.419098143236075e-05,
+      "loss": 99.0576,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.445623342175067e-05,
+      "loss": 1.0127,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.47214854111406e-05,
+      "loss": 0.9713,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.498673740053051e-05,
+      "loss": 1.0488,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.525198938992043e-05,
+      "loss": 0.9644,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.551724137931034e-05,
+      "loss": 0.9583,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.578249336870027e-05,
+      "loss": 102.9239,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.604774535809019e-05,
+      "loss": 0.9651,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.631299734748011e-05,
+      "loss": 1.0349,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.657824933687002e-05,
+      "loss": 0.9955,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.684350132625995e-05,
+      "loss": 102.8532,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.710875331564987e-05,
+      "loss": 0.9851,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.737400530503978e-05,
+      "loss": 89.9916,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.763925729442972e-05,
+      "loss": 0.9276,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.790450928381963e-05,
+      "loss": 96.0695,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.816976127320955e-05,
+      "loss": 0.9995,
+      "step": 257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.843501326259947e-05,
+      "loss": 81.5731,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.870026525198939e-05,
+      "loss": 0.8199,
+      "step": 259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.896551724137931e-05,
+      "loss": 0.9651,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.923076923076924e-05,
+      "loss": 93.5618,
+      "step": 261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.949602122015916e-05,
+      "loss": 1.0194,
+      "step": 262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.976127320954907e-05,
+      "loss": 1.0407,
+      "step": 263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.0026525198939e-05,
+      "loss": 97.9513,
+      "step": 264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.02917771883289e-05,
+      "loss": 0.8069,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.055702917771883e-05,
+      "loss": 90.9445,
+      "step": 266
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.082228116710877e-05,
+      "loss": 0.9834,
+      "step": 267
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.108753315649868e-05,
+      "loss": 91.383,
+      "step": 268
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.13527851458886e-05,
+      "loss": 1.0119,
+      "step": 269
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.161803713527851e-05,
+      "loss": 78.8975,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.188328912466844e-05,
+      "loss": 0.8321,
+      "step": 271
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.214854111405836e-05,
+      "loss": 1.0666,
+      "step": 272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.241379310344828e-05,
+      "loss": 84.7708,
+      "step": 273
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.267904509283821e-05,
+      "loss": 1.0042,
+      "step": 274
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.294429708222812e-05,
+      "loss": 88.0736,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.320954907161804e-05,
+      "loss": 1.0018,
+      "step": 276
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.347480106100795e-05,
+      "loss": 1.0183,
+      "step": 277
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.374005305039788e-05,
+      "loss": 0.9837,
+      "step": 278
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.40053050397878e-05,
+      "loss": 88.5162,
+      "step": 279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.427055702917773e-05,
+      "loss": 0.7799,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.453580901856765e-05,
+      "loss": 92.5571,
+      "step": 281
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.480106100795756e-05,
+      "loss": 0.9718,
+      "step": 282
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.506631299734748e-05,
+      "loss": 1.0291,
+      "step": 283
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.53315649867374e-05,
+      "loss": 79.1633,
+      "step": 284
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.559681697612733e-05,
+      "loss": 0.9944,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.586206896551724e-05,
+      "loss": 92.5145,
+      "step": 286
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.612732095490717e-05,
+      "loss": 93.2599,
+      "step": 287
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.639257294429708e-05,
+      "loss": 82.0209,
+      "step": 288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.6657824933687e-05,
+      "loss": 82.7437,
+      "step": 289
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.692307692307693e-05,
+      "loss": 0.7462,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.718832891246685e-05,
+      "loss": 81.1237,
+      "step": 291
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.745358090185677e-05,
+      "loss": 81.443,
+      "step": 292
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.771883289124668e-05,
+      "loss": 82.9081,
+      "step": 293
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.798408488063661e-05,
+      "loss": 81.9265,
+      "step": 294
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.824933687002652e-05,
+      "loss": 1.0398,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.851458885941644e-05,
+      "loss": 88.7073,
+      "step": 296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.877984084880638e-05,
+      "loss": 1.0163,
+      "step": 297
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.904509283819629e-05,
+      "loss": 0.6908,
+      "step": 298
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.931034482758621e-05,
+      "loss": 88.1923,
+      "step": 299
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.957559681697613e-05,
+      "loss": 0.7226,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.984084880636605e-05,
+      "loss": 87.261,
+      "step": 301
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.010610079575597e-05,
+      "loss": 0.9945,
+      "step": 302
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.03713527851459e-05,
+      "loss": 88.3708,
+      "step": 303
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.063660477453582e-05,
+      "loss": 89.1978,
+      "step": 304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.090185676392573e-05,
+      "loss": 0.9935,
+      "step": 305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.116710875331566e-05,
+      "loss": 1.0075,
+      "step": 306
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.143236074270557e-05,
+      "loss": 0.9789,
+      "step": 307
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.169761273209549e-05,
+      "loss": 331.5893,
+      "step": 308
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.196286472148541e-05,
+      "loss": 235.2816,
+      "step": 309
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.222811671087534e-05,
+      "loss": 116.4192,
+      "step": 310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.249336870026526e-05,
+      "loss": 1.0003,
+      "step": 311
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.275862068965517e-05,
+      "loss": 81.619,
+      "step": 312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.30238726790451e-05,
+      "loss": 84.3221,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.328912466843502e-05,
+      "loss": 1.0682,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.355437665782495e-05,
+      "loss": 0.6522,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.381962864721486e-05,
+      "loss": 0.6493,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.408488063660478e-05,
+      "loss": 83.2849,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.43501326259947e-05,
+      "loss": 80.939,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.461538461538461e-05,
+      "loss": 78.8991,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.488063660477454e-05,
+      "loss": 80.1076,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.514588859416446e-05,
+      "loss": 78.3018,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.541114058355439e-05,
+      "loss": 83.9089,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.56763925729443e-05,
+      "loss": 75.7313,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.594164456233422e-05,
+      "loss": 71.6388,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.620689655172413e-05,
+      "loss": 77.0513,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.647214854111406e-05,
+      "loss": 74.1877,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.673740053050399e-05,
+      "loss": 80.7892,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.70026525198939e-05,
+      "loss": 0.9984,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.726790450928383e-05,
+      "loss": 1.0402,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.753315649867374e-05,
+      "loss": 93.0129,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.779840848806366e-05,
+      "loss": 0.6148,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.806366047745359e-05,
+      "loss": 0.6229,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.832891246684351e-05,
+      "loss": 0.5419,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.859416445623343e-05,
+      "loss": 78.0339,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.885941644562334e-05,
+      "loss": 0.5632,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.912466843501327e-05,
+      "loss": 77.7043,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.938992042440318e-05,
+      "loss": 81.5605,
+      "step": 337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.96551724137931e-05,
+      "loss": 73.4402,
+      "step": 338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.992042440318303e-05,
+      "loss": 78.5478,
+      "step": 339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.018567639257295e-05,
+      "loss": 0.957,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.045092838196288e-05,
+      "loss": 96.4714,
+      "step": 341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.071618037135279e-05,
+      "loss": 0.555,
+      "step": 342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.098143236074271e-05,
+      "loss": 1.0372,
+      "step": 343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.124668435013263e-05,
+      "loss": 93.8684,
+      "step": 344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.151193633952256e-05,
+      "loss": 0.9774,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.177718832891247e-05,
+      "loss": 0.4805,
+      "step": 346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.204244031830239e-05,
+      "loss": 79.8396,
+      "step": 347
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.230769230769232e-05,
+      "loss": 78.6151,
+      "step": 348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.257294429708223e-05,
+      "loss": 0.4689,
+      "step": 349
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.283819628647215e-05,
+      "loss": 0.989,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.310344827586207e-05,
+      "loss": 1.0326,
+      "step": 351
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.3368700265252e-05,
+      "loss": 168.3048,
+      "step": 352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.363395225464191e-05,
+      "loss": 134.5511,
+      "step": 353
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.389920424403183e-05,
+      "loss": 89.2469,
+      "step": 354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.416445623342176e-05,
+      "loss": 77.1235,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.442970822281168e-05,
+      "loss": 0.4797,
+      "step": 356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.46949602122016e-05,
+      "loss": 81.3354,
+      "step": 357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.496021220159152e-05,
+      "loss": 85.8969,
+      "step": 358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.522546419098144e-05,
+      "loss": 78.8194,
+      "step": 359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.549071618037135e-05,
+      "loss": 1.0121,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.575596816976127e-05,
+      "loss": 1.0313,
+      "step": 361
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.60212201591512e-05,
+      "loss": 0.9854,
+      "step": 362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.628647214854112e-05,
+      "loss": 204.1228,
+      "step": 363
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.655172413793105e-05,
+      "loss": 0.981,
+      "step": 364
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.681697612732096e-05,
+      "loss": 0.9922,
+      "step": 365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.708222811671088e-05,
+      "loss": 181.7419,
+      "step": 366
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.734748010610079e-05,
+      "loss": 0.9919,
+      "step": 367
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.761273209549072e-05,
+      "loss": 116.2884,
+      "step": 368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.787798408488064e-05,
+      "loss": 0.4841,
+      "step": 369
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.814323607427056e-05,
+      "loss": 83.0612,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.840848806366049e-05,
+      "loss": 0.3841,
+      "step": 371
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.86737400530504e-05,
+      "loss": 88.5745,
+      "step": 372
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.893899204244032e-05,
+      "loss": 82.5545,
+      "step": 373
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.920424403183025e-05,
+      "loss": 1.0076,
+      "step": 374
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.946949602122017e-05,
+      "loss": 0.4153,
+      "step": 375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973474801061008e-05,
+      "loss": 0.9772,
+      "step": 376
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001,
+      "loss": 0.9869,
+      "step": 377
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999999833488396e-05,
+      "loss": 82.8654,
+      "step": 378
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999999333953591e-05,
+      "loss": 78.5671,
+      "step": 379
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99999850139562e-05,
+      "loss": 1.0519,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999997335814539e-05,
+      "loss": 0.4435,
+      "step": 381
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999995837210424e-05,
+      "loss": 83.5828,
+      "step": 382
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999994005583376e-05,
+      "loss": 79.008,
+      "step": 383
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999991840933518e-05,
+      "loss": 1.0493,
+      "step": 384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999989343260992e-05,
+      "loss": 0.4483,
+      "step": 385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999986512565964e-05,
+      "loss": 80.6994,
+      "step": 386
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999983348848625e-05,
+      "loss": 75.5787,
+      "step": 387
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999979852109185e-05,
+      "loss": 72.5209,
+      "step": 388
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999976022347876e-05,
+      "loss": 1.0269,
+      "step": 389
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999971859564954e-05,
+      "loss": 72.8698,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999967363760696e-05,
+      "loss": 0.9817,
+      "step": 391
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999962534935401e-05,
+      "loss": 74.721,
+      "step": 392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99995737308939e-05,
+      "loss": 0.3467,
+      "step": 393
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999951878223008e-05,
+      "loss": 0.3649,
+      "step": 394
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999946050336621e-05,
+      "loss": 0.3912,
+      "step": 395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999939889430618e-05,
+      "loss": 82.2541,
+      "step": 396
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999933395505406e-05,
+      "loss": 79.772,
+      "step": 397
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99992656856142e-05,
+      "loss": 65.3157,
+      "step": 398
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999919408599115e-05,
+      "loss": 0.3794,
+      "step": 399
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999911915618968e-05,
+      "loss": 80.6593,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999904089621476e-05,
+      "loss": 77.4088,
+      "step": 401
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999895930607163e-05,
+      "loss": 0.3846,
+      "step": 402
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99988743857657e-05,
+      "loss": 78.2599,
+      "step": 403
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999878613530263e-05,
+      "loss": 1.022,
+      "step": 404
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999869455468833e-05,
+      "loss": 0.976,
+      "step": 405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999859964392885e-05,
+      "loss": 74.0803,
+      "step": 406
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999850140303053e-05,
+      "loss": 75.5381,
+      "step": 407
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999839983199995e-05,
+      "loss": 0.9493,
+      "step": 408
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99982949308438e-05,
+      "loss": 74.3843,
+      "step": 409
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999818669956915e-05,
+      "loss": 78.993,
+      "step": 410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999807513818315e-05,
+      "loss": 72.1346,
+      "step": 411
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999796024669326e-05,
+      "loss": 1.013,
+      "step": 412
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999784202510711e-05,
+      "loss": 0.9933,
+      "step": 413
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99977204734326e-05,
+      "loss": 0.9986,
+      "step": 414
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99975955916778e-05,
+      "loss": 0.307,
+      "step": 415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999746737985103e-05,
+      "loss": 1.0194,
+      "step": 416
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999733583796087e-05,
+      "loss": 0.99,
+      "step": 417
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999720096601603e-05,
+      "loss": 0.367,
+      "step": 418
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999706276402552e-05,
+      "loss": 1.0226,
+      "step": 419
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999692123199853e-05,
+      "loss": 0.3755,
+      "step": 420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999677636994451e-05,
+      "loss": 265.6756,
+      "step": 421
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999662817787309e-05,
+      "loss": 132.2463,
+      "step": 422
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999647665579414e-05,
+      "loss": 89.5892,
+      "step": 423
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999632180371776e-05,
+      "loss": 0.3471,
+      "step": 424
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999616362165425e-05,
+      "loss": 93.9717,
+      "step": 425
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999600210961416e-05,
+      "loss": 0.3322,
+      "step": 426
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999583726760825e-05,
+      "loss": 92.9762,
+      "step": 427
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999566909564748e-05,
+      "loss": 1.0223,
+      "step": 428
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999549759374307e-05,
+      "loss": 86.295,
+      "step": 429
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999532276190644e-05,
+      "loss": 86.4494,
+      "step": 430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999514460014923e-05,
+      "loss": 0.4141,
+      "step": 431
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999496310848331e-05,
+      "loss": 85.1278,
+      "step": 432
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999477828692077e-05,
+      "loss": 77.4984,
+      "step": 433
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99945901354739e-05,
+      "loss": 0.984,
+      "step": 434
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999439865415525e-05,
+      "loss": 1.0146,
+      "step": 435
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999420384297758e-05,
+      "loss": 84.0189,
+      "step": 436
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999400570195385e-05,
+      "loss": 0.3588,
+      "step": 437
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999380423109726e-05,
+      "loss": 77.2802,
+      "step": 438
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999359943042124e-05,
+      "loss": 0.9773,
+      "step": 439
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999339129993942e-05,
+      "loss": 82.0883,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999317983966566e-05,
+      "loss": 0.3804,
+      "step": 441
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999296504961406e-05,
+      "loss": 76.5529,
+      "step": 442
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999274692979891e-05,
+      "loss": 75.5094,
+      "step": 443
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999252548023474e-05,
+      "loss": 75.4063,
+      "step": 444
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99923007009363e-05,
+      "loss": 1.0023,
+      "step": 445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999207259191859e-05,
+      "loss": 0.3312,
+      "step": 446
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999184115319675e-05,
+      "loss": 1.0156,
+      "step": 447
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999160638478621e-05,
+      "loss": 72.2757,
+      "step": 448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999136828670266e-05,
+      "loss": 0.9921,
+      "step": 449
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999112685896188e-05,
+      "loss": 0.3171,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999088210158001e-05,
+      "loss": 1.0295,
+      "step": 451
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999063401457333e-05,
+      "loss": 0.2773,
+      "step": 452
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999038259795834e-05,
+      "loss": 77.071,
+      "step": 453
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999012785175183e-05,
+      "loss": 1.0093,
+      "step": 454
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998986977597075e-05,
+      "loss": 0.3021,
+      "step": 455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998960837063227e-05,
+      "loss": 0.3259,
+      "step": 456
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998934363575383e-05,
+      "loss": 72.4336,
+      "step": 457
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998907557135304e-05,
+      "loss": 1.0245,
+      "step": 458
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998880417744776e-05,
+      "loss": 69.0884,
+      "step": 459
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998852945405607e-05,
+      "loss": 68.9443,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998825140119628e-05,
+      "loss": 71.189,
+      "step": 461
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998797001888687e-05,
+      "loss": 70.9115,
+      "step": 462
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998768530714663e-05,
+      "loss": 68.693,
+      "step": 463
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998739726599449e-05,
+      "loss": 64.8195,
+      "step": 464
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998710589544965e-05,
+      "loss": 0.2915,
+      "step": 465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998681119553151e-05,
+      "loss": 65.072,
+      "step": 466
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99865131662597e-05,
+      "loss": 0.3001,
+      "step": 467
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998621180765409e-05,
+      "loss": 0.3178,
+      "step": 468
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99859071197347e-05,
+      "loss": 69.5364,
+      "step": 469
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998559910252188e-05,
+      "loss": 70.0809,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998528775603611e-05,
+      "loss": 64.8861,
+      "step": 471
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998497308029815e-05,
+      "loss": 0.9871,
+      "step": 472
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998465507532893e-05,
+      "loss": 65.371,
+      "step": 473
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998433374114966e-05,
+      "loss": 70.8288,
+      "step": 474
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998400907778174e-05,
+      "loss": 0.9779,
+      "step": 475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998368108524676e-05,
+      "loss": 69.6419,
+      "step": 476
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998334976356661e-05,
+      "loss": 1.0372,
+      "step": 477
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998301511276333e-05,
+      "loss": 0.9742,
+      "step": 478
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998267713285923e-05,
+      "loss": 0.9882,
+      "step": 479
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99823358238768e-05,
+      "loss": 0.3679,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998199118583877e-05,
+      "loss": 1.0097,
+      "step": 481
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998164321876813e-05,
+      "loss": 68.7706,
+      "step": 482
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998129192268801e-05,
+      "loss": 68.3933,
+      "step": 483
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998093729762183e-05,
+      "loss": 0.2785,
+      "step": 484
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998057934359323e-05,
+      "loss": 75.2287,
+      "step": 485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998021806062602e-05,
+      "loss": 0.3694,
+      "step": 486
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997985344874428e-05,
+      "loss": 96.1501,
+      "step": 487
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997948550797227e-05,
+      "loss": 0.3224,
+      "step": 488
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997911423833453e-05,
+      "loss": 78.2589,
+      "step": 489
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997873963985578e-05,
+      "loss": 63.9594,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997836171256094e-05,
+      "loss": 68.9724,
+      "step": 491
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997798045647523e-05,
+      "loss": 0.2895,
+      "step": 492
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997759587162401e-05,
+      "loss": 1.0333,
+      "step": 493
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997720795803291e-05,
+      "loss": 0.2879,
+      "step": 494
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997681671572775e-05,
+      "loss": 0.3138,
+      "step": 495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997642214473461e-05,
+      "loss": 1.019,
+      "step": 496
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997602424507975e-05,
+      "loss": 0.2743,
+      "step": 497
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997562301678969e-05,
+      "loss": 1.0202,
+      "step": 498
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997521845989113e-05,
+      "loss": 86.5886,
+      "step": 499
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997481057441105e-05,
+      "loss": 81.1785,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997439936037658e-05,
+      "loss": 83.5304,
+      "step": 501
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997398481781515e-05,
+      "loss": 1.0159,
+      "step": 502
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997356694675433e-05,
+      "loss": 0.2699,
+      "step": 503
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997314574722196e-05,
+      "loss": 75.1134,
+      "step": 504
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997272121924612e-05,
+      "loss": 1.0201,
+      "step": 505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997229336285507e-05,
+      "loss": 0.2797,
+      "step": 506
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99718621780773e-05,
+      "loss": 81.6194,
+      "step": 507
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997142766494154e-05,
+      "loss": 75.9182,
+      "step": 508
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997098982347672e-05,
+      "loss": 73.4405,
+      "step": 509
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9970548653712e-05,
+      "loss": 1.005,
+      "step": 510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997010415567678e-05,
+      "loss": 71.4915,
+      "step": 511
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996965632940065e-05,
+      "loss": 0.2925,
+      "step": 512
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996920517491345e-05,
+      "loss": 0.9302,
+      "step": 513
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996875069224523e-05,
+      "loss": 0.3028,
+      "step": 514
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996829288142625e-05,
+      "loss": 1.0049,
+      "step": 515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9967831742487e-05,
+      "loss": 0.2801,
+      "step": 516
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996736727545822e-05,
+      "loss": 73.9744,
+      "step": 517
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99668994803708e-05,
+      "loss": 78.0719,
+      "step": 518
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996642835725594e-05,
+      "loss": 0.2848,
+      "step": 519
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996595390614499e-05,
+      "loss": 0.2848,
+      "step": 520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996547612706957e-05,
+      "loss": 74.5416,
+      "step": 521
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99649950200615e-05,
+      "loss": 0.2939,
+      "step": 522
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996451058515282e-05,
+      "loss": 0.2706,
+      "step": 523
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996402282237578e-05,
+      "loss": 0.9519,
+      "step": 524
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996353173176289e-05,
+      "loss": 0.9758,
+      "step": 525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996303731334686e-05,
+      "loss": 0.9856,
+      "step": 526
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996253956716061e-05,
+      "loss": 75.301,
+      "step": 527
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996203849323729e-05,
+      "loss": 0.2969,
+      "step": 528
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996153409161027e-05,
+      "loss": 68.3434,
+      "step": 529
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996102636231316e-05,
+      "loss": 0.9349,
+      "step": 530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996051530537977e-05,
+      "loss": 69.8016,
+      "step": 531
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996000092084414e-05,
+      "loss": 0.9248,
+      "step": 532
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995948320874054e-05,
+      "loss": 72.0766,
+      "step": 533
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99589621691034e-05,
+      "loss": 68.2865,
+      "step": 534
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995843780196752e-05,
+      "loss": 67.5214,
+      "step": 535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995791010736773e-05,
+      "loss": 65.7419,
+      "step": 536
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995737908533924e-05,
+      "loss": 0.2982,
+      "step": 537
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995684473591738e-05,
+      "loss": 67.1611,
+      "step": 538
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995630705913776e-05,
+      "loss": 0.3084,
+      "step": 539
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99557660550362e-05,
+      "loss": 64.8243,
+      "step": 540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995522172364872e-05,
+      "loss": 66.9518,
+      "step": 541
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995467406501158e-05,
+      "loss": 1.0126,
+      "step": 542
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995412307916123e-05,
+      "loss": 62.7886,
+      "step": 543
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99535687661344e-05,
+      "loss": 0.2669,
+      "step": 544
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995301112596801e-05,
+      "loss": 66.3049,
+      "step": 545
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995245015869918e-05,
+      "loss": 65.7317,
+      "step": 546
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995188586436531e-05,
+      "loss": 62.7328,
+      "step": 547
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995131824300394e-05,
+      "loss": 0.2776,
+      "step": 548
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995074729465289e-05,
+      "loss": 0.3127,
+      "step": 549
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995017301935021e-05,
+      "loss": 69.1995,
+      "step": 550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994959541713412e-05,
+      "loss": 66.495,
+      "step": 551
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994901448804312e-05,
+      "loss": 61.3668,
+      "step": 552
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994843023211587e-05,
+      "loss": 67.7914,
+      "step": 553
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99478426493913e-05,
+      "loss": 57.6027,
+      "step": 554
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994725173990855e-05,
+      "loss": 56.6251,
+      "step": 555
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994665750370699e-05,
+      "loss": 64.4527,
+      "step": 556
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994605994082617e-05,
+      "loss": 61.9552,
+      "step": 557
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99454590513059e-05,
+      "loss": 0.2671,
+      "step": 558
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994485483518622e-05,
+      "loss": 0.9973,
+      "step": 559
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994424729250736e-05,
+      "loss": 67.0643,
+      "step": 560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994363642330977e-05,
+      "loss": 0.2757,
+      "step": 561
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994302222763414e-05,
+      "loss": 0.2903,
+      "step": 562
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994240470552141e-05,
+      "loss": 1.0492,
+      "step": 563
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994178385701269e-05,
+      "loss": 64.0873,
+      "step": 564
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994115968214932e-05,
+      "loss": 64.4263,
+      "step": 565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99405321809729e-05,
+      "loss": 62.2643,
+      "step": 566
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993990135352518e-05,
+      "loss": 65.9128,
+      "step": 567
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993926719984823e-05,
+      "loss": 60.4061,
+      "step": 568
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993862971998425e-05,
+      "loss": 62.4924,
+      "step": 569
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993798891397571e-05,
+      "loss": 0.9612,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993734478186529e-05,
+      "loss": 0.9719,
+      "step": 571
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993669732369591e-05,
+      "loss": 0.9461,
+      "step": 572
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993604653951065e-05,
+      "loss": 65.6464,
+      "step": 573
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99353924293529e-05,
+      "loss": 66.6071,
+      "step": 574
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99347349932662e-05,
+      "loss": 62.0228,
+      "step": 575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993407423129437e-05,
+      "loss": 0.261,
+      "step": 576
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993341014348136e-05,
+      "loss": 62.6333,
+      "step": 577
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993274272987148e-05,
+      "loss": 62.2651,
+      "step": 578
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993207199050911e-05,
+      "loss": 62.5122,
+      "step": 579
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993139792543898e-05,
+      "loss": 0.9634,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993072053470593e-05,
+      "loss": 64.3238,
+      "step": 581
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993003981835514e-05,
+      "loss": 61.0733,
+      "step": 582
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99293557764319e-05,
+      "loss": 63.1102,
+      "step": 583
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992866840898179e-05,
+      "loss": 1.0038,
+      "step": 584
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99279777160506e-05,
+      "loss": 0.2872,
+      "step": 585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992728369768432e-05,
+      "loss": 0.9478,
+      "step": 586
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992658635392917e-05,
+      "loss": 72.3014,
+      "step": 587
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992588568483162e-05,
+      "loss": 64.1739,
+      "step": 588
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99251816904383e-05,
+      "loss": 64.3805,
+      "step": 589
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992447437079614e-05,
+      "loss": 1.0188,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992376372595223e-05,
+      "loss": 73.3975,
+      "step": 591
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99230497559539e-05,
+      "loss": 90.1765,
+      "step": 592
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992233246084872e-05,
+      "loss": 0.9884,
+      "step": 593
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992161184068445e-05,
+      "loss": 76.3311,
+      "step": 594
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992088789550909e-05,
+      "loss": 66.6462,
+      "step": 595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992016062537084e-05,
+      "loss": 0.9812,
+      "step": 596
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991943003031818e-05,
+      "loss": 62.3417,
+      "step": 597
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991869611039975e-05,
+      "loss": 0.978,
+      "step": 598
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991795886566441e-05,
+      "loss": 67.2238,
+      "step": 599
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99172182961613e-05,
+      "loss": 0.243,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991647440193973e-05,
+      "loss": 69.1839,
+      "step": 601
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991572718304924e-05,
+      "loss": 61.3824,
+      "step": 602
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991497663953962e-05,
+      "loss": 61.3291,
+      "step": 603
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991422277146084e-05,
+      "loss": 1.0047,
+      "step": 604
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991346557886312e-05,
+      "loss": 62.9807,
+      "step": 605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991270506179688e-05,
+      "loss": 1.0033,
+      "step": 606
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991194122031279e-05,
+      "loss": 62.9973,
+      "step": 607
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99111740544617e-05,
+      "loss": 64.6191,
+      "step": 608
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991040356429476e-05,
+      "loss": 61.0964,
+      "step": 609
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990962974986324e-05,
+      "loss": 0.2544,
+      "step": 610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990885261121868e-05,
+      "loss": 61.3976,
+      "step": 611
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990807214841287e-05,
+      "loss": 0.295,
+      "step": 612
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990728836149777e-05,
+      "loss": 63.5144,
+      "step": 613
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990650125052559e-05,
+      "loss": 59.2344,
+      "step": 614
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990571081554875e-05,
+      "loss": 62.3906,
+      "step": 615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990491705661992e-05,
+      "loss": 64.7064,
+      "step": 616
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990411997379194e-05,
+      "loss": 63.2732,
+      "step": 617
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990331956711792e-05,
+      "loss": 62.6972,
+      "step": 618
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990251583665114e-05,
+      "loss": 0.2892,
+      "step": 619
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990170878244516e-05,
+      "loss": 1.0048,
+      "step": 620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990089840455373e-05,
+      "loss": 58.443,
+      "step": 621
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990008470303083e-05,
+      "loss": 0.2199,
+      "step": 622
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989926767793063e-05,
+      "loss": 60.3842,
+      "step": 623
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989844732930757e-05,
+      "loss": 1.0028,
+      "step": 624
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989762365721629e-05,
+      "loss": 62.493,
+      "step": 625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989679666171165e-05,
+      "loss": 1.0111,
+      "step": 626
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989596634284872e-05,
+      "loss": 0.9548,
+      "step": 627
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989513270068283e-05,
+      "loss": 62.1525,
+      "step": 628
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989429573526945e-05,
+      "loss": 0.9287,
+      "step": 629
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989345544666438e-05,
+      "loss": 61.8642,
+      "step": 630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989261183492356e-05,
+      "loss": 0.9846,
+      "step": 631
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989176490010319e-05,
+      "loss": 0.2005,
+      "step": 632
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989091464225967e-05,
+      "loss": 0.9934,
+      "step": 633
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989006106144963e-05,
+      "loss": 59.2656,
+      "step": 634
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988920415772994e-05,
+      "loss": 55.8585,
+      "step": 635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988834393115767e-05,
+      "loss": 0.9598,
+      "step": 636
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988748038179009e-05,
+      "loss": 59.5097,
+      "step": 637
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988661350968473e-05,
+      "loss": 0.23,
+      "step": 638
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988574331489935e-05,
+      "loss": 55.9103,
+      "step": 639
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988486979749189e-05,
+      "loss": 60.0181,
+      "step": 640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988399295752052e-05,
+      "loss": 0.2606,
+      "step": 641
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988311279504366e-05,
+      "loss": 0.2519,
+      "step": 642
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988222931011993e-05,
+      "loss": 61.5162,
+      "step": 643
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988134250280817e-05,
+      "loss": 0.962,
+      "step": 644
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988045237316743e-05,
+      "loss": 0.2723,
+      "step": 645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987955892125702e-05,
+      "loss": 61.2125,
+      "step": 646
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987866214713643e-05,
+      "loss": 0.2254,
+      "step": 647
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987776205086542e-05,
+      "loss": 60.0206,
+      "step": 648
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987685863250391e-05,
+      "loss": 0.2639,
+      "step": 649
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987595189211208e-05,
+      "loss": 1.0155,
+      "step": 650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987504182975033e-05,
+      "loss": 0.9383,
+      "step": 651
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987412844547927e-05,
+      "loss": 63.5999,
+      "step": 652
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987321173935973e-05,
+      "loss": 0.9866,
+      "step": 653
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987229171145277e-05,
+      "loss": 60.9953,
+      "step": 654
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987136836181968e-05,
+      "loss": 0.2409,
+      "step": 655
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987044169052193e-05,
+      "loss": 55.72,
+      "step": 656
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986951169762128e-05,
+      "loss": 57.4278,
+      "step": 657
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986857838317965e-05,
+      "loss": 0.2744,
+      "step": 658
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986764174725919e-05,
+      "loss": 0.2458,
+      "step": 659
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98667017899223e-05,
+      "loss": 0.9857,
+      "step": 660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98657585112316e-05,
+      "loss": 60.6201,
+      "step": 661
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98648119112499e-05,
+      "loss": 59.656,
+      "step": 662
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986386199004025e-05,
+      "loss": 63.3504,
+      "step": 663
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98629087476659e-05,
+      "loss": 60.3867,
+      "step": 664
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986195218419037e-05,
+      "loss": 0.9608,
+      "step": 665
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986099229967736e-05,
+      "loss": 0.2691,
+      "step": 666
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986002909419081e-05,
+      "loss": 62.924,
+      "step": 667
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985906256779487e-05,
+      "loss": 58.9446,
+      "step": 668
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98580927205539e-05,
+      "loss": 61.8899,
+      "step": 669
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985711955253252e-05,
+      "loss": 52.8588,
+      "step": 670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985614306379551e-05,
+      "loss": 1.0174,
+      "step": 671
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985516325440796e-05,
+      "loss": 52.4062,
+      "step": 672
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98541801244351e-05,
+      "loss": 0.967,
+      "step": 673
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985319367394241e-05,
+      "loss": 1.0159,
+      "step": 674
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985220390299561e-05,
+      "loss": 59.9132,
+      "step": 675
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98512108116606e-05,
+      "loss": 63.1111,
+      "step": 676
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985021440000354e-05,
+      "loss": 51.5684,
+      "step": 677
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98492146680908e-05,
+      "loss": 58.0687,
+      "step": 678
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984821161598894e-05,
+      "loss": 0.9654,
+      "step": 679
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984720524376481e-05,
+      "loss": 57.6385,
+      "step": 680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98461955514854e-05,
+      "loss": 57.8896,
+      "step": 681
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984518253921797e-05,
+      "loss": 55.2421,
+      "step": 682
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984416620703e-05,
+      "loss": 59.7844,
+      "step": 683
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984314655498919e-05,
+      "loss": 0.9568,
+      "step": 684
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984212358316342e-05,
+      "loss": 1.0145,
+      "step": 685
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984109729162088e-05,
+      "loss": 0.989,
+      "step": 686
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984006768042987e-05,
+      "loss": 59.845,
+      "step": 687
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983903474965899e-05,
+      "loss": 0.2516,
+      "step": 688
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983799849937705e-05,
+      "loss": 57.1969,
+      "step": 689
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983695892965306e-05,
+      "loss": 0.968,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983591604055624e-05,
+      "loss": 0.2696,
+      "step": 691
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983486983215607e-05,
+      "loss": 58.1152,
+      "step": 692
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983382030452224e-05,
+      "loss": 0.9279,
+      "step": 693
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983276745772465e-05,
+      "loss": 58.2799,
+      "step": 694
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98317112918334e-05,
+      "loss": 55.8836,
+      "step": 695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983065180691886e-05,
+      "loss": 0.2373,
+      "step": 696
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982958900305161e-05,
+      "loss": 55.8487,
+      "step": 697
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98285228803024e-05,
+      "loss": 0.9513,
+      "step": 698
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982745343874227e-05,
+      "loss": 53.4159,
+      "step": 699
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982638067844243e-05,
+      "loss": 0.9967,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982530459947432e-05,
+      "loss": 0.3034,
+      "step": 701
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982422520190966e-05,
+      "loss": 0.219,
+      "step": 702
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98231424858203e-05,
+      "loss": 51.515,
+      "step": 703
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982205645127835e-05,
+      "loss": 0.9855,
+      "step": 704
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982096709835618e-05,
+      "loss": 0.2962,
+      "step": 705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981987442712633e-05,
+      "loss": 1.0022,
+      "step": 706
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981877843766157e-05,
+      "loss": 56.4238,
+      "step": 707
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98176791300349e-05,
+      "loss": 1.0203,
+      "step": 708
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981657650431955e-05,
+      "loss": 56.2517,
+      "step": 709
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981547056058893e-05,
+      "loss": 0.9527,
+      "step": 710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981436129891675e-05,
+      "loss": 0.2686,
+      "step": 711
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981324871937686e-05,
+      "loss": 59.4706,
+      "step": 712
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981213282204335e-05,
+      "loss": 0.228,
+      "step": 713
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981101360699057e-05,
+      "loss": 1.0072,
+      "step": 714
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980989107429305e-05,
+      "loss": 0.2284,
+      "step": 715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980876522402557e-05,
+      "loss": 52.6122,
+      "step": 716
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980763605626311e-05,
+      "loss": 0.2427,
+      "step": 717
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980650357108088e-05,
+      "loss": 54.7157,
+      "step": 718
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98053677685543e-05,
+      "loss": 56.5811,
+      "step": 719
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9804228648759e-05,
+      "loss": 0.2772,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980308621177091e-05,
+      "loss": 55.5849,
+      "step": 721
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980194045766609e-05,
+      "loss": 0.9741,
+      "step": 722
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980079138652084e-05,
+      "loss": 54.635,
+      "step": 723
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979963899841169e-05,
+      "loss": 0.9565,
+      "step": 724
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979848329341543e-05,
+      "loss": 52.8728,
+      "step": 725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979732427160899e-05,
+      "loss": 1.0134,
+      "step": 726
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97961619330696e-05,
+      "loss": 0.2599,
+      "step": 727
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979499627787468e-05,
+      "loss": 0.2791,
+      "step": 728
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979382730610184e-05,
+      "loss": 0.3036,
+      "step": 729
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979265501782895e-05,
+      "loss": 53.7531,
+      "step": 730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979147941313412e-05,
+      "loss": 53.5505,
+      "step": 731
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97903004920956e-05,
+      "loss": 0.9667,
+      "step": 732
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978911825479193e-05,
+      "loss": 0.9401,
+      "step": 733
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978793270130188e-05,
+      "loss": 57.1942,
+      "step": 734
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978674383170437e-05,
+      "loss": 56.6465,
+      "step": 735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978555164607862e-05,
+      "loss": 0.9684,
+      "step": 736
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978435614450402e-05,
+      "loss": 57.527,
+      "step": 737
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97831573270602e-05,
+      "loss": 0.2779,
+      "step": 738
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978195519382699e-05,
+      "loss": 0.9818,
+      "step": 739
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978074974488449e-05,
+      "loss": 0.9645,
+      "step": 740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977954098031295e-05,
+      "loss": 57.28,
+      "step": 741
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977832890019292e-05,
+      "loss": 50.2221,
+      "step": 742
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977711350460509e-05,
+      "loss": 0.2656,
+      "step": 743
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977589479363044e-05,
+      "loss": 59.3013,
+      "step": 744
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977467276735013e-05,
+      "loss": 60.0254,
+      "step": 745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977344742584555e-05,
+      "loss": 57.2797,
+      "step": 746
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977221876919833e-05,
+      "loss": 50.3694,
+      "step": 747
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977098679749028e-05,
+      "loss": 58.4564,
+      "step": 748
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976975151080348e-05,
+      "loss": 0.2673,
+      "step": 749
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97685129092202e-05,
+      "loss": 0.975,
+      "step": 750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97672709928229e-05,
+      "loss": 0.9617,
+      "step": 751
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976602576169437e-05,
+      "loss": 0.2797,
+      "step": 752
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976477721591745e-05,
+      "loss": 56.1347,
+      "step": 753
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976352535557541e-05,
+      "loss": 55.0225,
+      "step": 754
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976227018075154e-05,
+      "loss": 0.9822,
+      "step": 755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976101169152948e-05,
+      "loss": 56.245,
+      "step": 756
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975974988799306e-05,
+      "loss": 0.2252,
+      "step": 757
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97584847702263e-05,
+      "loss": 1.0088,
+      "step": 758
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975721633831345e-05,
+      "loss": 56.7539,
+      "step": 759
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975594459233904e-05,
+      "loss": 54.9889,
+      "step": 760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975466953238775e-05,
+      "loss": 54.3288,
+      "step": 761
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975339115854448e-05,
+      "loss": 53.3199,
+      "step": 762
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975210947089442e-05,
+      "loss": 0.2549,
+      "step": 763
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975082446952291e-05,
+      "loss": 0.9926,
+      "step": 764
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974953615451553e-05,
+      "loss": 55.955,
+      "step": 765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974824452595811e-05,
+      "loss": 0.9554,
+      "step": 766
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974694958393666e-05,
+      "loss": 52.1878,
+      "step": 767
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974565132853747e-05,
+      "loss": 0.2702,
+      "step": 768
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974434975984694e-05,
+      "loss": 0.9855,
+      "step": 769
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97430448779518e-05,
+      "loss": 0.2557,
+      "step": 770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974173668293898e-05,
+      "loss": 0.9399,
+      "step": 771
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974042517489558e-05,
+      "loss": 60.2154,
+      "step": 772
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973911035390897e-05,
+      "loss": 0.2578,
+      "step": 773
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973779222006671e-05,
+      "loss": 56.7649,
+      "step": 774
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97364707734566e-05,
+      "loss": 53.3764,
+      "step": 775
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973514601416665e-05,
+      "loss": 54.3095,
+      "step": 776
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973381794228511e-05,
+      "loss": 51.0485,
+      "step": 777
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973248655790042e-05,
+      "loss": 0.9766,
+      "step": 778
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973115186110127e-05,
+      "loss": 58.2661,
+      "step": 779
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972981385197654e-05,
+      "loss": 0.2164,
+      "step": 780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972847253061536e-05,
+      "loss": 0.9551,
+      "step": 781
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972712789710708e-05,
+      "loss": 52.6205,
+      "step": 782
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972577995154122e-05,
+      "loss": 53.6391,
+      "step": 783
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972442869400759e-05,
+      "loss": 0.9389,
+      "step": 784
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972307412459618e-05,
+      "loss": 48.8824,
+      "step": 785
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972171624339724e-05,
+      "loss": 53.5929,
+      "step": 786
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972035505050115e-05,
+      "loss": 50.1047,
+      "step": 787
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971899054599863e-05,
+      "loss": 0.2357,
+      "step": 788
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971762272998053e-05,
+      "loss": 0.9844,
+      "step": 789
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971625160253796e-05,
+      "loss": 56.6615,
+      "step": 790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971487716376224e-05,
+      "loss": 57.3819,
+      "step": 791
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971349941374494e-05,
+      "loss": 51.7561,
+      "step": 792
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971211835257779e-05,
+      "loss": 1.0468,
+      "step": 793
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97107339803528e-05,
+      "loss": 50.7734,
+      "step": 794
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970934629716216e-05,
+      "loss": 53.0359,
+      "step": 795
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970795530309829e-05,
+      "loss": 53.1843,
+      "step": 796
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970656099825386e-05,
+      "loss": 0.9875,
+      "step": 797
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970516338272171e-05,
+      "loss": 49.7602,
+      "step": 798
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970376245659496e-05,
+      "loss": 52.2703,
+      "step": 799
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97023582199669e-05,
+      "loss": 48.3629,
+      "step": 800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970095067293104e-05,
+      "loss": 52.482,
+      "step": 801
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969953981558116e-05,
+      "loss": 52.1638,
+      "step": 802
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969812564801122e-05,
+      "loss": 52.7011,
+      "step": 803
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969670817031541e-05,
+      "loss": 0.2562,
+      "step": 804
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969528738258813e-05,
+      "loss": 52.5658,
+      "step": 805
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969386328492404e-05,
+      "loss": 48.4901,
+      "step": 806
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969243587741796e-05,
+      "loss": 0.9846,
+      "step": 807
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969100516016496e-05,
+      "loss": 52.1179,
+      "step": 808
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968957113326036e-05,
+      "loss": 45.6252,
+      "step": 809
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968813379679966e-05,
+      "loss": 0.2064,
+      "step": 810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96866931508786e-05,
+      "loss": 50.9742,
+      "step": 811
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96852491955931e-05,
+      "loss": 44.2102,
+      "step": 812
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968380193103938e-05,
+      "loss": 48.5798,
+      "step": 813
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968235135731381e-05,
+      "loss": 53.8097,
+      "step": 814
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968089747451301e-05,
+      "loss": 0.9704,
+      "step": 815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96794402827338e-05,
+      "loss": 52.0576,
+      "step": 816
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967797978207328e-05,
+      "loss": 0.9334,
+      "step": 817
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967651597262868e-05,
+      "loss": 46.5265,
+      "step": 818
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96750488544975e-05,
+      "loss": 50.1412,
+      "step": 819
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967357842777748e-05,
+      "loss": 1.0051,
+      "step": 820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967210469256656e-05,
+      "loss": 51.5663,
+      "step": 821
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967062764896287e-05,
+      "loss": 46.1717,
+      "step": 822
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96691472970648e-05,
+      "loss": 0.2429,
+      "step": 823
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966766363697096e-05,
+      "loss": 1.0348,
+      "step": 824
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966617666878015e-05,
+      "loss": 53.124,
+      "step": 825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966468639259143e-05,
+      "loss": 52.6015,
+      "step": 826
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966319280850405e-05,
+      "loss": 1.0202,
+      "step": 827
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966169591661748e-05,
+      "loss": 0.9511,
+      "step": 828
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966019571703143e-05,
+      "loss": 48.4102,
+      "step": 829
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965869220984581e-05,
+      "loss": 0.2676,
+      "step": 830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965718539516078e-05,
+      "loss": 47.6623,
+      "step": 831
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96556752730767e-05,
+      "loss": 48.8877,
+      "step": 832
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965416184369412e-05,
+      "loss": 49.8145,
+      "step": 833
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965264510711385e-05,
+      "loss": 0.2492,
+      "step": 834
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965112506343694e-05,
+      "loss": 1.0281,
+      "step": 835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964960171276462e-05,
+      "loss": 50.3797,
+      "step": 836
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964807505519835e-05,
+      "loss": 48.4168,
+      "step": 837
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964654509083979e-05,
+      "loss": 49.7582,
+      "step": 838
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964501181979089e-05,
+      "loss": 47.8568,
+      "step": 839
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964347524215373e-05,
+      "loss": 48.2526,
+      "step": 840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964193535803067e-05,
+      "loss": 49.0474,
+      "step": 841
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964039216752427e-05,
+      "loss": 45.1445,
+      "step": 842
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963884567073732e-05,
+      "loss": 48.9057,
+      "step": 843
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963729586777282e-05,
+      "loss": 47.4408,
+      "step": 844
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9635742758734e-05,
+      "loss": 0.2506,
+      "step": 845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963418634372428e-05,
+      "loss": 0.9805,
+      "step": 846
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963262662284736e-05,
+      "loss": 48.2303,
+      "step": 847
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963106359620711e-05,
+      "loss": 49.5093,
+      "step": 848
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962949726390763e-05,
+      "loss": 48.93,
+      "step": 849
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962792762605323e-05,
+      "loss": 46.3411,
+      "step": 850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962635468274847e-05,
+      "loss": 1.0141,
+      "step": 851
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962477843409814e-05,
+      "loss": 1.0209,
+      "step": 852
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96231988802072e-05,
+      "loss": 0.9558,
+      "step": 853
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962161602118086e-05,
+      "loss": 1.0115,
+      "step": 854
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962002985712453e-05,
+      "loss": 0.2181,
+      "step": 855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961844038814387e-05,
+      "loss": 55.9986,
+      "step": 856
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961684761434476e-05,
+      "loss": 49.4712,
+      "step": 857
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961525153583327e-05,
+      "loss": 50.5011,
+      "step": 858
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961365215271572e-05,
+      "loss": 0.9389,
+      "step": 859
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961204946509861e-05,
+      "loss": 56.1149,
+      "step": 860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96104434730887e-05,
+      "loss": 52.3832,
+      "step": 861
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960883417679296e-05,
+      "loss": 46.094,
+      "step": 862
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960722157631858e-05,
+      "loss": 1.0066,
+      "step": 863
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960560567177295e-05,
+      "loss": 47.7516,
+      "step": 864
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960398646326372e-05,
+      "loss": 0.942,
+      "step": 865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960236395089873e-05,
+      "loss": 49.3104,
+      "step": 866
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960073813478604e-05,
+      "loss": 0.2754,
+      "step": 867
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959910901503393e-05,
+      "loss": 50.6017,
+      "step": 868
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959747659175092e-05,
+      "loss": 0.9776,
+      "step": 869
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959584086504572e-05,
+      "loss": 45.0632,
+      "step": 870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95942018350273e-05,
+      "loss": 1.0091,
+      "step": 871
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959255950180482e-05,
+      "loss": 45.1806,
+      "step": 872
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959091386548767e-05,
+      "loss": 45.9731,
+      "step": 873
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958926492618544e-05,
+      "loss": 0.283,
+      "step": 874
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958761268400796e-05,
+      "loss": 46.6923,
+      "step": 875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958595713906529e-05,
+      "loss": 0.2292,
+      "step": 876
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95842982914677e-05,
+      "loss": 1.0117,
+      "step": 877
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958263614132567e-05,
+      "loss": 0.2621,
+      "step": 878
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95809706887499e-05,
+      "loss": 0.2319,
+      "step": 879
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957930193385132e-05,
+      "loss": 0.2575,
+      "step": 880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957762987674108e-05,
+      "loss": 0.2542,
+      "step": 881
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957595451753055e-05,
+      "loss": 0.9021,
+      "step": 882
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957427585633131e-05,
+      "loss": 0.2413,
+      "step": 883
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957259389325517e-05,
+      "loss": 0.9699,
+      "step": 884
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957090862841416e-05,
+      "loss": 52.1502,
+      "step": 885
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95692200619205e-05,
+      "loss": 48.2267,
+      "step": 886
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95675281938867e-05,
+      "loss": 50.3511,
+      "step": 887
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956583302442544e-05,
+      "loss": 44.0241,
+      "step": 888
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956413455364959e-05,
+      "loss": 0.949,
+      "step": 889
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95624327816723e-05,
+      "loss": 44.7475,
+      "step": 890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956072770860692e-05,
+      "loss": 46.572,
+      "step": 891
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955901933456699e-05,
+      "loss": 0.9489,
+      "step": 892
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955730765966633e-05,
+      "loss": 0.2387,
+      "step": 893
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955559268401893e-05,
+      "loss": 54.8626,
+      "step": 894
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9553874407739e-05,
+      "loss": 0.222,
+      "step": 895
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955215283094102e-05,
+      "loss": 0.2561,
+      "step": 896
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955042795373963e-05,
+      "loss": 49.3994,
+      "step": 897
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954869977624973e-05,
+      "loss": 52.4479,
+      "step": 898
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95469682985864e-05,
+      "loss": 0.9827,
+      "step": 899
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954523352086498e-05,
+      "loss": 46.9813,
+      "step": 900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954349544320101e-05,
+      "loss": 0.9657,
+      "step": 901
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954175406571027e-05,
+      "loss": 1.0133,
+      "step": 902
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954000938850873e-05,
+      "loss": 0.9689,
+      "step": 903
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953826141171258e-05,
+      "loss": 44.3747,
+      "step": 904
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953651013543828e-05,
+      "loss": 45.7444,
+      "step": 905
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953475555980244e-05,
+      "loss": 0.2231,
+      "step": 906
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953299768492194e-05,
+      "loss": 49.1966,
+      "step": 907
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953123651091385e-05,
+      "loss": 48.3032,
+      "step": 908
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952947203789548e-05,
+      "loss": 0.9956,
+      "step": 909
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952770426598435e-05,
+      "loss": 49.875,
+      "step": 910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952593319529821e-05,
+      "loss": 47.0181,
+      "step": 911
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952415882595501e-05,
+      "loss": 0.231,
+      "step": 912
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952238115807296e-05,
+      "loss": 0.2543,
+      "step": 913
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952060019177041e-05,
+      "loss": 0.9835,
+      "step": 914
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9518815927166e-05,
+      "loss": 49.3486,
+      "step": 915
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95170283643786e-05,
+      "loss": 0.2652,
+      "step": 916
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951523750352725e-05,
+      "loss": 47.893,
+      "step": 917
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95134433447312e-05,
+      "loss": 0.9749,
+      "step": 918
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951164588811001e-05,
+      "loss": 0.2464,
+      "step": 919
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950984513378334e-05,
+      "loss": 0.9696,
+      "step": 920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950804108187118e-05,
+      "loss": 47.8699,
+      "step": 921
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950623373249364e-05,
+      "loss": 0.9499,
+      "step": 922
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950442308577113e-05,
+      "loss": 0.9916,
+      "step": 923
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950260914182423e-05,
+      "loss": 42.2458,
+      "step": 924
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950079190077378e-05,
+      "loss": 46.4848,
+      "step": 925
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.94989713627408e-05,
+      "loss": 0.9706,
+      "step": 926
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949714752784653e-05,
+      "loss": 44.7281,
+      "step": 927
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949532039621249e-05,
+      "loss": 0.9843,
+      "step": 928
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949348996796033e-05,
+      "loss": 44.4407,
+      "step": 929
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9491656243212e-05,
+      "loss": 0.2412,
+      "step": 930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.94898192220896e-05,
+      "loss": 45.1936,
+      "step": 931
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948797890471551e-05,
+      "loss": 0.1999,
+      "step": 932
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948613529121232e-05,
+      "loss": 0.2565,
+      "step": 933
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948428838170277e-05,
+      "loss": 39.9801,
+      "step": 934
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948243817630991e-05,
+      "loss": 0.9699,
+      "step": 935
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948058467515698e-05,
+      "loss": 43.2089,
+      "step": 936
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.94787278783674e-05,
+      "loss": 0.2376,
+      "step": 937
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947686778606487e-05,
+      "loss": 46.1772,
+      "step": 938
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947500439837325e-05,
+      "loss": 0.2398,
+      "step": 939
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947313771541668e-05,
+      "loss": 44.9082,
+      "step": 940
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947126773731948e-05,
+      "loss": 46.4001,
+      "step": 941
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94693944642062e-05,
+      "loss": 1.0144,
+      "step": 942
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94675178962016e-05,
+      "loss": 0.2585,
+      "step": 943
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946563803343069e-05,
+      "loss": 45.2341,
+      "step": 944
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946375487601866e-05,
+      "loss": 0.9461,
+      "step": 945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946186842409093e-05,
+      "loss": 44.3099,
+      "step": 946
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945997867777317e-05,
+      "loss": 42.9972,
+      "step": 947
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945808563719121e-05,
+      "loss": 45.0994,
+      "step": 948
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945618930247118e-05,
+      "loss": 0.2446,
+      "step": 949
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945428967373934e-05,
+      "loss": 41.8009,
+      "step": 950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945238675112225e-05,
+      "loss": 46.6565,
+      "step": 951
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945048053474664e-05,
+      "loss": 41.8784,
+      "step": 952
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944857102473947e-05,
+      "loss": 0.9877,
+      "step": 953
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944665822122791e-05,
+      "loss": 44.8941,
+      "step": 954
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94447421243394e-05,
+      "loss": 0.2699,
+      "step": 955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944282273420154e-05,
+      "loss": 0.9935,
+      "step": 956
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944090005094215e-05,
+      "loss": 42.5263,
+      "step": 957
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943897407468932e-05,
+      "loss": 42.7949,
+      "step": 958
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94370448055713e-05,
+      "loss": 44.9576,
+      "step": 959
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943511224371661e-05,
+      "loss": 39.296,
+      "step": 960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943317638925397e-05,
+      "loss": 1.0081,
+      "step": 961
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943123724231229e-05,
+      "loss": 44.3067,
+      "step": 962
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942929480302077e-05,
+      "loss": 41.329,
+      "step": 963
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942734907150874e-05,
+      "loss": 40.2668,
+      "step": 964
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942540004790583e-05,
+      "loss": 46.1401,
+      "step": 965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942344773234183e-05,
+      "loss": 42.8369,
+      "step": 966
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942149212494679e-05,
+      "loss": 42.4101,
+      "step": 967
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941953322585096e-05,
+      "loss": 1.0163,
+      "step": 968
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941757103518478e-05,
+      "loss": 43.4805,
+      "step": 969
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9415605553079e-05,
+      "loss": 41.201,
+      "step": 970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941363677966448e-05,
+      "loss": 0.9473,
+      "step": 971
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941166471507237e-05,
+      "loss": 40.0193,
+      "step": 972
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940968935943402e-05,
+      "loss": 41.0844,
+      "step": 973
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9407710712881e-05,
+      "loss": 0.9644,
+      "step": 974
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940572877554508e-05,
+      "loss": 0.923,
+      "step": 975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940374354755828e-05,
+      "loss": 0.2618,
+      "step": 976
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940175502905282e-05,
+      "loss": 41.7853,
+      "step": 977
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939976322016116e-05,
+      "loss": 40.3158,
+      "step": 978
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939776812101593e-05,
+      "loss": 1.0243,
+      "step": 979
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939576973175005e-05,
+      "loss": 0.2246,
+      "step": 980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93937680524966e-05,
+      "loss": 46.7679,
+      "step": 981
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939176308338891e-05,
+      "loss": 0.2441,
+      "step": 982
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938975482456052e-05,
+      "loss": 0.2382,
+      "step": 983
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938774327614518e-05,
+      "loss": 40.2712,
+      "step": 984
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938572843827689e-05,
+      "loss": 0.9105,
+      "step": 985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938371031108982e-05,
+      "loss": 0.9598,
+      "step": 986
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93816888947184e-05,
+      "loss": 44.2942,
+      "step": 987
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937966418929726e-05,
+      "loss": 39.187,
+      "step": 988
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937763619496125e-05,
+      "loss": 0.9967,
+      "step": 989
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937560491184547e-05,
+      "loss": 41.5463,
+      "step": 990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93735703400852e-05,
+      "loss": 42.4376,
+      "step": 991
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937153247981594e-05,
+      "loss": 42.3424,
+      "step": 992
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936949133117344e-05,
+      "loss": 0.9727,
+      "step": 993
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936744689429362e-05,
+      "loss": 39.2011,
+      "step": 994
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936539916931268e-05,
+      "loss": 39.9348,
+      "step": 995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936334815636701e-05,
+      "loss": 0.9663,
+      "step": 996
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936129385559318e-05,
+      "loss": 0.2456,
+      "step": 997
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935923626712806e-05,
+      "loss": 0.2552,
+      "step": 998
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935717539110866e-05,
+      "loss": 0.2468,
+      "step": 999
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935511122767227e-05,
+      "loss": 44.7939,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935304377695634e-05,
+      "loss": 44.7982,
+      "step": 1001
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935097303909861e-05,
+      "loss": 0.9551,
+      "step": 1002
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934889901423698e-05,
+      "loss": 0.2638,
+      "step": 1003
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934682170250959e-05,
+      "loss": 0.2747,
+      "step": 1004
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934474110405481e-05,
+      "loss": 0.9225,
+      "step": 1005
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93426572190112e-05,
+      "loss": 0.2467,
+      "step": 1006
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934057004751758e-05,
+      "loss": 43.8349,
+      "step": 1007
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933847958971293e-05,
+      "loss": 45.3195,
+      "step": 1008
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933638584573652e-05,
+      "loss": 44.127,
+      "step": 1009
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933428881572778e-05,
+      "loss": 0.996,
+      "step": 1010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93321884998264e-05,
+      "loss": 0.1862,
+      "step": 1011
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933008489817226e-05,
+      "loss": 44.3946,
+      "step": 1012
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932797801090545e-05,
+      "loss": 45.9503,
+      "step": 1013
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932586783816634e-05,
+      "loss": 40.7693,
+      "step": 1014
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932375438009546e-05,
+      "loss": 41.4502,
+      "step": 1015
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932163763683356e-05,
+      "loss": 43.8845,
+      "step": 1016
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931951760852165e-05,
+      "loss": 0.9923,
+      "step": 1017
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931739429530092e-05,
+      "loss": 0.9468,
+      "step": 1018
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931526769731277e-05,
+      "loss": 0.2588,
+      "step": 1019
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931313781469889e-05,
+      "loss": 0.2442,
+      "step": 1020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93110046476011e-05,
+      "loss": 45.1013,
+      "step": 1021
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930886819616151e-05,
+      "loss": 0.9255,
+      "step": 1022
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930672846052239e-05,
+      "loss": 1.0336,
+      "step": 1023
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930458544082627e-05,
+      "loss": 43.8214,
+      "step": 1024
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930243913721588e-05,
+      "loss": 44.8325,
+      "step": 1025
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93002895498342e-05,
+      "loss": 0.9513,
+      "step": 1026
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929813667882436e-05,
+      "loss": 45.2684,
+      "step": 1027
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929598052432976e-05,
+      "loss": 0.2728,
+      "step": 1028
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929382108649403e-05,
+      "loss": 0.2248,
+      "step": 1029
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929165836546099e-05,
+      "loss": 0.1914,
+      "step": 1030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928949236137468e-05,
+      "loss": 46.248,
+      "step": 1031
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928732307437938e-05,
+      "loss": 42.8375,
+      "step": 1032
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928515050461957e-05,
+      "loss": 46.3739,
+      "step": 1033
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928297465223994e-05,
+      "loss": 41.6847,
+      "step": 1034
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928079551738543e-05,
+      "loss": 0.9687,
+      "step": 1035
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927861310020116e-05,
+      "loss": 0.9915,
+      "step": 1036
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927642740083252e-05,
+      "loss": 49.1873,
+      "step": 1037
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927423841942506e-05,
+      "loss": 0.9886,
+      "step": 1038
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927204615612457e-05,
+      "loss": 0.253,
+      "step": 1039
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926985061107711e-05,
+      "loss": 0.9339,
+      "step": 1040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926765178442888e-05,
+      "loss": 47.9861,
+      "step": 1041
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926544967632632e-05,
+      "loss": 44.5299,
+      "step": 1042
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926324428691611e-05,
+      "loss": 0.2064,
+      "step": 1043
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926103561634517e-05,
+      "loss": 0.2263,
+      "step": 1044
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925882366476056e-05,
+      "loss": 43.2069,
+      "step": 1045
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925660843230963e-05,
+      "loss": 0.2148,
+      "step": 1046
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925438991913994e-05,
+      "loss": 43.682,
+      "step": 1047
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925216812539924e-05,
+      "loss": 0.2461,
+      "step": 1048
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924994305123549e-05,
+      "loss": 0.2072,
+      "step": 1049
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924771469679694e-05,
+      "loss": 40.8886,
+      "step": 1050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924548306223196e-05,
+      "loss": 39.4508,
+      "step": 1051
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924324814768919e-05,
+      "loss": 0.9398,
+      "step": 1052
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924100995331753e-05,
+      "loss": 1.0347,
+      "step": 1053
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923876847926601e-05,
+      "loss": 41.6177,
+      "step": 1054
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923652372568395e-05,
+      "loss": 39.9077,
+      "step": 1055
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923427569272083e-05,
+      "loss": 0.9996,
+      "step": 1056
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923202438052641e-05,
+      "loss": 0.2552,
+      "step": 1057
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922976978925062e-05,
+      "loss": 40.1189,
+      "step": 1058
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922751191904364e-05,
+      "loss": 45.5104,
+      "step": 1059
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922525077005584e-05,
+      "loss": 42.0512,
+      "step": 1060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922298634243785e-05,
+      "loss": 36.9163,
+      "step": 1061
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922071863634046e-05,
+      "loss": 38.8977,
+      "step": 1062
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921844765191472e-05,
+      "loss": 0.9808,
+      "step": 1063
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921617338931188e-05,
+      "loss": 43.1161,
+      "step": 1064
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921389584868343e-05,
+      "loss": 0.2295,
+      "step": 1065
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921161503018108e-05,
+      "loss": 39.968,
+      "step": 1066
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.92093309339567e-05,
+      "loss": 42.1593,
+      "step": 1067
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920704356016246e-05,
+      "loss": 0.2338,
+      "step": 1068
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920475290895069e-05,
+      "loss": 41.2075,
+      "step": 1069
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920245898047395e-05,
+      "loss": 0.9632,
+      "step": 1070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920016177488506e-05,
+      "loss": 38.6055,
+      "step": 1071
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919786129233699e-05,
+      "loss": 42.3736,
+      "step": 1072
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919555753298299e-05,
+      "loss": 0.9574,
+      "step": 1073
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919325049697647e-05,
+      "loss": 43.2483,
+      "step": 1074
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919094018447112e-05,
+      "loss": 42.3972,
+      "step": 1075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918862659562081e-05,
+      "loss": 37.9486,
+      "step": 1076
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918630973057961e-05,
+      "loss": 0.9547,
+      "step": 1077
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918398958950188e-05,
+      "loss": 42.1074,
+      "step": 1078
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918166617254211e-05,
+      "loss": 1.0292,
+      "step": 1079
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917933947985507e-05,
+      "loss": 39.2308,
+      "step": 1080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917700951159572e-05,
+      "loss": 38.4809,
+      "step": 1081
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917467626791925e-05,
+      "loss": 0.259,
+      "step": 1082
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917233974898108e-05,
+      "loss": 0.9743,
+      "step": 1083
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91699999549368e-05,
+      "loss": 39.166,
+      "step": 1084
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91676568859423e-05,
+      "loss": 0.2334,
+      "step": 1085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91653105421536e-05,
+      "loss": 38.6978,
+      "step": 1086
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916296092372699e-05,
+      "loss": 0.9708,
+      "step": 1087
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916060803081895e-05,
+      "loss": 0.2114,
+      "step": 1088
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915825186358621e-05,
+      "loss": 0.2456,
+      "step": 1089
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91558924221857e-05,
+      "loss": 0.9748,
+      "step": 1090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915352970677457e-05,
+      "loss": 0.193,
+      "step": 1091
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915116371751018e-05,
+      "loss": 1.0043,
+      "step": 1092
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914879445455014e-05,
+      "loss": 0.9751,
+      "step": 1093
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914642191805223e-05,
+      "loss": 39.2019,
+      "step": 1094
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914404610817447e-05,
+      "loss": 0.9881,
+      "step": 1095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91416670250751e-05,
+      "loss": 36.5987,
+      "step": 1096
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91392846689126e-05,
+      "loss": 38.2961,
+      "step": 1097
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913689903984562e-05,
+      "loss": 0.9627,
+      "step": 1098
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913451013803307e-05,
+      "loss": 40.7441,
+      "step": 1099
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913211796363406e-05,
+      "loss": 39.8736,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912972251680792e-05,
+      "loss": 41.4032,
+      "step": 1101
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91273237977142e-05,
+      "loss": 38.2687,
+      "step": 1102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912492180651265e-05,
+      "loss": 0.2064,
+      "step": 1103
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912251654336326e-05,
+      "loss": 0.9553,
+      "step": 1104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912010800842626e-05,
+      "loss": 41.5291,
+      "step": 1105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911769620186204e-05,
+      "loss": 0.9961,
+      "step": 1106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911528112383123e-05,
+      "loss": 41.1861,
+      "step": 1107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911286277449469e-05,
+      "loss": 0.2228,
+      "step": 1108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911044115401354e-05,
+      "loss": 40.4576,
+      "step": 1109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910801626254899e-05,
+      "loss": 40.3135,
+      "step": 1110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910558810026263e-05,
+      "loss": 42.4333,
+      "step": 1111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910315666731613e-05,
+      "loss": 0.9845,
+      "step": 1112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910072196387145e-05,
+      "loss": 39.191,
+      "step": 1113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909828399009078e-05,
+      "loss": 0.2569,
+      "step": 1114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909584274613645e-05,
+      "loss": 40.4635,
+      "step": 1115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909339823217111e-05,
+      "loss": 37.0656,
+      "step": 1116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909095044835754e-05,
+      "loss": 1.0054,
+      "step": 1117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908849939485878e-05,
+      "loss": 37.2406,
+      "step": 1118
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90860450718381e-05,
+      "loss": 0.222,
+      "step": 1119
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908358747945893e-05,
+      "loss": 0.9806,
+      "step": 1120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908112661788501e-05,
+      "loss": 39.9162,
+      "step": 1121
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907866248728022e-05,
+      "loss": 42.1686,
+      "step": 1122
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907619508780866e-05,
+      "loss": 0.2711,
+      "step": 1123
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907372441963469e-05,
+      "loss": 0.9565,
+      "step": 1124
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907125048292288e-05,
+      "loss": 0.226,
+      "step": 1125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.9068773277838e-05,
+      "loss": 49.5394,
+      "step": 1126
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906629280454503e-05,
+      "loss": 50.665,
+      "step": 1127
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906380906320919e-05,
+      "loss": 45.3829,
+      "step": 1128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90613220539959e-05,
+      "loss": 0.2174,
+      "step": 1129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905883177707083e-05,
+      "loss": 0.2395,
+      "step": 1130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905633823259981e-05,
+      "loss": 40.4784,
+      "step": 1131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905384142074896e-05,
+      "loss": 0.9866,
+      "step": 1132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905134134168454e-05,
+      "loss": 40.3078,
+      "step": 1133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90488379955731e-05,
+      "loss": 0.2469,
+      "step": 1134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904633138258134e-05,
+      "loss": 38.058,
+      "step": 1135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904382150287624e-05,
+      "loss": 39.4252,
+      "step": 1136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904130835662497e-05,
+      "loss": 38.256,
+      "step": 1137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90387919439949e-05,
+      "loss": 1.0154,
+      "step": 1138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903627226515365e-05,
+      "loss": 0.2568,
+      "step": 1139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903374932026903e-05,
+      "loss": 39.256,
+      "step": 1140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903122310950909e-05,
+      "loss": 42.4229,
+      "step": 1141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902869363304209e-05,
+      "loss": 41.3086,
+      "step": 1142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902616089103649e-05,
+      "loss": 38.4074,
+      "step": 1143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902362488366099e-05,
+      "loss": 36.6132,
+      "step": 1144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902108561108449e-05,
+      "loss": 0.2633,
+      "step": 1145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901854307347616e-05,
+      "loss": 0.9671,
+      "step": 1146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901599727100528e-05,
+      "loss": 1.0022,
+      "step": 1147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901344820384147e-05,
+      "loss": 38.9359,
+      "step": 1148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901089587215446e-05,
+      "loss": 39.7603,
+      "step": 1149
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900834027611429e-05,
+      "loss": 39.6973,
+      "step": 1150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900578141589115e-05,
+      "loss": 35.4934,
+      "step": 1151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900321929165549e-05,
+      "loss": 0.9996,
+      "step": 1152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900065390357791e-05,
+      "loss": 0.2298,
+      "step": 1153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899808525182935e-05,
+      "loss": 0.2667,
+      "step": 1154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899551333658084e-05,
+      "loss": 0.9813,
+      "step": 1155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89929381580037e-05,
+      "loss": 38.4522,
+      "step": 1156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899035971626947e-05,
+      "loss": 0.2169,
+      "step": 1157
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898777801154984e-05,
+      "loss": 0.9636,
+      "step": 1158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898519304401681e-05,
+      "loss": 0.2236,
+      "step": 1159
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898260481384252e-05,
+      "loss": 37.4417,
+      "step": 1160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898001332119936e-05,
+      "loss": 35.0797,
+      "step": 1161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897741856625995e-05,
+      "loss": 39.9977,
+      "step": 1162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89748205491971e-05,
+      "loss": 36.323,
+      "step": 1163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897221927018387e-05,
+      "loss": 0.2103,
+      "step": 1164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896961472939348e-05,
+      "loss": 0.9224,
+      "step": 1165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896700692699945e-05,
+      "loss": 0.2436,
+      "step": 1166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896439586317543e-05,
+      "loss": 44.4688,
+      "step": 1167
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896178153809537e-05,
+      "loss": 41.0975,
+      "step": 1168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895916395193334e-05,
+      "loss": 39.7826,
+      "step": 1169
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895654310486375e-05,
+      "loss": 0.2496,
+      "step": 1170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895391899706111e-05,
+      "loss": 0.239,
+      "step": 1171
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895129162870023e-05,
+      "loss": 0.2371,
+      "step": 1172
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894866099995608e-05,
+      "loss": 0.9762,
+      "step": 1173
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89460271110039e-05,
+      "loss": 0.2481,
+      "step": 1174
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894338996201909e-05,
+      "loss": 40.4717,
+      "step": 1175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89407495531773e-05,
+      "loss": 37.5122,
+      "step": 1176
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893810588465439e-05,
+      "loss": 0.9526,
+      "step": 1177
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893545895662647e-05,
+      "loss": 0.974,
+      "step": 1178
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893280876926984e-05,
+      "loss": 41.7851,
+      "step": 1179
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893015532276098e-05,
+      "loss": 1.0232,
+      "step": 1180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892749861727664e-05,
+      "loss": 38.9891,
+      "step": 1181
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892483865299376e-05,
+      "loss": 0.9748,
+      "step": 1182
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892217543008951e-05,
+      "loss": 41.0904,
+      "step": 1183
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891950894874127e-05,
+      "loss": 0.2092,
+      "step": 1184
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891683920912664e-05,
+      "loss": 41.1706,
+      "step": 1185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891416621142346e-05,
+      "loss": 0.2784,
+      "step": 1186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891148995580974e-05,
+      "loss": 0.9808,
+      "step": 1187
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890881044246374e-05,
+      "loss": 39.277,
+      "step": 1188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890612767156393e-05,
+      "loss": 0.9517,
+      "step": 1189
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890344164328897e-05,
+      "loss": 39.4601,
+      "step": 1190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890075235781779e-05,
+      "loss": 0.9397,
+      "step": 1191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.889805981532952e-05,
+      "loss": 0.2332,
+      "step": 1192
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889536401600345e-05,
+      "loss": 0.2282,
+      "step": 1193
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889266496001918e-05,
+      "loss": 0.9797,
+      "step": 1194
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888996264755646e-05,
+      "loss": 0.9796,
+      "step": 1195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888725707879525e-05,
+      "loss": 0.2426,
+      "step": 1196
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88845482539158e-05,
+      "loss": 0.98,
+      "step": 1197
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888183617309851e-05,
+      "loss": 34.4497,
+      "step": 1198
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887912083652401e-05,
+      "loss": 39.3318,
+      "step": 1199
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887640224437315e-05,
+      "loss": 0.9579,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887368039682703e-05,
+      "loss": 1.0013,
+      "step": 1201
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887095529406691e-05,
+      "loss": 0.2722,
+      "step": 1202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886822693627431e-05,
+      "loss": 0.2126,
+      "step": 1203
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886549532363094e-05,
+      "loss": 0.9843,
+      "step": 1204
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886276045631874e-05,
+      "loss": 0.2753,
+      "step": 1205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886002233451988e-05,
+      "loss": 42.4759,
+      "step": 1206
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885728095841671e-05,
+      "loss": 0.2468,
+      "step": 1207
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885453632819183e-05,
+      "loss": 0.9652,
+      "step": 1208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885178844402805e-05,
+      "loss": 0.9868,
+      "step": 1209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884903730610837e-05,
+      "loss": 0.9807,
+      "step": 1210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884628291461606e-05,
+      "loss": 42.9436,
+      "step": 1211
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884352526973454e-05,
+      "loss": 42.5738,
+      "step": 1212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884076437164752e-05,
+      "loss": 0.9302,
+      "step": 1213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883800022053886e-05,
+      "loss": 40.1155,
+      "step": 1214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883523281659268e-05,
+      "loss": 36.8151,
+      "step": 1215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883246215999329e-05,
+      "loss": 36.2971,
+      "step": 1216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882968825092522e-05,
+      "loss": 40.4584,
+      "step": 1217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882691108957327e-05,
+      "loss": 39.7867,
+      "step": 1218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882413067612236e-05,
+      "loss": 0.9955,
+      "step": 1219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882134701075772e-05,
+      "loss": 0.9567,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881856009366473e-05,
+      "loss": 42.0719,
+      "step": 1221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8815769925029e-05,
+      "loss": 40.3441,
+      "step": 1222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881297650503641e-05,
+      "loss": 37.9645,
+      "step": 1223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881017983387296e-05,
+      "loss": 0.2679,
+      "step": 1224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880737991172497e-05,
+      "loss": 40.4283,
+      "step": 1225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880457673877892e-05,
+      "loss": 38.9387,
+      "step": 1226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880177031522149e-05,
+      "loss": 0.233,
+      "step": 1227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879896064123961e-05,
+      "loss": 1.0081,
+      "step": 1228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879614771702044e-05,
+      "loss": 40.0192,
+      "step": 1229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87933315427513e-05,
+      "loss": 36.707,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879051211861978e-05,
+      "loss": 39.2562,
+      "step": 1231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878768944481366e-05,
+      "loss": 39.8391,
+      "step": 1232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878486352152094e-05,
+      "loss": 37.3688,
+      "step": 1233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878203434892986e-05,
+      "loss": 39.3168,
+      "step": 1234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877920192722883e-05,
+      "loss": 1.0095,
+      "step": 1235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877636625660652e-05,
+      "loss": 0.989,
+      "step": 1236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87735273372518e-05,
+      "loss": 38.7147,
+      "step": 1237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877068516935373e-05,
+      "loss": 32.6513,
+      "step": 1238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876783975310165e-05,
+      "loss": 0.9229,
+      "step": 1239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876499108868504e-05,
+      "loss": 0.232,
+      "step": 1240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876213917629366e-05,
+      "loss": 38.008,
+      "step": 1241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875928401611745e-05,
+      "loss": 0.9666,
+      "step": 1242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875642560834657e-05,
+      "loss": 0.2106,
+      "step": 1243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875356395317143e-05,
+      "loss": 36.5233,
+      "step": 1244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875069905078261e-05,
+      "loss": 42.3355,
+      "step": 1245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874783090137092e-05,
+      "loss": 36.9385,
+      "step": 1246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874495950512739e-05,
+      "loss": 35.0483,
+      "step": 1247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87420848622433e-05,
+      "loss": 37.4544,
+      "step": 1248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873920697291007e-05,
+      "loss": 0.9054,
+      "step": 1249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873632583731942e-05,
+      "loss": 37.8007,
+      "step": 1250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873344145566324e-05,
+      "loss": 0.217,
+      "step": 1251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873055382813361e-05,
+      "loss": 0.9539,
+      "step": 1252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872766295492288e-05,
+      "loss": 34.9902,
+      "step": 1253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872476883622362e-05,
+      "loss": 0.958,
+      "step": 1254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872187147222856e-05,
+      "loss": 0.9675,
+      "step": 1255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871897086313069e-05,
+      "loss": 38.2742,
+      "step": 1256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871606700912322e-05,
+      "loss": 34.8488,
+      "step": 1257
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871315991039951e-05,
+      "loss": 36.6386,
+      "step": 1258
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871024956715323e-05,
+      "loss": 36.2852,
+      "step": 1259
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870733597957821e-05,
+      "loss": 0.2534,
+      "step": 1260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870441914786852e-05,
+      "loss": 1.0147,
+      "step": 1261
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870149907221842e-05,
+      "loss": 35.219,
+      "step": 1262
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869857575282241e-05,
+      "loss": 0.2663,
+      "step": 1263
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869564918987519e-05,
+      "loss": 0.9747,
+      "step": 1264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869271938357167e-05,
+      "loss": 36.8821,
+      "step": 1265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868978633410702e-05,
+      "loss": 37.1921,
+      "step": 1266
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868685004167656e-05,
+      "loss": 34.5692,
+      "step": 1267
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868391050647588e-05,
+      "loss": 36.9454,
+      "step": 1268
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868096772870077e-05,
+      "loss": 0.2883,
+      "step": 1269
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867802170854724e-05,
+      "loss": 38.6262,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867507244621148e-05,
+      "loss": 38.4759,
+      "step": 1271
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867211994188995e-05,
+      "loss": 34.8777,
+      "step": 1272
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86691641957793e-05,
+      "loss": 0.9964,
+      "step": 1273
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866620520807637e-05,
+      "loss": 0.9998,
+      "step": 1274
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866324297897826e-05,
+      "loss": 38.6639,
+      "step": 1275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866027750868227e-05,
+      "loss": 38.2855,
+      "step": 1276
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865730879738591e-05,
+      "loss": 36.0875,
+      "step": 1277
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865433684528691e-05,
+      "loss": 0.9923,
+      "step": 1278
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865136165258323e-05,
+      "loss": 0.9367,
+      "step": 1279
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864838321947302e-05,
+      "loss": 36.7003,
+      "step": 1280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864540154615465e-05,
+      "loss": 0.997,
+      "step": 1281
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864241663282672e-05,
+      "loss": 37.2022,
+      "step": 1282
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863942847968804e-05,
+      "loss": 34.3547,
+      "step": 1283
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863643708693764e-05,
+      "loss": 1.0051,
+      "step": 1284
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863344245477475e-05,
+      "loss": 36.0165,
+      "step": 1285
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863044458339883e-05,
+      "loss": 37.4201,
+      "step": 1286
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862744347300957e-05,
+      "loss": 34.7606,
+      "step": 1287
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862443912380681e-05,
+      "loss": 39.4492,
+      "step": 1288
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86214315359907e-05,
+      "loss": 0.9173,
+      "step": 1289
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861842070976155e-05,
+      "loss": 0.2091,
+      "step": 1290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861540664531989e-05,
+      "loss": 40.8834,
+      "step": 1291
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861238934286647e-05,
+      "loss": 0.181,
+      "step": 1292
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860936880260226e-05,
+      "loss": 44.2806,
+      "step": 1293
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860634502472842e-05,
+      "loss": 41.8532,
+      "step": 1294
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860331800944638e-05,
+      "loss": 0.2712,
+      "step": 1295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860028775695773e-05,
+      "loss": 0.2207,
+      "step": 1296
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859725426746431e-05,
+      "loss": 0.2209,
+      "step": 1297
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859421754116818e-05,
+      "loss": 36.0672,
+      "step": 1298
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859117757827156e-05,
+      "loss": 0.2914,
+      "step": 1299
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858813437897697e-05,
+      "loss": 37.231,
+      "step": 1300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858508794348705e-05,
+      "loss": 0.9832,
+      "step": 1301
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858203827200476e-05,
+      "loss": 38.1826,
+      "step": 1302
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857898536473318e-05,
+      "loss": 0.9899,
+      "step": 1303
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857592922187568e-05,
+      "loss": 34.4871,
+      "step": 1304
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85728698436358e-05,
+      "loss": 0.2553,
+      "step": 1305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856980723021732e-05,
+      "loss": 35.44,
+      "step": 1306
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856674138182418e-05,
+      "loss": 38.0982,
+      "step": 1307
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856367229866064e-05,
+      "loss": 0.9588,
+      "step": 1308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856059998093107e-05,
+      "loss": 35.1571,
+      "step": 1309
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855752442884013e-05,
+      "loss": 0.976,
+      "step": 1310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855444564259266e-05,
+      "loss": 1.0083,
+      "step": 1311
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85513636223937e-05,
+      "loss": 38.0767,
+      "step": 1312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854827836844855e-05,
+      "loss": 0.9671,
+      "step": 1313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854518988096269e-05,
+      "loss": 1.0052,
+      "step": 1314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854209816014183e-05,
+      "loss": 33.8943,
+      "step": 1315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853900320619192e-05,
+      "loss": 37.1083,
+      "step": 1316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853590501931904e-05,
+      "loss": 0.9288,
+      "step": 1317
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.85328035997296e-05,
+      "loss": 34.2571,
+      "step": 1318
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852969894763014e-05,
+      "loss": 33.7014,
+      "step": 1319
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852659106322745e-05,
+      "loss": 1.016,
+      "step": 1320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852347994672852e-05,
+      "loss": 36.8071,
+      "step": 1321
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.85203655983406e-05,
+      "loss": 37.1229,
+      "step": 1322
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851724801827107e-05,
+      "loss": 0.2368,
+      "step": 1323
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851412720672761e-05,
+      "loss": 34.9659,
+      "step": 1324
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851100316391808e-05,
+      "loss": 0.2505,
+      "step": 1325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850787589005052e-05,
+      "loss": 36.6025,
+      "step": 1326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850474538533328e-05,
+      "loss": 1.0015,
+      "step": 1327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850161164997482e-05,
+      "loss": 35.461,
+      "step": 1328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849847468418388e-05,
+      "loss": 35.0558,
+      "step": 1329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849533448816938e-05,
+      "loss": 0.2452,
+      "step": 1330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849219106214048e-05,
+      "loss": 0.9596,
+      "step": 1331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848904440630657e-05,
+      "loss": 0.2384,
+      "step": 1332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84858945208772e-05,
+      "loss": 0.2683,
+      "step": 1333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848274140606218e-05,
+      "loss": 38.6451,
+      "step": 1334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84795850620715e-05,
+      "loss": 0.2662,
+      "step": 1335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847642548911543e-05,
+      "loss": 38.4079,
+      "step": 1336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847326268740438e-05,
+      "loss": 33.6351,
+      "step": 1337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847009665714904e-05,
+      "loss": 0.2059,
+      "step": 1338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846692739856024e-05,
+      "loss": 34.4984,
+      "step": 1339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846375491184909e-05,
+      "loss": 34.5274,
+      "step": 1340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846057919722688e-05,
+      "loss": 36.268,
+      "step": 1341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845740025490516e-05,
+      "loss": 0.2171,
+      "step": 1342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845421808509563e-05,
+      "loss": 34.6124,
+      "step": 1343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845103268801024e-05,
+      "loss": 0.9852,
+      "step": 1344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844784406386116e-05,
+      "loss": 0.9603,
+      "step": 1345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844465221286077e-05,
+      "loss": 36.409,
+      "step": 1346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844145713522165e-05,
+      "loss": 35.0799,
+      "step": 1347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843825883115662e-05,
+      "loss": 34.7625,
+      "step": 1348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84350573008787e-05,
+      "loss": 36.3814,
+      "step": 1349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843185254460111e-05,
+      "loss": 0.9316,
+      "step": 1350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842864456253734e-05,
+      "loss": 33.5849,
+      "step": 1351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8425433354901e-05,
+      "loss": 30.475,
+      "step": 1352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842221892190602e-05,
+      "loss": 36.3373,
+      "step": 1353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841900126376648e-05,
+      "loss": 35.0954,
+      "step": 1354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841578038069668e-05,
+      "loss": 35.3088,
+      "step": 1355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841255627291118e-05,
+      "loss": 0.9635,
+      "step": 1356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840932894062467e-05,
+      "loss": 0.1832,
+      "step": 1357
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840609838405215e-05,
+      "loss": 33.7195,
+      "step": 1358
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840286460340876e-05,
+      "loss": 34.1572,
+      "step": 1359
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83996275989099e-05,
+      "loss": 34.0639,
+      "step": 1360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839638737077117e-05,
+      "loss": 35.2556,
+      "step": 1361
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839314391920837e-05,
+      "loss": 0.2311,
+      "step": 1362
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838989724443756e-05,
+      "loss": 32.7057,
+      "step": 1363
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838664734667495e-05,
+      "loss": 0.2567,
+      "step": 1364
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8383394226137e-05,
+      "loss": 0.2269,
+      "step": 1365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838013788304041e-05,
+      "loss": 32.7868,
+      "step": 1366
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837687831760205e-05,
+      "loss": 0.2517,
+      "step": 1367
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837361553003902e-05,
+      "loss": 32.3546,
+      "step": 1368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837034952056865e-05,
+      "loss": 30.9944,
+      "step": 1369
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836708028940844e-05,
+      "loss": 0.9556,
+      "step": 1370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836380783677617e-05,
+      "loss": 0.9252,
+      "step": 1371
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83605321628898e-05,
+      "loss": 33.4515,
+      "step": 1372
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835725326796748e-05,
+      "loss": 37.9604,
+      "step": 1373
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835397115222762e-05,
+      "loss": 0.9924,
+      "step": 1374
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835068581588883e-05,
+      "loss": 32.4672,
+      "step": 1375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834739725916988e-05,
+      "loss": 36.5131,
+      "step": 1376
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834410548228987e-05,
+      "loss": 0.1766,
+      "step": 1377
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8340810485468e-05,
+      "loss": 0.9836,
+      "step": 1378
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833751226892376e-05,
+      "loss": 34.4091,
+      "step": 1379
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833421083287682e-05,
+      "loss": 36.6336,
+      "step": 1380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833090617754705e-05,
+      "loss": 35.7056,
+      "step": 1381
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832759830315457e-05,
+      "loss": 35.7621,
+      "step": 1382
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832428720991972e-05,
+      "loss": 0.9282,
+      "step": 1383
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8320972898063e-05,
+      "loss": 33.9711,
+      "step": 1384
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831765536780518e-05,
+      "loss": 0.9491,
+      "step": 1385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831433461936722e-05,
+      "loss": 35.3751,
+      "step": 1386
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83110106529703e-05,
+      "loss": 35.5329,
+      "step": 1387
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830768346883579e-05,
+      "loss": 0.2273,
+      "step": 1388
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830435306718532e-05,
+      "loss": 35.5525,
+      "step": 1389
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83010194482407e-05,
+      "loss": 0.2368,
+      "step": 1390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829768261222399e-05,
+      "loss": 0.9901,
+      "step": 1391
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829434255935739e-05,
+      "loss": 33.6457,
+      "step": 1392
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829099928986338e-05,
+      "loss": 0.983,
+      "step": 1393
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828765280396467e-05,
+      "loss": 38.3107,
+      "step": 1394
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828430310188412e-05,
+      "loss": 32.2074,
+      "step": 1395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828095018384484e-05,
+      "loss": 35.4988,
+      "step": 1396
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827759405007016e-05,
+      "loss": 33.3467,
+      "step": 1397
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827423470078359e-05,
+      "loss": 0.9605,
+      "step": 1398
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82708721362089e-05,
+      "loss": 31.6257,
+      "step": 1399
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826750635657005e-05,
+      "loss": 34.3765,
+      "step": 1400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826413736209122e-05,
+      "loss": 27.7347,
+      "step": 1401
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82607651529968e-05,
+      "loss": 0.2152,
+      "step": 1402
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825738972951138e-05,
+      "loss": 1.051,
+      "step": 1403
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825401109185978e-05,
+      "loss": 0.9742,
+      "step": 1404
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825062924026707e-05,
+      "loss": 32.5468,
+      "step": 1405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824724417495845e-05,
+      "loss": 36.9039,
+      "step": 1406
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824385589615938e-05,
+      "loss": 31.3324,
+      "step": 1407
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824046440409558e-05,
+      "loss": 0.2249,
+      "step": 1408
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823706969899292e-05,
+      "loss": 0.9949,
+      "step": 1409
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823367178107747e-05,
+      "loss": 37.6938,
+      "step": 1410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82302706505756e-05,
+      "loss": 34.2757,
+      "step": 1411
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82268663077138e-05,
+      "loss": 32.6118,
+      "step": 1412
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822345875271883e-05,
+      "loss": 31.9864,
+      "step": 1413
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822004798581765e-05,
+      "loss": 31.6684,
+      "step": 1414
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821663400723744e-05,
+      "loss": 0.9891,
+      "step": 1415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821321681720558e-05,
+      "loss": 0.2216,
+      "step": 1416
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820979641594966e-05,
+      "loss": 0.2328,
+      "step": 1417
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820637280369751e-05,
+      "loss": 37.761,
+      "step": 1418
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820294598067714e-05,
+      "loss": 0.9763,
+      "step": 1419
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819951594711683e-05,
+      "loss": 34.2438,
+      "step": 1420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819608270324499e-05,
+      "loss": 0.9726,
+      "step": 1421
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819264624929033e-05,
+      "loss": 0.2065,
+      "step": 1422
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818920658548169e-05,
+      "loss": 35.2635,
+      "step": 1423
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81857637120482e-05,
+      "loss": 1.0144,
+      "step": 1424
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818231762921917e-05,
+      "loss": 36.0404,
+      "step": 1425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817886833722411e-05,
+      "loss": 0.224,
+      "step": 1426
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817541583629278e-05,
+      "loss": 0.949,
+      "step": 1427
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817196012665512e-05,
+      "loss": 0.2261,
+      "step": 1428
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816850120854129e-05,
+      "loss": 0.2176,
+      "step": 1429
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816503908218168e-05,
+      "loss": 0.2359,
+      "step": 1430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816157374780688e-05,
+      "loss": 32.5967,
+      "step": 1431
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81581052056477e-05,
+      "loss": 36.2716,
+      "step": 1432
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815463345593515e-05,
+      "loss": 33.7492,
+      "step": 1433
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815115849890049e-05,
+      "loss": 33.1194,
+      "step": 1434
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814768033477512e-05,
+      "loss": 31.569,
+      "step": 1435
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814419896379077e-05,
+      "loss": 32.8146,
+      "step": 1436
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814071438617926e-05,
+      "loss": 0.2718,
+      "step": 1437
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813722660217271e-05,
+      "loss": 0.262,
+      "step": 1438
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813373561200339e-05,
+      "loss": 35.4849,
+      "step": 1439
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813024141590384e-05,
+      "loss": 35.5685,
+      "step": 1440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81267440141068e-05,
+      "loss": 32.0916,
+      "step": 1441
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81232434068452e-05,
+      "loss": 33.9387,
+      "step": 1442
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.811973959435219e-05,
+      "loss": 31.2486,
+      "step": 1443
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811623257686114e-05,
+      "loss": 0.2435,
+      "step": 1444
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811272235460564e-05,
+      "loss": 34.7344,
+      "step": 1445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810920892781951e-05,
+      "loss": 34.3665,
+      "step": 1446
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810569229673671e-05,
+      "loss": 0.2263,
+      "step": 1447
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.81021724615915e-05,
+      "loss": 34.6323,
+      "step": 1448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809864942261832e-05,
+      "loss": 33.3948,
+      "step": 1449
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809512318005181e-05,
+      "loss": 34.5085,
+      "step": 1450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809159373412683e-05,
+      "loss": 0.9796,
+      "step": 1451
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808806108507846e-05,
+      "loss": 33.2468,
+      "step": 1452
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808452523314199e-05,
+      "loss": 0.9778,
+      "step": 1453
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808098617855292e-05,
+      "loss": 34.5614,
+      "step": 1454
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807744392154699e-05,
+      "loss": 0.9621,
+      "step": 1455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80738984623601e-05,
+      "loss": 0.2613,
+      "step": 1456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807034980122842e-05,
+      "loss": 0.9692,
+      "step": 1457
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806679793838829e-05,
+      "loss": 0.2083,
+      "step": 1458
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806324287407629e-05,
+      "loss": 34.9837,
+      "step": 1459
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805968460852919e-05,
+      "loss": 31.7012,
+      "step": 1460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805612314198401e-05,
+      "loss": 32.1196,
+      "step": 1461
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805255847467794e-05,
+      "loss": 31.9956,
+      "step": 1462
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804899060684842e-05,
+      "loss": 0.9813,
+      "step": 1463
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804541953873305e-05,
+      "loss": 31.2999,
+      "step": 1464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804184527056974e-05,
+      "loss": 35.2551,
+      "step": 1465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80382678025965e-05,
+      "loss": 0.9422,
+      "step": 1466
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803468713505163e-05,
+      "loss": 34.4575,
+      "step": 1467
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803110326817363e-05,
+      "loss": 35.1968,
+      "step": 1468
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802751620220117e-05,
+      "loss": 33.8854,
+      "step": 1469
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80239259373732e-05,
+      "loss": 34.6193,
+      "step": 1470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802033247392881e-05,
+      "loss": 32.8792,
+      "step": 1471
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801673581210738e-05,
+      "loss": 31.9711,
+      "step": 1472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801313595214842e-05,
+      "loss": 36.4603,
+      "step": 1473
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800953289429175e-05,
+      "loss": 0.9814,
+      "step": 1474
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800592663877733e-05,
+      "loss": 35.8018,
+      "step": 1475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800231718584533e-05,
+      "loss": 32.023,
+      "step": 1476
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.799870453573618e-05,
+      "loss": 34.1777,
+      "step": 1477
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.799508868869049e-05,
+      "loss": 33.0868,
+      "step": 1478
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79914696449491e-05,
+      "loss": 31.2575,
+      "step": 1479
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798784740475304e-05,
+      "loss": 0.1933,
+      "step": 1480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79842219683436e-05,
+      "loss": 34.6152,
+      "step": 1481
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798059333596221e-05,
+      "loss": 0.9858,
+      "step": 1482
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797696150785058e-05,
+      "loss": 30.2425,
+      "step": 1483
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79733264842506e-05,
+      "loss": 31.6854,
+      "step": 1484
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796968826540438e-05,
+      "loss": 33.4633,
+      "step": 1485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796604685155424e-05,
+      "loss": 0.9429,
+      "step": 1486
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796240224294271e-05,
+      "loss": 29.3228,
+      "step": 1487
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795875443981256e-05,
+      "loss": 29.1921,
+      "step": 1488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795510344240672e-05,
+      "loss": 32.5472,
+      "step": 1489
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795144925096837e-05,
+      "loss": 34.8198,
+      "step": 1490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794779186574092e-05,
+      "loss": 0.9367,
+      "step": 1491
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794413128696796e-05,
+      "loss": 32.8966,
+      "step": 1492
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79404675148933e-05,
+      "loss": 0.2245,
+      "step": 1493
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793680054976093e-05,
+      "loss": 29.379,
+      "step": 1494
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793313039181514e-05,
+      "loss": 0.9845,
+      "step": 1495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792945704130034e-05,
+      "loss": 0.9834,
+      "step": 1496
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792578049846123e-05,
+      "loss": 32.9589,
+      "step": 1497
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792210076354264e-05,
+      "loss": 32.1899,
+      "step": 1498
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79184178367897e-05,
+      "loss": 32.5261,
+      "step": 1499
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791473171844767e-05,
+      "loss": 33.2431,
+      "step": 1500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79110424087621e-05,
+      "loss": 0.2251,
+      "step": 1501
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790734990797869e-05,
+      "loss": 31.915,
+      "step": 1502
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790365421634337e-05,
+      "loss": 29.5992,
+      "step": 1503
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789995533410233e-05,
+      "loss": 31.2195,
+      "step": 1504
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78962532615019e-05,
+      "loss": 31.1736,
+      "step": 1505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789254799878866e-05,
+      "loss": 29.9862,
+      "step": 1506
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788883954620942e-05,
+      "loss": 0.2369,
+      "step": 1507
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788512790401114e-05,
+      "loss": 0.9606,
+      "step": 1508
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788141307244105e-05,
+      "loss": 30.8511,
+      "step": 1509
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78776950517466e-05,
+      "loss": 0.2064,
+      "step": 1510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787397384217539e-05,
+      "loss": 1.0123,
+      "step": 1511
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787024944397529e-05,
+      "loss": 0.1789,
+      "step": 1512
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786652185739438e-05,
+      "loss": 0.2572,
+      "step": 1513
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786279108268088e-05,
+      "loss": 0.966,
+      "step": 1514
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785905712008333e-05,
+      "loss": 0.922,
+      "step": 1515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78553199698504e-05,
+      "loss": 34.0705,
+      "step": 1516
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785157963223102e-05,
+      "loss": 1.008,
+      "step": 1517
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78478361074743e-05,
+      "loss": 0.2462,
+      "step": 1518
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784408939582959e-05,
+      "loss": 0.9462,
+      "step": 1519
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784033949754641e-05,
+      "loss": 0.9538,
+      "step": 1520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783658641287455e-05,
+      "loss": 37.8837,
+      "step": 1521
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783283014206397e-05,
+      "loss": 0.997,
+      "step": 1522
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782907068536487e-05,
+      "loss": 28.0233,
+      "step": 1523
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782530804302763e-05,
+      "loss": 0.9493,
+      "step": 1524
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782154221530285e-05,
+      "loss": 31.3336,
+      "step": 1525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781777320244139e-05,
+      "loss": 31.8628,
+      "step": 1526
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781400100469425e-05,
+      "loss": 32.282,
+      "step": 1527
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781022562231268e-05,
+      "loss": 0.9283,
+      "step": 1528
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780644705554813e-05,
+      "loss": 0.2162,
+      "step": 1529
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780266530465231e-05,
+      "loss": 0.9797,
+      "step": 1530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779888036987707e-05,
+      "loss": 34.4667,
+      "step": 1531
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77950922514745e-05,
+      "loss": 35.2586,
+      "step": 1532
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779130094969693e-05,
+      "loss": 0.9575,
+      "step": 1533
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778750646479686e-05,
+      "loss": 1.0144,
+      "step": 1534
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778370879702702e-05,
+      "loss": 33.2126,
+      "step": 1535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777990794664036e-05,
+      "loss": 0.2511,
+      "step": 1536
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777610391389003e-05,
+      "loss": 35.3159,
+      "step": 1537
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777229669902941e-05,
+      "loss": 31.0651,
+      "step": 1538
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776848630231206e-05,
+      "loss": 0.2392,
+      "step": 1539
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776467272399177e-05,
+      "loss": 30.0857,
+      "step": 1540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776085596432255e-05,
+      "loss": 29.1782,
+      "step": 1541
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775703602355861e-05,
+      "loss": 0.2304,
+      "step": 1542
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775321290195438e-05,
+      "loss": 0.9747,
+      "step": 1543
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77493865997645e-05,
+      "loss": 34.1639,
+      "step": 1544
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774555711724382e-05,
+      "loss": 31.5956,
+      "step": 1545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774172445464739e-05,
+      "loss": 29.3386,
+      "step": 1546
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773788861223049e-05,
+      "loss": 30.8204,
+      "step": 1547
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77340495902486e-05,
+      "loss": 0.9708,
+      "step": 1548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773020738895742e-05,
+      "loss": 32.2992,
+      "step": 1549
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772636200861287e-05,
+      "loss": 33.4888,
+      "step": 1550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772251344947105e-05,
+      "loss": 30.2792,
+      "step": 1551
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771866171178831e-05,
+      "loss": 29.8238,
+      "step": 1552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771480679582118e-05,
+      "loss": 28.5033,
+      "step": 1553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771094870182642e-05,
+      "loss": 29.6054,
+      "step": 1554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770708743006101e-05,
+      "loss": 1.0052,
+      "step": 1555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77032229807821e-05,
+      "loss": 0.272,
+      "step": 1556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76993553542471e-05,
+      "loss": 0.9888,
+      "step": 1557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769548455071362e-05,
+      "loss": 33.6853,
+      "step": 1558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769161057043945e-05,
+      "loss": 1.0145,
+      "step": 1559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768773341368262e-05,
+      "loss": 30.7125,
+      "step": 1560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768385308070138e-05,
+      "loss": 31.9611,
+      "step": 1561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767996957175418e-05,
+      "loss": 0.2251,
+      "step": 1562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767608288709967e-05,
+      "loss": 31.0846,
+      "step": 1563
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767219302699671e-05,
+      "loss": 32.2555,
+      "step": 1564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766829999170441e-05,
+      "loss": 0.9501,
+      "step": 1565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766440378148205e-05,
+      "loss": 29.7321,
+      "step": 1566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766050439658912e-05,
+      "loss": 30.7637,
+      "step": 1567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.765660183728537e-05,
+      "loss": 30.0701,
+      "step": 1568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76526961038307e-05,
+      "loss": 0.2483,
+      "step": 1569
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764878719648527e-05,
+      "loss": 31.8788,
+      "step": 1570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764487511550941e-05,
+      "loss": 32.5654,
+      "step": 1571
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76409598611637e-05,
+      "loss": 32.9212,
+      "step": 1572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763704143370891e-05,
+      "loss": 32.0161,
+      "step": 1573
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763311983340603e-05,
+      "loss": 35.0584,
+      "step": 1574
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762919506051624e-05,
+      "loss": 35.9652,
+      "step": 1575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762526711530098e-05,
+      "loss": 33.4433,
+      "step": 1576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762133599802184e-05,
+      "loss": 33.0581,
+      "step": 1577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761740170894065e-05,
+      "loss": 31.8144,
+      "step": 1578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761346424831946e-05,
+      "loss": 0.9676,
+      "step": 1579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760952361642053e-05,
+      "loss": 0.9409,
+      "step": 1580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760557981350633e-05,
+      "loss": 29.9056,
+      "step": 1581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760163283983951e-05,
+      "loss": 27.9656,
+      "step": 1582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759768269568298e-05,
+      "loss": 28.6107,
+      "step": 1583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759372938129982e-05,
+      "loss": 28.4693,
+      "step": 1584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758977289695335e-05,
+      "loss": 29.5586,
+      "step": 1585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758581324290709e-05,
+      "loss": 0.2253,
+      "step": 1586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758185041942478e-05,
+      "loss": 32.6752,
+      "step": 1587
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757788442677036e-05,
+      "loss": 30.6391,
+      "step": 1588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757391526520797e-05,
+      "loss": 31.3529,
+      "step": 1589
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756994293500196e-05,
+      "loss": 31.2178,
+      "step": 1590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756596743641694e-05,
+      "loss": 0.9944,
+      "step": 1591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756198876971767e-05,
+      "loss": 0.9721,
+      "step": 1592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755800693516918e-05,
+      "loss": 0.9738,
+      "step": 1593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755402193303665e-05,
+      "loss": 0.9612,
+      "step": 1594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755003376358551e-05,
+      "loss": 29.472,
+      "step": 1595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75460424270814e-05,
+      "loss": 31.57,
+      "step": 1596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754204792379014e-05,
+      "loss": 0.9424,
+      "step": 1597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753805025397779e-05,
+      "loss": 28.4375,
+      "step": 1598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753404941791062e-05,
+      "loss": 30.7228,
+      "step": 1599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753004541585511e-05,
+      "loss": 30.3348,
+      "step": 1600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752603824807792e-05,
+      "loss": 0.9431,
+      "step": 1601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752202791484596e-05,
+      "loss": 0.2191,
+      "step": 1602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751801441642636e-05,
+      "loss": 0.2088,
+      "step": 1603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751399775308641e-05,
+      "loss": 0.2261,
+      "step": 1604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.750997792509364e-05,
+      "loss": 30.4781,
+      "step": 1605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75059549327158e-05,
+      "loss": 0.191,
+      "step": 1606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.750192877622084e-05,
+      "loss": 0.9436,
+      "step": 1607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749789945587689e-05,
+      "loss": 0.2559,
+      "step": 1608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749386697195237e-05,
+      "loss": 31.2739,
+      "step": 1609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748983132471583e-05,
+      "loss": 30.2669,
+      "step": 1610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748579251443605e-05,
+      "loss": 0.9277,
+      "step": 1611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748175054138209e-05,
+      "loss": 32.9017,
+      "step": 1612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747770540582311e-05,
+      "loss": 0.2046,
+      "step": 1613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747365710802856e-05,
+      "loss": 0.2631,
+      "step": 1614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746960564826808e-05,
+      "loss": 29.0636,
+      "step": 1615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746555102681149e-05,
+      "loss": 28.4451,
+      "step": 1616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746149324392886e-05,
+      "loss": 32.7949,
+      "step": 1617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745743229989046e-05,
+      "loss": 1.0167,
+      "step": 1618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745336819496677e-05,
+      "loss": 32.7803,
+      "step": 1619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744930092942848e-05,
+      "loss": 28.8161,
+      "step": 1620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744523050354648e-05,
+      "loss": 31.2332,
+      "step": 1621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74411569175919e-05,
+      "loss": 0.9663,
+      "step": 1622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743708017183602e-05,
+      "loss": 0.965,
+      "step": 1623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74330002665504e-05,
+      "loss": 0.953,
+      "step": 1624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742891720200677e-05,
+      "loss": 26.7893,
+      "step": 1625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742483097847709e-05,
+      "loss": 0.9879,
+      "step": 1626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742074159623352e-05,
+      "loss": 0.2066,
+      "step": 1627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.741664905554842e-05,
+      "loss": 1.0017,
+      "step": 1628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.741255335669436e-05,
+      "loss": 1.0318,
+      "step": 1629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740845449994419e-05,
+      "loss": 0.9247,
+      "step": 1630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740435248557086e-05,
+      "loss": 32.6053,
+      "step": 1631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74002473138476e-05,
+      "loss": 0.9412,
+      "step": 1632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739613898504782e-05,
+      "loss": 30.669,
+      "step": 1633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739202749944518e-05,
+      "loss": 32.8165,
+      "step": 1634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.738791285731352e-05,
+      "loss": 31.4196,
+      "step": 1635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.738379505892687e-05,
+      "loss": 0.9613,
+      "step": 1636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73796741045595e-05,
+      "loss": 0.995,
+      "step": 1637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737554999448592e-05,
+      "loss": 0.9921,
+      "step": 1638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737142272898077e-05,
+      "loss": 32.851,
+      "step": 1639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.736729230831895e-05,
+      "loss": 0.9594,
+      "step": 1640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.736315873277562e-05,
+      "loss": 0.2474,
+      "step": 1641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735902200262603e-05,
+      "loss": 31.6664,
+      "step": 1642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735488211814572e-05,
+      "loss": 31.655,
+      "step": 1643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735073907961044e-05,
+      "loss": 30.2136,
+      "step": 1644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734659288729613e-05,
+      "loss": 0.224,
+      "step": 1645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734244354147895e-05,
+      "loss": 1.0315,
+      "step": 1646
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733829104243526e-05,
+      "loss": 0.9638,
+      "step": 1647
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733413539044166e-05,
+      "loss": 30.8626,
+      "step": 1648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732997658577488e-05,
+      "loss": 0.9512,
+      "step": 1649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732581462871196e-05,
+      "loss": 31.8122,
+      "step": 1650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73216495195301e-05,
+      "loss": 0.2116,
+      "step": 1651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731748125850671e-05,
+      "loss": 29.0798,
+      "step": 1652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731330984591942e-05,
+      "loss": 0.2186,
+      "step": 1653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730913528204605e-05,
+      "loss": 28.8798,
+      "step": 1654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730495756716467e-05,
+      "loss": 0.2357,
+      "step": 1655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73007767015535e-05,
+      "loss": 0.9475,
+      "step": 1656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729659268549106e-05,
+      "loss": 0.2203,
+      "step": 1657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729240551925595e-05,
+      "loss": 30.6741,
+      "step": 1658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728821520312712e-05,
+      "loss": 0.2535,
+      "step": 1659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728402173738365e-05,
+      "loss": 31.437,
+      "step": 1660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727982512230482e-05,
+      "loss": 31.0375,
+      "step": 1661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727562535817016e-05,
+      "loss": 29.1372,
+      "step": 1662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72714224452594e-05,
+      "loss": 1.0169,
+      "step": 1663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726721638385246e-05,
+      "loss": 0.9304,
+      "step": 1664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726300717422949e-05,
+      "loss": 0.2219,
+      "step": 1665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725879481667085e-05,
+      "loss": 33.4427,
+      "step": 1666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72545793114571e-05,
+      "loss": 0.2833,
+      "step": 1667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725036065886899e-05,
+      "loss": 0.2213,
+      "step": 1668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724613885918753e-05,
+      "loss": 0.9606,
+      "step": 1669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724191391269391e-05,
+      "loss": 0.2192,
+      "step": 1670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72376858196695e-05,
+      "loss": 33.6498,
+      "step": 1671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.723345458039594e-05,
+      "loss": 34.5869,
+      "step": 1672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722922019515505e-05,
+      "loss": 0.216,
+      "step": 1673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722498266422885e-05,
+      "loss": 29.7249,
+      "step": 1674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722074198789958e-05,
+      "loss": 0.943,
+      "step": 1675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.721649816644969e-05,
+      "loss": 33.2873,
+      "step": 1676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.721225120016184e-05,
+      "loss": 1.0393,
+      "step": 1677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720800108931888e-05,
+      "loss": 34.1456,
+      "step": 1678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720374783420394e-05,
+      "loss": 31.0931,
+      "step": 1679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719949143510024e-05,
+      "loss": 32.4908,
+      "step": 1680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719523189229132e-05,
+      "loss": 0.9868,
+      "step": 1681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719096920606085e-05,
+      "loss": 33.2242,
+      "step": 1682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718670337669279e-05,
+      "loss": 28.2096,
+      "step": 1683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718243440447124e-05,
+      "loss": 0.2532,
+      "step": 1684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717816228968052e-05,
+      "loss": 0.2396,
+      "step": 1685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717388703260519e-05,
+      "loss": 0.9665,
+      "step": 1686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716960863353e-05,
+      "loss": 0.2425,
+      "step": 1687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71653270927399e-05,
+      "loss": 31.0153,
+      "step": 1688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716104241052007e-05,
+      "loss": 0.9548,
+      "step": 1689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71567545871559e-05,
+      "loss": 31.5674,
+      "step": 1690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715246362293297e-05,
+      "loss": 29.1495,
+      "step": 1691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.714816951813708e-05,
+      "loss": 0.9677,
+      "step": 1692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.714387227305422e-05,
+      "loss": 32.622,
+      "step": 1693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.713957188797062e-05,
+      "loss": 30.5831,
+      "step": 1694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.71352683631727e-05,
+      "loss": 0.931,
+      "step": 1695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713096169894712e-05,
+      "loss": 1.0359,
+      "step": 1696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.71266518955807e-05,
+      "loss": 31.9551,
+      "step": 1697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.712233895336049e-05,
+      "loss": 29.8093,
+      "step": 1698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711802287257376e-05,
+      "loss": 1.005,
+      "step": 1699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711370365350798e-05,
+      "loss": 30.3059,
+      "step": 1700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.710938129645083e-05,
+      "loss": 0.231,
+      "step": 1701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.71050558016902e-05,
+      "loss": 0.1809,
+      "step": 1702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.71007271695142e-05,
+      "loss": 28.9306,
+      "step": 1703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70963954002111e-05,
+      "loss": 27.7648,
+      "step": 1704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709206049406946e-05,
+      "loss": 28.5283,
+      "step": 1705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708772245137798e-05,
+      "loss": 34.9292,
+      "step": 1706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70833812724256e-05,
+      "loss": 0.2313,
+      "step": 1707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707903695750145e-05,
+      "loss": 0.9767,
+      "step": 1708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707468950689491e-05,
+      "loss": 31.9238,
+      "step": 1709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707033892089552e-05,
+      "loss": 0.9706,
+      "step": 1710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706598519979304e-05,
+      "loss": 28.2174,
+      "step": 1711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706162834387747e-05,
+      "loss": 28.9427,
+      "step": 1712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705726835343899e-05,
+      "loss": 29.9714,
+      "step": 1713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.7052905228768e-05,
+      "loss": 28.7441,
+      "step": 1714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.704853897015508e-05,
+      "loss": 30.0772,
+      "step": 1715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.704416957789106e-05,
+      "loss": 28.7979,
+      "step": 1716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703979705226698e-05,
+      "loss": 0.9503,
+      "step": 1717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703542139357404e-05,
+      "loss": 0.9369,
+      "step": 1718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703104260210369e-05,
+      "loss": 0.993,
+      "step": 1719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70266606781476e-05,
+      "loss": 0.2283,
+      "step": 1720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702227562199759e-05,
+      "loss": 31.1839,
+      "step": 1721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.701788743394574e-05,
+      "loss": 33.0278,
+      "step": 1722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.701349611428434e-05,
+      "loss": 0.9545,
+      "step": 1723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700910166330583e-05,
+      "loss": 0.193,
+      "step": 1724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700470408130295e-05,
+      "loss": 30.5029,
+      "step": 1725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700030336856857e-05,
+      "loss": 0.2061,
+      "step": 1726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699589952539582e-05,
+      "loss": 0.2367,
+      "step": 1727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699149255207799e-05,
+      "loss": 28.6362,
+      "step": 1728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.698708244890862e-05,
+      "loss": 30.6511,
+      "step": 1729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.698266921618144e-05,
+      "loss": 31.1724,
+      "step": 1730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697825285419039e-05,
+      "loss": 28.5522,
+      "step": 1731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697383336322963e-05,
+      "loss": 28.3992,
+      "step": 1732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.696941074359349e-05,
+      "loss": 32.1893,
+      "step": 1733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.696498499557657e-05,
+      "loss": 28.4134,
+      "step": 1734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.696055611947364e-05,
+      "loss": 29.0351,
+      "step": 1735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695612411557968e-05,
+      "loss": 31.3575,
+      "step": 1736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695168898418986e-05,
+      "loss": 30.0397,
+      "step": 1737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694725072559963e-05,
+      "loss": 0.2248,
+      "step": 1738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694280934010453e-05,
+      "loss": 0.9951,
+      "step": 1739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693836482800044e-05,
+      "loss": 29.2142,
+      "step": 1740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693391718958336e-05,
+      "loss": 29.4648,
+      "step": 1741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69294664251495e-05,
+      "loss": 31.061,
+      "step": 1742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.692501253499535e-05,
+      "loss": 29.6854,
+      "step": 1743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.692055551941752e-05,
+      "loss": 30.6864,
+      "step": 1744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69160953787129e-05,
+      "loss": 31.6822,
+      "step": 1745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.691163211317853e-05,
+      "loss": 0.9817,
+      "step": 1746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690716572311169e-05,
+      "loss": 30.0107,
+      "step": 1747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690269620880986e-05,
+      "loss": 28.8744,
+      "step": 1748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689822357057074e-05,
+      "loss": 27.5453,
+      "step": 1749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689374780869223e-05,
+      "loss": 0.2127,
+      "step": 1750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688926892347243e-05,
+      "loss": 29.9132,
+      "step": 1751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688478691520965e-05,
+      "loss": 29.7215,
+      "step": 1752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688030178420243e-05,
+      "loss": 30.4583,
+      "step": 1753
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687581353074946e-05,
+      "loss": 0.9642,
+      "step": 1754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687132215514973e-05,
+      "loss": 29.202,
+      "step": 1755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.686682765770235e-05,
+      "loss": 30.0826,
+      "step": 1756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68623300387067e-05,
+      "loss": 29.4222,
+      "step": 1757
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685782929846232e-05,
+      "loss": 29.8047,
+      "step": 1758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685332543726899e-05,
+      "loss": 0.941,
+      "step": 1759
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68488184554267e-05,
+      "loss": 28.4185,
+      "step": 1760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.684430835323561e-05,
+      "loss": 31.7389,
+      "step": 1761
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683979513099615e-05,
+      "loss": 29.1238,
+      "step": 1762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683527878900888e-05,
+      "loss": 0.9761,
+      "step": 1763
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683075932757463e-05,
+      "loss": 28.0693,
+      "step": 1764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682623674699443e-05,
+      "loss": 29.0135,
+      "step": 1765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682171104756947e-05,
+      "loss": 31.9516,
+      "step": 1766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68171822296012e-05,
+      "loss": 1.0331,
+      "step": 1767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68126502933913e-05,
+      "loss": 0.2238,
+      "step": 1768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680811523924154e-05,
+      "loss": 31.1436,
+      "step": 1769
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680357706745404e-05,
+      "loss": 28.2603,
+      "step": 1770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.679903577833106e-05,
+      "loss": 0.9588,
+      "step": 1771
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.679449137217501e-05,
+      "loss": 0.9758,
+      "step": 1772
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678994384928863e-05,
+      "loss": 28.8561,
+      "step": 1773
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678539320997477e-05,
+      "loss": 27.5738,
+      "step": 1774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678083945453657e-05,
+      "loss": 29.9497,
+      "step": 1775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677628258327728e-05,
+      "loss": 27.9495,
+      "step": 1776
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677172259650043e-05,
+      "loss": 31.6125,
+      "step": 1777
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676715949450974e-05,
+      "loss": 32.3372,
+      "step": 1778
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676259327760913e-05,
+      "loss": 27.2042,
+      "step": 1779
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675802394610273e-05,
+      "loss": 28.2697,
+      "step": 1780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675345150029489e-05,
+      "loss": 0.2303,
+      "step": 1781
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674887594049013e-05,
+      "loss": 29.009,
+      "step": 1782
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674429726699323e-05,
+      "loss": 25.5935,
+      "step": 1783
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.673971548010914e-05,
+      "loss": 0.2513,
+      "step": 1784
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.673513058014303e-05,
+      "loss": 0.2664,
+      "step": 1785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.673054256740027e-05,
+      "loss": 32.2384,
+      "step": 1786
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672595144218646e-05,
+      "loss": 0.2755,
+      "step": 1787
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672135720480736e-05,
+      "loss": 29.1947,
+      "step": 1788
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.671675985556898e-05,
+      "loss": 30.095,
+      "step": 1789
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.671215939477755e-05,
+      "loss": 28.5795,
+      "step": 1790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670755582273945e-05,
+      "loss": 0.2274,
+      "step": 1791
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670294913976132e-05,
+      "loss": 27.189,
+      "step": 1792
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669833934614996e-05,
+      "loss": 32.1515,
+      "step": 1793
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669372644221245e-05,
+      "loss": 27.9666,
+      "step": 1794
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668911042825597e-05,
+      "loss": 0.2056,
+      "step": 1795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668449130458802e-05,
+      "loss": 0.2611,
+      "step": 1796
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667986907151622e-05,
+      "loss": 0.9415,
+      "step": 1797
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667524372934846e-05,
+      "loss": 0.9743,
+      "step": 1798
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66706152783928e-05,
+      "loss": 0.2503,
+      "step": 1799
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.666598371895751e-05,
+      "loss": 33.0643,
+      "step": 1800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.666134905135107e-05,
+      "loss": 29.5455,
+      "step": 1801
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665671127588218e-05,
+      "loss": 28.606,
+      "step": 1802
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665207039285974e-05,
+      "loss": 29.43,
+      "step": 1803
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.664742640259284e-05,
+      "loss": 27.2397,
+      "step": 1804
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.664277930539081e-05,
+      "loss": 28.7676,
+      "step": 1805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663812910156314e-05,
+      "loss": 0.9656,
+      "step": 1806
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663347579141959e-05,
+      "loss": 28.1731,
+      "step": 1807
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.662881937527006e-05,
+      "loss": 28.9698,
+      "step": 1808
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.662415985342473e-05,
+      "loss": 0.9642,
+      "step": 1809
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661949722619388e-05,
+      "loss": 24.6906,
+      "step": 1810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661483149388814e-05,
+      "loss": 26.9557,
+      "step": 1811
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661016265681822e-05,
+      "loss": 0.9921,
+      "step": 1812
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660549071529509e-05,
+      "loss": 31.604,
+      "step": 1813
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660081566962993e-05,
+      "loss": 0.2521,
+      "step": 1814
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.659613752013414e-05,
+      "loss": 29.0717,
+      "step": 1815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.659145626711927e-05,
+      "loss": 0.9531,
+      "step": 1816
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.658677191089714e-05,
+      "loss": 0.9718,
+      "step": 1817
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.658208445177972e-05,
+      "loss": 0.2147,
+      "step": 1818
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.657739389007927e-05,
+      "loss": 32.8076,
+      "step": 1819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.657270022610813e-05,
+      "loss": 27.3026,
+      "step": 1820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.6568003460179e-05,
+      "loss": 29.9413,
+      "step": 1821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.656330359260463e-05,
+      "loss": 32.1947,
+      "step": 1822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.655860062369812e-05,
+      "loss": 29.2776,
+      "step": 1823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.655389455377266e-05,
+      "loss": 29.2353,
+      "step": 1824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654918538314173e-05,
+      "loss": 0.9754,
+      "step": 1825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654447311211896e-05,
+      "loss": 28.2386,
+      "step": 1826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653975774101822e-05,
+      "loss": 28.9041,
+      "step": 1827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653503927015357e-05,
+      "loss": 0.2525,
+      "step": 1828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653031769983928e-05,
+      "loss": 27.2485,
+      "step": 1829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.652559303038984e-05,
+      "loss": 27.8043,
+      "step": 1830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.652086526211992e-05,
+      "loss": 28.2988,
+      "step": 1831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.651613439534443e-05,
+      "loss": 26.0419,
+      "step": 1832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.651140043037846e-05,
+      "loss": 27.8485,
+      "step": 1833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650666336753728e-05,
+      "loss": 27.8707,
+      "step": 1834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650192320713646e-05,
+      "loss": 26.876,
+      "step": 1835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.649717994949168e-05,
+      "loss": 31.1403,
+      "step": 1836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.649243359491886e-05,
+      "loss": 0.1755,
+      "step": 1837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.648768414373416e-05,
+      "loss": 30.5776,
+      "step": 1838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.648293159625389e-05,
+      "loss": 0.2144,
+      "step": 1839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.647817595279459e-05,
+      "loss": 0.2291,
+      "step": 1840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.647341721367303e-05,
+      "loss": 29.2387,
+      "step": 1841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646865537920613e-05,
+      "loss": 0.9826,
+      "step": 1842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646389044971108e-05,
+      "loss": 0.9424,
+      "step": 1843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.645912242550523e-05,
+      "loss": 28.6928,
+      "step": 1844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.645435130690617e-05,
+      "loss": 0.2348,
+      "step": 1845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644957709423165e-05,
+      "loss": 29.7786,
+      "step": 1846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64447997877997e-05,
+      "loss": 31.726,
+      "step": 1847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644001938792846e-05,
+      "loss": 27.2737,
+      "step": 1848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.643523589493636e-05,
+      "loss": 26.7944,
+      "step": 1849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.643044930914199e-05,
+      "loss": 30.9229,
+      "step": 1850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642565963086416e-05,
+      "loss": 27.0694,
+      "step": 1851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642086686042189e-05,
+      "loss": 30.1572,
+      "step": 1852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64160709981344e-05,
+      "loss": 0.2063,
+      "step": 1853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64112720443211e-05,
+      "loss": 29.8617,
+      "step": 1854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640646999930163e-05,
+      "loss": 30.3804,
+      "step": 1855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640166486339585e-05,
+      "loss": 0.9333,
+      "step": 1856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63968566369238e-05,
+      "loss": 0.9755,
+      "step": 1857
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63920453202057e-05,
+      "loss": 0.9895,
+      "step": 1858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.638723091356203e-05,
+      "loss": 30.2006,
+      "step": 1859
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.638241341731344e-05,
+      "loss": 27.4298,
+      "step": 1860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637759283178082e-05,
+      "loss": 29.3763,
+      "step": 1861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637276915728522e-05,
+      "loss": 29.3162,
+      "step": 1862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636794239414793e-05,
+      "loss": 29.5097,
+      "step": 1863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636311254269043e-05,
+      "loss": 1.0006,
+      "step": 1864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63582796032344e-05,
+      "loss": 0.2536,
+      "step": 1865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.635344357610177e-05,
+      "loss": 29.7558,
+      "step": 1866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634860446161461e-05,
+      "loss": 30.7831,
+      "step": 1867
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634376226009525e-05,
+      "loss": 0.9484,
+      "step": 1868
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.633891697186618e-05,
+      "loss": 0.9587,
+      "step": 1869
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.633406859725013e-05,
+      "loss": 0.9498,
+      "step": 1870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632921713657001e-05,
+      "loss": 0.9686,
+      "step": 1871
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632436259014898e-05,
+      "loss": 31.7337,
+      "step": 1872
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.631950495831037e-05,
+      "loss": 0.9861,
+      "step": 1873
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63146442413777e-05,
+      "loss": 0.9934,
+      "step": 1874
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.630978043967472e-05,
+      "loss": 28.1761,
+      "step": 1875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63049135535254e-05,
+      "loss": 0.2227,
+      "step": 1876
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.630004358325389e-05,
+      "loss": 0.9677,
+      "step": 1877
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629517052918454e-05,
+      "loss": 1.0204,
+      "step": 1878
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629029439164191e-05,
+      "loss": 32.2434,
+      "step": 1879
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62854151709508e-05,
+      "loss": 29.4037,
+      "step": 1880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.628053286743619e-05,
+      "loss": 26.4291,
+      "step": 1881
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627564748142324e-05,
+      "loss": 32.0067,
+      "step": 1882
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627075901323734e-05,
+      "loss": 27.8447,
+      "step": 1883
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.626586746320412e-05,
+      "loss": 27.5538,
+      "step": 1884
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.626097283164936e-05,
+      "loss": 28.8209,
+      "step": 1885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.625607511889905e-05,
+      "loss": 29.672,
+      "step": 1886
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62511743252794e-05,
+      "loss": 29.1266,
+      "step": 1887
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.624627045111685e-05,
+      "loss": 0.2446,
+      "step": 1888
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.6241363496738e-05,
+      "loss": 27.1077,
+      "step": 1889
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62364534624697e-05,
+      "loss": 31.8891,
+      "step": 1890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.623154034863896e-05,
+      "loss": 25.7761,
+      "step": 1891
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.622662415557302e-05,
+      "loss": 0.9378,
+      "step": 1892
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.622170488359932e-05,
+      "loss": 0.2128,
+      "step": 1893
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62167825330455e-05,
+      "loss": 30.5924,
+      "step": 1894
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.621185710423943e-05,
+      "loss": 0.257,
+      "step": 1895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.620692859750916e-05,
+      "loss": 29.9713,
+      "step": 1896
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.620199701318296e-05,
+      "loss": 30.276,
+      "step": 1897
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.619706235158925e-05,
+      "loss": 28.439,
+      "step": 1898
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.619212461305676e-05,
+      "loss": 26.9075,
+      "step": 1899
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618718379791434e-05,
+      "loss": 27.7466,
+      "step": 1900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618223990649107e-05,
+      "loss": 30.2629,
+      "step": 1901
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617729293911625e-05,
+      "loss": 28.2452,
+      "step": 1902
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617234289611935e-05,
+      "loss": 0.2248,
+      "step": 1903
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.616738977783009e-05,
+      "loss": 0.9697,
+      "step": 1904
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.616243358457833e-05,
+      "loss": 29.2327,
+      "step": 1905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615747431669422e-05,
+      "loss": 32.0436,
+      "step": 1906
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615251197450805e-05,
+      "loss": 0.1962,
+      "step": 1907
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.614754655835035e-05,
+      "loss": 28.4,
+      "step": 1908
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.614257806855182e-05,
+      "loss": 0.2221,
+      "step": 1909
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.613760650544339e-05,
+      "loss": 0.9526,
+      "step": 1910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.613263186935619e-05,
+      "loss": 0.2244,
+      "step": 1911
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.612765416062156e-05,
+      "loss": 29.0094,
+      "step": 1912
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.612267337957104e-05,
+      "loss": 27.1047,
+      "step": 1913
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.611768952653635e-05,
+      "loss": 25.8279,
+      "step": 1914
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.611270260184945e-05,
+      "loss": 26.9881,
+      "step": 1915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.610771260584253e-05,
+      "loss": 28.5143,
+      "step": 1916
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.610271953884788e-05,
+      "loss": 30.9942,
+      "step": 1917
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.609772340119811e-05,
+      "loss": 27.6188,
+      "step": 1918
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.609272419322598e-05,
+      "loss": 26.4025,
+      "step": 1919
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608772191526446e-05,
+      "loss": 28.1319,
+      "step": 1920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60827165676467e-05,
+      "loss": 0.2471,
+      "step": 1921
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607770815070607e-05,
+      "loss": 0.953,
+      "step": 1922
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607269666477621e-05,
+      "loss": 0.964,
+      "step": 1923
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.606768211019088e-05,
+      "loss": 27.4281,
+      "step": 1924
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.606266448728407e-05,
+      "loss": 26.3726,
+      "step": 1925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.605764379638997e-05,
+      "loss": 29.1056,
+      "step": 1926
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.605262003784299e-05,
+      "loss": 28.5141,
+      "step": 1927
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604759321197773e-05,
+      "loss": 0.9982,
+      "step": 1928
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604256331912901e-05,
+      "loss": 0.9758,
+      "step": 1929
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.603753035963184e-05,
+      "loss": 0.9415,
+      "step": 1930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.603249433382144e-05,
+      "loss": 0.9385,
+      "step": 1931
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602745524203323e-05,
+      "loss": 0.9658,
+      "step": 1932
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602241308460283e-05,
+      "loss": 0.9548,
+      "step": 1933
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.601736786186608e-05,
+      "loss": 0.957,
+      "step": 1934
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.601231957415902e-05,
+      "loss": 31.4339,
+      "step": 1935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.600726822181788e-05,
+      "loss": 33.9088,
+      "step": 1936
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.600221380517911e-05,
+      "loss": 30.8545,
+      "step": 1937
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.599715632457935e-05,
+      "loss": 27.3028,
+      "step": 1938
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.599209578035545e-05,
+      "loss": 0.2189,
+      "step": 1939
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.598703217284448e-05,
+      "loss": 0.1853,
+      "step": 1940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.598196550238368e-05,
+      "loss": 0.2272,
+      "step": 1941
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.597689576931052e-05,
+      "loss": 34.2602,
+      "step": 1942
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.597182297396269e-05,
+      "loss": 34.5423,
+      "step": 1943
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.596674711667805e-05,
+      "loss": 0.9735,
+      "step": 1944
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.596166819779465e-05,
+      "loss": 33.0613,
+      "step": 1945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.59565862176508e-05,
+      "loss": 0.1874,
+      "step": 1946
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.595150117658496e-05,
+      "loss": 0.9796,
+      "step": 1947
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.594641307493585e-05,
+      "loss": 32.8737,
+      "step": 1948
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.594132191304233e-05,
+      "loss": 0.9871,
+      "step": 1949
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.59362276912435e-05,
+      "loss": 0.9745,
+      "step": 1950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.593113040987868e-05,
+      "loss": 31.3802,
+      "step": 1951
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.592603006928735e-05,
+      "loss": 29.5726,
+      "step": 1952
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.592092666980924e-05,
+      "loss": 29.3989,
+      "step": 1953
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.591582021178423e-05,
+      "loss": 0.9705,
+      "step": 1954
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.591071069555244e-05,
+      "loss": 28.5332,
+      "step": 1955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.59055981214542e-05,
+      "loss": 29.1749,
+      "step": 1956
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.590048248983003e-05,
+      "loss": 0.9963,
+      "step": 1957
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.589536380102067e-05,
+      "loss": 0.9856,
+      "step": 1958
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.589024205536701e-05,
+      "loss": 0.2202,
+      "step": 1959
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58851172532102e-05,
+      "loss": 0.9913,
+      "step": 1960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.587998939489158e-05,
+      "loss": 0.9912,
+      "step": 1961
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58748584807527e-05,
+      "loss": 0.2125,
+      "step": 1962
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.586972451113528e-05,
+      "loss": 0.2342,
+      "step": 1963
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.586458748638127e-05,
+      "loss": 0.2601,
+      "step": 1964
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.585944740683283e-05,
+      "loss": 31.7586,
+      "step": 1965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58543042728323e-05,
+      "loss": 29.9774,
+      "step": 1966
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.584915808472226e-05,
+      "loss": 0.2373,
+      "step": 1967
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.584400884284545e-05,
+      "loss": 29.6876,
+      "step": 1968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.583885654754486e-05,
+      "loss": 30.9967,
+      "step": 1969
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58337011991636e-05,
+      "loss": 0.9765,
+      "step": 1970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58285427980451e-05,
+      "loss": 0.2328,
+      "step": 1971
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.582338134453289e-05,
+      "loss": 0.2034,
+      "step": 1972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.581821683897078e-05,
+      "loss": 0.9625,
+      "step": 1973
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.581304928170273e-05,
+      "loss": 0.1907,
+      "step": 1974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.580787867307293e-05,
+      "loss": 0.9509,
+      "step": 1975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.580270501342576e-05,
+      "loss": 31.9289,
+      "step": 1976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.579752830310582e-05,
+      "loss": 0.2381,
+      "step": 1977
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.579234854245791e-05,
+      "loss": 0.2274,
+      "step": 1978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.5787165731827e-05,
+      "loss": 0.2212,
+      "step": 1979
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57819798715583e-05,
+      "loss": 0.2171,
+      "step": 1980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.577679096199723e-05,
+      "loss": 30.218,
+      "step": 1981
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.577159900348938e-05,
+      "loss": 0.9489,
+      "step": 1982
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576640399638054e-05,
+      "loss": 34.3843,
+      "step": 1983
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576120594101677e-05,
+      "loss": 0.1713,
+      "step": 1984
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575600483774426e-05,
+      "loss": 31.8439,
+      "step": 1985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575080068690939e-05,
+      "loss": 31.7423,
+      "step": 1986
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.574559348885883e-05,
+      "loss": 0.9214,
+      "step": 1987
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.574038324393939e-05,
+      "loss": 26.3809,
+      "step": 1988
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.573516995249811e-05,
+      "loss": 0.2567,
+      "step": 1989
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.572995361488219e-05,
+      "loss": 29.5952,
+      "step": 1990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.572473423143907e-05,
+      "loss": 0.2137,
+      "step": 1991
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.571951180251639e-05,
+      "loss": 29.4584,
+      "step": 1992
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.5714286328462e-05,
+      "loss": 0.2296,
+      "step": 1993
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.570905780962394e-05,
+      "loss": 28.4617,
+      "step": 1994
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.570382624635042e-05,
+      "loss": 0.9648,
+      "step": 1995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.569859163898993e-05,
+      "loss": 29.6773,
+      "step": 1996
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.569335398789109e-05,
+      "loss": 28.7684,
+      "step": 1997
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.568811329340276e-05,
+      "loss": 28.363,
+      "step": 1998
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.5682869555874e-05,
+      "loss": 32.1427,
+      "step": 1999
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.567762277565406e-05,
+      "loss": 28.2643,
+      "step": 2000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.567237295309241e-05,
+      "loss": 0.9857,
+      "step": 2001
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56671200885387e-05,
+      "loss": 0.9365,
+      "step": 2002
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56618641823428e-05,
+      "loss": 30.2043,
+      "step": 2003
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.565660523485478e-05,
+      "loss": 26.1842,
+      "step": 2004
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56513432464249e-05,
+      "loss": 0.2291,
+      "step": 2005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.564607821740366e-05,
+      "loss": 27.4334,
+      "step": 2006
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.564081014814171e-05,
+      "loss": 29.3596,
+      "step": 2007
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.563553903898995e-05,
+      "loss": 0.9312,
+      "step": 2008
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.563026489029941e-05,
+      "loss": 28.7684,
+      "step": 2009
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.562498770242143e-05,
+      "loss": 28.4385,
+      "step": 2010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.561970747570747e-05,
+      "loss": 0.9624,
+      "step": 2011
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.561442421050922e-05,
+      "loss": 0.2229,
+      "step": 2012
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.560913790717858e-05,
+      "loss": 30.0062,
+      "step": 2013
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.560384856606762e-05,
+      "loss": 28.061,
+      "step": 2014
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.559855618752865e-05,
+      "loss": 0.9115,
+      "step": 2015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.559326077191417e-05,
+      "loss": 26.6809,
+      "step": 2016
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.558796231957687e-05,
+      "loss": 30.9961,
+      "step": 2017
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.558266083086966e-05,
+      "loss": 30.3266,
+      "step": 2018
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.557735630614564e-05,
+      "loss": 28.3318,
+      "step": 2019
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.557204874575811e-05,
+      "loss": 0.9228,
+      "step": 2020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.556673815006059e-05,
+      "loss": 26.495,
+      "step": 2021
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55614245194068e-05,
+      "loss": 29.5284,
+      "step": 2022
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.555610785415061e-05,
+      "loss": 27.4983,
+      "step": 2023
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.555078815464618e-05,
+      "loss": 0.2106,
+      "step": 2024
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.554546542124779e-05,
+      "loss": 26.0752,
+      "step": 2025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.554013965430998e-05,
+      "loss": 28.5392,
+      "step": 2026
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.553481085418747e-05,
+      "loss": 28.1087,
+      "step": 2027
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.552947902123518e-05,
+      "loss": 27.5485,
+      "step": 2028
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.552414415580824e-05,
+      "loss": 0.2066,
+      "step": 2029
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.551880625826197e-05,
+      "loss": 29.7081,
+      "step": 2030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.551346532895189e-05,
+      "loss": 0.9402,
+      "step": 2031
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.550812136823375e-05,
+      "loss": 28.8646,
+      "step": 2032
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.550277437646346e-05,
+      "loss": 0.2641,
+      "step": 2033
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54974243539972e-05,
+      "loss": 0.9568,
+      "step": 2034
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.549207130119124e-05,
+      "loss": 29.2592,
+      "step": 2035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.548671521840218e-05,
+      "loss": 27.2134,
+      "step": 2036
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.548135610598672e-05,
+      "loss": 26.9966,
+      "step": 2037
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.547599396430182e-05,
+      "loss": 28.494,
+      "step": 2038
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.547062879370461e-05,
+      "loss": 27.4585,
+      "step": 2039
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.546526059455244e-05,
+      "loss": 26.1411,
+      "step": 2040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.545988936720287e-05,
+      "loss": 0.9697,
+      "step": 2041
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.545451511201364e-05,
+      "loss": 0.9672,
+      "step": 2042
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.544913782934269e-05,
+      "loss": 0.2246,
+      "step": 2043
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.544375751954819e-05,
+      "loss": 0.9714,
+      "step": 2044
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.543837418298847e-05,
+      "loss": 0.2489,
+      "step": 2045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.543298782002213e-05,
+      "loss": 30.1556,
+      "step": 2046
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.542759843100787e-05,
+      "loss": 26.713,
+      "step": 2047
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.542220601630468e-05,
+      "loss": 28.015,
+      "step": 2048
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.541681057627172e-05,
+      "loss": 0.2192,
+      "step": 2049
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.541141211126834e-05,
+      "loss": 0.958,
+      "step": 2050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.540601062165412e-05,
+      "loss": 29.6134,
+      "step": 2051
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54006061077888e-05,
+      "loss": 30.5993,
+      "step": 2052
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.539519857003237e-05,
+      "loss": 0.2339,
+      "step": 2053
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.538978800874498e-05,
+      "loss": 28.963,
+      "step": 2054
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.538437442428702e-05,
+      "loss": 29.5538,
+      "step": 2055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.537895781701902e-05,
+      "loss": 0.1898,
+      "step": 2056
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.53735381873018e-05,
+      "loss": 28.3277,
+      "step": 2057
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.536811553549629e-05,
+      "loss": 27.7302,
+      "step": 2058
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.536268986196368e-05,
+      "loss": 29.8034,
+      "step": 2059
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.535726116706537e-05,
+      "loss": 28.2272,
+      "step": 2060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.53518294511629e-05,
+      "loss": 26.791,
+      "step": 2061
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.534639471461806e-05,
+      "loss": 27.3063,
+      "step": 2062
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.534095695779283e-05,
+      "loss": 26.4585,
+      "step": 2063
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.53355161810494e-05,
+      "loss": 0.2658,
+      "step": 2064
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533007238475011e-05,
+      "loss": 0.2602,
+      "step": 2065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.532462556925759e-05,
+      "loss": 29.571,
+      "step": 2066
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.53191757349346e-05,
+      "loss": 0.9602,
+      "step": 2067
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.531372288214415e-05,
+      "loss": 27.2769,
+      "step": 2068
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.530826701124939e-05,
+      "loss": 0.9296,
+      "step": 2069
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.530280812261372e-05,
+      "loss": 0.2622,
+      "step": 2070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.529734621660075e-05,
+      "loss": 26.0562,
+      "step": 2071
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.529188129357423e-05,
+      "loss": 28.279,
+      "step": 2072
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.528641335389816e-05,
+      "loss": 0.9291,
+      "step": 2073
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.528094239793674e-05,
+      "loss": 1.0144,
+      "step": 2074
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.527546842605437e-05,
+      "loss": 0.977,
+      "step": 2075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.526999143861562e-05,
+      "loss": 0.9625,
+      "step": 2076
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52645114359853e-05,
+      "loss": 25.8451,
+      "step": 2077
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.525902841852839e-05,
+      "loss": 0.971,
+      "step": 2078
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.525354238661009e-05,
+      "loss": 27.8629,
+      "step": 2079
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52480533405958e-05,
+      "loss": 0.9633,
+      "step": 2080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.524256128085112e-05,
+      "loss": 29.9539,
+      "step": 2081
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.523706620774183e-05,
+      "loss": 27.3314,
+      "step": 2082
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.523156812163392e-05,
+      "loss": 28.3844,
+      "step": 2083
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.522606702289362e-05,
+      "loss": 0.9702,
+      "step": 2084
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52205629118873e-05,
+      "loss": 1.0066,
+      "step": 2085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.521505578898159e-05,
+      "loss": 29.7006,
+      "step": 2086
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.520954565454325e-05,
+      "loss": 0.9756,
+      "step": 2087
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.520403250893933e-05,
+      "loss": 0.243,
+      "step": 2088
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.519851635253699e-05,
+      "loss": 0.1825,
+      "step": 2089
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.519299718570363e-05,
+      "loss": 26.0073,
+      "step": 2090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.518747500880689e-05,
+      "loss": 0.992,
+      "step": 2091
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.518194982221453e-05,
+      "loss": 24.7862,
+      "step": 2092
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.517642162629459e-05,
+      "loss": 0.9959,
+      "step": 2093
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.517089042141524e-05,
+      "loss": 28.0063,
+      "step": 2094
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.516535620794492e-05,
+      "loss": 29.3884,
+      "step": 2095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51598189862522e-05,
+      "loss": 0.9657,
+      "step": 2096
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51542787567059e-05,
+      "loss": 27.9877,
+      "step": 2097
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.514873551967504e-05,
+      "loss": 28.3402,
+      "step": 2098
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.514318927552878e-05,
+      "loss": 0.2073,
+      "step": 2099
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.513764002463658e-05,
+      "loss": 0.1748,
+      "step": 2100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.513208776736802e-05,
+      "loss": 28.2683,
+      "step": 2101
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51265325040929e-05,
+      "loss": 29.7946,
+      "step": 2102
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.512097423518124e-05,
+      "loss": 27.2855,
+      "step": 2103
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.511541296100323e-05,
+      "loss": 26.0072,
+      "step": 2104
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51098486819293e-05,
+      "loss": 0.9783,
+      "step": 2105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.510428139833003e-05,
+      "loss": 0.2496,
+      "step": 2106
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.509871111057624e-05,
+      "loss": 26.9011,
+      "step": 2107
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.509313781903894e-05,
+      "loss": 0.929,
+      "step": 2108
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.508756152408934e-05,
+      "loss": 0.9328,
+      "step": 2109
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.508198222609883e-05,
+      "loss": 27.5718,
+      "step": 2110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.507639992543905e-05,
+      "loss": 26.4096,
+      "step": 2111
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.507081462248177e-05,
+      "loss": 27.5614,
+      "step": 2112
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.506522631759902e-05,
+      "loss": 0.2647,
+      "step": 2113
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.5059635011163e-05,
+      "loss": 25.938,
+      "step": 2114
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.505404070354611e-05,
+      "loss": 0.2467,
+      "step": 2115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.504844339512095e-05,
+      "loss": 26.8012,
+      "step": 2116
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.504284308626037e-05,
+      "loss": 28.6777,
+      "step": 2117
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.503723977733734e-05,
+      "loss": 0.9561,
+      "step": 2118
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.503163346872506e-05,
+      "loss": 0.9873,
+      "step": 2119
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.502602416079696e-05,
+      "loss": 1.0141,
+      "step": 2120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.502041185392664e-05,
+      "loss": 0.9732,
+      "step": 2121
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.501479654848788e-05,
+      "loss": 31.1071,
+      "step": 2122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.500917824485473e-05,
+      "loss": 27.0289,
+      "step": 2123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.500355694340138e-05,
+      "loss": 0.9492,
+      "step": 2124
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49979326445022e-05,
+      "loss": 0.9914,
+      "step": 2125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.499230534853183e-05,
+      "loss": 0.2105,
+      "step": 2126
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.498667505586508e-05,
+      "loss": 0.2285,
+      "step": 2127
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.498104176687692e-05,
+      "loss": 30.7735,
+      "step": 2128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.497540548194257e-05,
+      "loss": 29.3607,
+      "step": 2129
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.496976620143745e-05,
+      "loss": 27.8089,
+      "step": 2130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.496412392573714e-05,
+      "loss": 29.8048,
+      "step": 2131
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.495847865521744e-05,
+      "loss": 0.9941,
+      "step": 2132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.495283039025438e-05,
+      "loss": 28.3148,
+      "step": 2133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.494717913122413e-05,
+      "loss": 0.979,
+      "step": 2134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49415248785031e-05,
+      "loss": 0.9513,
+      "step": 2135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.493586763246788e-05,
+      "loss": 0.2447,
+      "step": 2136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49302073934953e-05,
+      "loss": 31.8863,
+      "step": 2137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.492454416196232e-05,
+      "loss": 33.3447,
+      "step": 2138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.491887793824617e-05,
+      "loss": 0.9788,
+      "step": 2139
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49132087227242e-05,
+      "loss": 0.9521,
+      "step": 2140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.490753651577406e-05,
+      "loss": 26.6712,
+      "step": 2141
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.490186131777354e-05,
+      "loss": 0.2245,
+      "step": 2142
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48961831291006e-05,
+      "loss": 0.8995,
+      "step": 2143
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.489050195013344e-05,
+      "loss": 0.9177,
+      "step": 2144
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.488481778125048e-05,
+      "loss": 28.3726,
+      "step": 2145
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48791306228303e-05,
+      "loss": 27.5186,
+      "step": 2146
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.487344047525166e-05,
+      "loss": 26.3202,
+      "step": 2147
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.486774733889361e-05,
+      "loss": 0.9615,
+      "step": 2148
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48620512141353e-05,
+      "loss": 25.9778,
+      "step": 2149
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48563521013561e-05,
+      "loss": 27.9993,
+      "step": 2150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.485065000093565e-05,
+      "loss": 0.2291,
+      "step": 2151
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48449449132537e-05,
+      "loss": 28.4361,
+      "step": 2152
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.483923683869024e-05,
+      "loss": 0.921,
+      "step": 2153
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.483352577762545e-05,
+      "loss": 0.9717,
+      "step": 2154
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.482781173043975e-05,
+      "loss": 0.2233,
+      "step": 2155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.482209469751367e-05,
+      "loss": 26.8202,
+      "step": 2156
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.481637467922803e-05,
+      "loss": 27.8933,
+      "step": 2157
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.481065167596378e-05,
+      "loss": 27.2803,
+      "step": 2158
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.480492568810212e-05,
+      "loss": 26.9458,
+      "step": 2159
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.479919671602442e-05,
+      "loss": 0.9775,
+      "step": 2160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.479346476011226e-05,
+      "loss": 25.9241,
+      "step": 2161
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.47877298207474e-05,
+      "loss": 29.5127,
+      "step": 2162
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.478199189831183e-05,
+      "loss": 25.9201,
+      "step": 2163
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.477625099318772e-05,
+      "loss": 1.0261,
+      "step": 2164
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.477050710575744e-05,
+      "loss": 24.0185,
+      "step": 2165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.476476023640356e-05,
+      "loss": 0.9657,
+      "step": 2166
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.475901038550885e-05,
+      "loss": 0.9844,
+      "step": 2167
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.475325755345626e-05,
+      "loss": 25.7755,
+      "step": 2168
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.474750174062896e-05,
+      "loss": 0.9547,
+      "step": 2169
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.474174294741035e-05,
+      "loss": 26.6209,
+      "step": 2170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.473598117418394e-05,
+      "loss": 0.9716,
+      "step": 2171
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.473021642133353e-05,
+      "loss": 26.9201,
+      "step": 2172
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.472444868924305e-05,
+      "loss": 29.2032,
+      "step": 2173
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.471867797829668e-05,
+      "loss": 26.9147,
+      "step": 2174
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.471290428887877e-05,
+      "loss": 25.1732,
+      "step": 2175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.470712762137386e-05,
+      "loss": 0.9654,
+      "step": 2176
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.470134797616673e-05,
+      "loss": 25.0863,
+      "step": 2177
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.469556535364232e-05,
+      "loss": 29.7298,
+      "step": 2178
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.468977975418577e-05,
+      "loss": 0.9555,
+      "step": 2179
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.468399117818243e-05,
+      "loss": 29.2949,
+      "step": 2180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.467819962601785e-05,
+      "loss": 28.5336,
+      "step": 2181
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.467240509807778e-05,
+      "loss": 25.2038,
+      "step": 2182
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.466660759474816e-05,
+      "loss": 27.0847,
+      "step": 2183
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.466080711641512e-05,
+      "loss": 0.211,
+      "step": 2184
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4655003663465e-05,
+      "loss": 0.1759,
+      "step": 2185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.464919723628436e-05,
+      "loss": 25.706,
+      "step": 2186
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46433878352599e-05,
+      "loss": 0.9925,
+      "step": 2187
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.463757546077858e-05,
+      "loss": 28.6422,
+      "step": 2188
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.463176011322752e-05,
+      "loss": 0.9609,
+      "step": 2189
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.462594179299406e-05,
+      "loss": 0.9851,
+      "step": 2190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.462012050046571e-05,
+      "loss": 0.9358,
+      "step": 2191
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.461429623603023e-05,
+      "loss": 0.9351,
+      "step": 2192
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.460846900007548e-05,
+      "loss": 25.0324,
+      "step": 2193
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.460263879298963e-05,
+      "loss": 27.314,
+      "step": 2194
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4596805615161e-05,
+      "loss": 25.1213,
+      "step": 2195
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.459096946697808e-05,
+      "loss": 28.0674,
+      "step": 2196
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.458513034882959e-05,
+      "loss": 25.7973,
+      "step": 2197
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.457928826110446e-05,
+      "loss": 27.6635,
+      "step": 2198
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.457344320419179e-05,
+      "loss": 0.9675,
+      "step": 2199
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.456759517848089e-05,
+      "loss": 26.8032,
+      "step": 2200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.456174418436126e-05,
+      "loss": 0.9612,
+      "step": 2201
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.45558902222226e-05,
+      "loss": 0.9413,
+      "step": 2202
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.455003329245485e-05,
+      "loss": 27.2933,
+      "step": 2203
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.454417339544805e-05,
+      "loss": 0.2057,
+      "step": 2204
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.453831053159253e-05,
+      "loss": 27.4747,
+      "step": 2205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.453244470127878e-05,
+      "loss": 25.6991,
+      "step": 2206
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.452657590489747e-05,
+      "loss": 26.6938,
+      "step": 2207
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.452070414283953e-05,
+      "loss": 26.6472,
+      "step": 2208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.451482941549601e-05,
+      "loss": 0.9419,
+      "step": 2209
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.450895172325822e-05,
+      "loss": 0.2261,
+      "step": 2210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.450307106651763e-05,
+      "loss": 0.9781,
+      "step": 2211
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.449718744566591e-05,
+      "loss": 0.9649,
+      "step": 2212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.449130086109496e-05,
+      "loss": 0.9397,
+      "step": 2213
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.448541131319684e-05,
+      "loss": 27.0377,
+      "step": 2214
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.447951880236381e-05,
+      "loss": 27.1005,
+      "step": 2215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.447362332898835e-05,
+      "loss": 28.5756,
+      "step": 2216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.446772489346312e-05,
+      "loss": 0.9459,
+      "step": 2217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.4461823496181e-05,
+      "loss": 0.295,
+      "step": 2218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.445591913753505e-05,
+      "loss": 28.4026,
+      "step": 2219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.445001181791849e-05,
+      "loss": 0.2483,
+      "step": 2220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.444410153772481e-05,
+      "loss": 29.4054,
+      "step": 2221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.443818829734765e-05,
+      "loss": 0.9617,
+      "step": 2222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.443227209718088e-05,
+      "loss": 0.2381,
+      "step": 2223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.442635293761852e-05,
+      "loss": 29.3848,
+      "step": 2224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.442043081905482e-05,
+      "loss": 0.9868,
+      "step": 2225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.441450574188422e-05,
+      "loss": 0.2221,
+      "step": 2226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.440857770650138e-05,
+      "loss": 27.8409,
+      "step": 2227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.44026467133011e-05,
+      "loss": 27.2299,
+      "step": 2228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.439671276267842e-05,
+      "loss": 0.9849,
+      "step": 2229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.43907758550286e-05,
+      "loss": 0.9757,
+      "step": 2230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.438483599074703e-05,
+      "loss": 25.981,
+      "step": 2231
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.437889317022934e-05,
+      "loss": 0.1812,
+      "step": 2232
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.437294739387136e-05,
+      "loss": 0.1912,
+      "step": 2233
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.43669986620691e-05,
+      "loss": 29.1381,
+      "step": 2234
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.436104697521877e-05,
+      "loss": 0.9452,
+      "step": 2235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.435509233371678e-05,
+      "loss": 26.5969,
+      "step": 2236
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.434913473795975e-05,
+      "loss": 26.0401,
+      "step": 2237
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.434317418834447e-05,
+      "loss": 0.9852,
+      "step": 2238
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.433721068526793e-05,
+      "loss": 26.076,
+      "step": 2239
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.433124422912735e-05,
+      "loss": 24.0189,
+      "step": 2240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.432527482032012e-05,
+      "loss": 0.9322,
+      "step": 2241
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.431930245924381e-05,
+      "loss": 0.2576,
+      "step": 2242
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.431332714629623e-05,
+      "loss": 27.9483,
+      "step": 2243
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.430734888187535e-05,
+      "loss": 27.447,
+      "step": 2244
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.430136766637939e-05,
+      "loss": 26.1273,
+      "step": 2245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.429538350020665e-05,
+      "loss": 27.2407,
+      "step": 2246
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.428939638375575e-05,
+      "loss": 0.9735,
+      "step": 2247
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.428340631742547e-05,
+      "loss": 23.3378,
+      "step": 2248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.427741330161477e-05,
+      "loss": 1.0119,
+      "step": 2249
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.427141733672278e-05,
+      "loss": 0.9368,
+      "step": 2250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42654184231489e-05,
+      "loss": 29.0527,
+      "step": 2251
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.425941656129267e-05,
+      "loss": 0.199,
+      "step": 2252
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.425341175155383e-05,
+      "loss": 27.238,
+      "step": 2253
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.424740399433234e-05,
+      "loss": 0.9202,
+      "step": 2254
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.424139329002835e-05,
+      "loss": 27.1821,
+      "step": 2255
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.423537963904219e-05,
+      "loss": 0.952,
+      "step": 2256
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.422936304177439e-05,
+      "loss": 28.7868,
+      "step": 2257
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.422334349862571e-05,
+      "loss": 26.4879,
+      "step": 2258
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.421732100999705e-05,
+      "loss": 0.9477,
+      "step": 2259
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.421129557628956e-05,
+      "loss": 28.1438,
+      "step": 2260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.420526719790455e-05,
+      "loss": 25.6615,
+      "step": 2261
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.419923587524354e-05,
+      "loss": 24.4142,
+      "step": 2262
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.419320160870824e-05,
+      "loss": 27.3677,
+      "step": 2263
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.418716439870057e-05,
+      "loss": 26.8825,
+      "step": 2264
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.418112424562262e-05,
+      "loss": 0.9748,
+      "step": 2265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.417508114987672e-05,
+      "loss": 26.8952,
+      "step": 2266
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.416903511186533e-05,
+      "loss": 27.2266,
+      "step": 2267
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.416298613199117e-05,
+      "loss": 0.2114,
+      "step": 2268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.415693421065714e-05,
+      "loss": 25.5866,
+      "step": 2269
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.415087934826629e-05,
+      "loss": 25.0391,
+      "step": 2270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.414482154522193e-05,
+      "loss": 24.3508,
+      "step": 2271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.413876080192753e-05,
+      "loss": 25.5949,
+      "step": 2272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.413269711878677e-05,
+      "loss": 24.9872,
+      "step": 2273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.412663049620352e-05,
+      "loss": 24.2533,
+      "step": 2274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.412056093458182e-05,
+      "loss": 24.9302,
+      "step": 2275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.411448843432597e-05,
+      "loss": 27.0537,
+      "step": 2276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41084129958404e-05,
+      "loss": 26.764,
+      "step": 2277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.410233461952976e-05,
+      "loss": 0.9992,
+      "step": 2278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40962533057989e-05,
+      "loss": 0.9603,
+      "step": 2279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.409016905505288e-05,
+      "loss": 29.0226,
+      "step": 2280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.408408186769694e-05,
+      "loss": 27.3444,
+      "step": 2281
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40779917441365e-05,
+      "loss": 0.2302,
+      "step": 2282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.407189868477719e-05,
+      "loss": 24.5185,
+      "step": 2283
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.406580269002486e-05,
+      "loss": 24.4134,
+      "step": 2284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.405970376028548e-05,
+      "loss": 25.2689,
+      "step": 2285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.405360189596532e-05,
+      "loss": 26.1785,
+      "step": 2286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.404749709747077e-05,
+      "loss": 27.7799,
+      "step": 2287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.404138936520843e-05,
+      "loss": 0.9972,
+      "step": 2288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.403527869958513e-05,
+      "loss": 0.9807,
+      "step": 2289
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.402916510100786e-05,
+      "loss": 24.9009,
+      "step": 2290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.402304856988378e-05,
+      "loss": 29.4918,
+      "step": 2291
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.401692910662032e-05,
+      "loss": 27.092,
+      "step": 2292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.401080671162505e-05,
+      "loss": 0.21,
+      "step": 2293
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.400468138530575e-05,
+      "loss": 24.9454,
+      "step": 2294
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39985531280704e-05,
+      "loss": 26.9531,
+      "step": 2295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.399242194032716e-05,
+      "loss": 25.9761,
+      "step": 2296
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.398628782248441e-05,
+      "loss": 0.9718,
+      "step": 2297
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39801507749507e-05,
+      "loss": 25.6887,
+      "step": 2298
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.397401079813478e-05,
+      "loss": 28.226,
+      "step": 2299
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.396786789244562e-05,
+      "loss": 0.9159,
+      "step": 2300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.396172205829234e-05,
+      "loss": 0.1851,
+      "step": 2301
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.395557329608432e-05,
+      "loss": 27.0009,
+      "step": 2302
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.394942160623105e-05,
+      "loss": 0.2293,
+      "step": 2303
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39432669891423e-05,
+      "loss": 28.6112,
+      "step": 2304
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.393710944522796e-05,
+      "loss": 0.2374,
+      "step": 2305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.393094897489818e-05,
+      "loss": 0.9566,
+      "step": 2306
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.392478557856328e-05,
+      "loss": 29.2273,
+      "step": 2307
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.391861925663374e-05,
+      "loss": 27.0406,
+      "step": 2308
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39124500095203e-05,
+      "loss": 24.9302,
+      "step": 2309
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.390627783763382e-05,
+      "loss": 25.7424,
+      "step": 2310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.390010274138543e-05,
+      "loss": 26.0343,
+      "step": 2311
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.389392472118641e-05,
+      "loss": 0.2387,
+      "step": 2312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.388774377744825e-05,
+      "loss": 27.614,
+      "step": 2313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.388155991058261e-05,
+      "loss": 24.9527,
+      "step": 2314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.387537312100139e-05,
+      "loss": 26.603,
+      "step": 2315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.386918340911663e-05,
+      "loss": 0.2131,
+      "step": 2316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.386299077534063e-05,
+      "loss": 27.4183,
+      "step": 2317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.385679522008582e-05,
+      "loss": 25.0496,
+      "step": 2318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.385059674376487e-05,
+      "loss": 1.0219,
+      "step": 2319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.384439534679059e-05,
+      "loss": 0.2198,
+      "step": 2320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.383819102957607e-05,
+      "loss": 28.1544,
+      "step": 2321
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.383198379253453e-05,
+      "loss": 24.8901,
+      "step": 2322
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.382577363607938e-05,
+      "loss": 0.2322,
+      "step": 2323
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.381956056062429e-05,
+      "loss": 0.2226,
+      "step": 2324
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.381334456658303e-05,
+      "loss": 0.2486,
+      "step": 2325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.380712565436965e-05,
+      "loss": 0.9424,
+      "step": 2326
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.380090382439833e-05,
+      "loss": 28.6269,
+      "step": 2327
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.37946790770835e-05,
+      "loss": 25.6718,
+      "step": 2328
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.378845141283973e-05,
+      "loss": 0.2348,
+      "step": 2329
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.378222083208183e-05,
+      "loss": 29.3665,
+      "step": 2330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.37759873352248e-05,
+      "loss": 0.2513,
+      "step": 2331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.376975092268378e-05,
+      "loss": 0.9375,
+      "step": 2332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.376351159487416e-05,
+      "loss": 0.2131,
+      "step": 2333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.375726935221153e-05,
+      "loss": 0.2414,
+      "step": 2334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.375102419511163e-05,
+      "loss": 0.9026,
+      "step": 2335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.374477612399042e-05,
+      "loss": 27.345,
+      "step": 2336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.373852513926407e-05,
+      "loss": 24.5078,
+      "step": 2337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.373227124134888e-05,
+      "loss": 25.9618,
+      "step": 2338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.372601443066142e-05,
+      "loss": 27.4012,
+      "step": 2339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.371975470761842e-05,
+      "loss": 0.9684,
+      "step": 2340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.371349207263682e-05,
+      "loss": 0.9634,
+      "step": 2341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.370722652613372e-05,
+      "loss": 0.9948,
+      "step": 2342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.370095806852644e-05,
+      "loss": 28.47,
+      "step": 2343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.369468670023249e-05,
+      "loss": 0.9638,
+      "step": 2344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.368841242166957e-05,
+      "loss": 27.9566,
+      "step": 2345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.368213523325556e-05,
+      "loss": 27.0615,
+      "step": 2346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.367585513540858e-05,
+      "loss": 0.2397,
+      "step": 2347
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.36695721285469e-05,
+      "loss": 28.4708,
+      "step": 2348
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3663286213089e-05,
+      "loss": 26.8404,
+      "step": 2349
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.365699738945355e-05,
+      "loss": 1.0015,
+      "step": 2350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.365070565805941e-05,
+      "loss": 25.4509,
+      "step": 2351
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.364441101932566e-05,
+      "loss": 27.1219,
+      "step": 2352
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.363811347367151e-05,
+      "loss": 26.2292,
+      "step": 2353
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.363181302151643e-05,
+      "loss": 28.4672,
+      "step": 2354
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.362550966328007e-05,
+      "loss": 0.9035,
+      "step": 2355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.361920339938226e-05,
+      "loss": 26.7807,
+      "step": 2356
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.361289423024299e-05,
+      "loss": 0.2399,
+      "step": 2357
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.360658215628254e-05,
+      "loss": 24.4142,
+      "step": 2358
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.360026717792128e-05,
+      "loss": 26.4073,
+      "step": 2359
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.359394929557983e-05,
+      "loss": 0.2085,
+      "step": 2360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.358762850967899e-05,
+      "loss": 1.0026,
+      "step": 2361
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.358130482063976e-05,
+      "loss": 0.2391,
+      "step": 2362
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.357497822888331e-05,
+      "loss": 0.9305,
+      "step": 2363
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.356864873483103e-05,
+      "loss": 0.2213,
+      "step": 2364
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.35623163389045e-05,
+      "loss": 26.1566,
+      "step": 2365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.355598104152548e-05,
+      "loss": 0.9855,
+      "step": 2366
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.354964284311593e-05,
+      "loss": 26.6061,
+      "step": 2367
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3543301744098e-05,
+      "loss": 27.8621,
+      "step": 2368
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.353695774489406e-05,
+      "loss": 28.2946,
+      "step": 2369
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.353061084592661e-05,
+      "loss": 25.0786,
+      "step": 2370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.352426104761843e-05,
+      "loss": 27.0942,
+      "step": 2371
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.35179083503924e-05,
+      "loss": 1.0151,
+      "step": 2372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.351155275467164e-05,
+      "loss": 27.6847,
+      "step": 2373
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.350519426087952e-05,
+      "loss": 0.9695,
+      "step": 2374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34988328694395e-05,
+      "loss": 0.9591,
+      "step": 2375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.349246858077527e-05,
+      "loss": 0.1851,
+      "step": 2376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.348610139531075e-05,
+      "loss": 30.511,
+      "step": 2377
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.347973131347e-05,
+      "loss": 0.9431,
+      "step": 2378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.347335833567732e-05,
+      "loss": 25.0035,
+      "step": 2379
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.346698246235714e-05,
+      "loss": 0.9774,
+      "step": 2380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.346060369393417e-05,
+      "loss": 27.8217,
+      "step": 2381
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.345422203083324e-05,
+      "loss": 29.0148,
+      "step": 2382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.344783747347943e-05,
+      "loss": 0.9551,
+      "step": 2383
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.344145002229793e-05,
+      "loss": 0.9379,
+      "step": 2384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34350596777142e-05,
+      "loss": 0.2198,
+      "step": 2385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.342866644015386e-05,
+      "loss": 28.2428,
+      "step": 2386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.342227031004274e-05,
+      "loss": 0.9943,
+      "step": 2387
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.341587128780685e-05,
+      "loss": 26.3571,
+      "step": 2388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.340946937387241e-05,
+      "loss": 0.2381,
+      "step": 2389
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.340306456866576e-05,
+      "loss": 25.8688,
+      "step": 2390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.339665687261357e-05,
+      "loss": 27.5817,
+      "step": 2391
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.339024628614255e-05,
+      "loss": 27.5221,
+      "step": 2392
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.338383280967972e-05,
+      "loss": 0.949,
+      "step": 2393
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.337741644365222e-05,
+      "loss": 26.6921,
+      "step": 2394
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.337099718848743e-05,
+      "loss": 25.1838,
+      "step": 2395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.33645750446129e-05,
+      "loss": 0.9594,
+      "step": 2396
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.335815001245635e-05,
+      "loss": 26.632,
+      "step": 2397
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.335172209244575e-05,
+      "loss": 27.2027,
+      "step": 2398
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.334529128500924e-05,
+      "loss": 26.8529,
+      "step": 2399
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.333885759057509e-05,
+      "loss": 0.2219,
+      "step": 2400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.333242100957184e-05,
+      "loss": 22.4919,
+      "step": 2401
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.33259815424282e-05,
+      "loss": 28.1612,
+      "step": 2402
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.331953918957308e-05,
+      "loss": 0.964,
+      "step": 2403
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.331309395143555e-05,
+      "loss": 0.9931,
+      "step": 2404
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.330664582844489e-05,
+      "loss": 0.2049,
+      "step": 2405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.330019482103058e-05,
+      "loss": 28.5486,
+      "step": 2406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32937409296223e-05,
+      "loss": 0.2419,
+      "step": 2407
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.328728415464991e-05,
+      "loss": 0.2206,
+      "step": 2408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.328082449654344e-05,
+      "loss": 27.5994,
+      "step": 2409
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.327436195573314e-05,
+      "loss": 0.2199,
+      "step": 2410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.326789653264947e-05,
+      "loss": 24.1853,
+      "step": 2411
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.326142822772302e-05,
+      "loss": 25.7996,
+      "step": 2412
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.325495704138463e-05,
+      "loss": 0.224,
+      "step": 2413
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.324848297406531e-05,
+      "loss": 26.7533,
+      "step": 2414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.324200602619626e-05,
+      "loss": 25.3354,
+      "step": 2415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.323552619820888e-05,
+      "loss": 1.0099,
+      "step": 2416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.322904349053474e-05,
+      "loss": 23.5582,
+      "step": 2417
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.322255790360563e-05,
+      "loss": 1.0129,
+      "step": 2418
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.321606943785352e-05,
+      "loss": 24.7168,
+      "step": 2419
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.320957809371058e-05,
+      "loss": 24.8352,
+      "step": 2420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.320308387160916e-05,
+      "loss": 0.2118,
+      "step": 2421
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31965867719818e-05,
+      "loss": 27.2649,
+      "step": 2422
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.319008679526123e-05,
+      "loss": 0.9486,
+      "step": 2423
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.318358394188038e-05,
+      "loss": 28.1429,
+      "step": 2424
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31770782122724e-05,
+      "loss": 26.2369,
+      "step": 2425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.317056960687056e-05,
+      "loss": 24.5214,
+      "step": 2426
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.316405812610838e-05,
+      "loss": 0.22,
+      "step": 2427
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.315754377041956e-05,
+      "loss": 24.0261,
+      "step": 2428
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3151026540238e-05,
+      "loss": 26.4276,
+      "step": 2429
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.314450643599775e-05,
+      "loss": 0.231,
+      "step": 2430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31379834581331e-05,
+      "loss": 0.1865,
+      "step": 2431
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.313145760707848e-05,
+      "loss": 26.0269,
+      "step": 2432
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.312492888326857e-05,
+      "loss": 0.9844,
+      "step": 2433
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31183972871382e-05,
+      "loss": 26.924,
+      "step": 2434
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.311186281912243e-05,
+      "loss": 27.2405,
+      "step": 2435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.310532547965646e-05,
+      "loss": 25.5163,
+      "step": 2436
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.30987852691757e-05,
+      "loss": 26.8633,
+      "step": 2437
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.309224218811579e-05,
+      "loss": 25.2948,
+      "step": 2438
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.30856962369125e-05,
+      "loss": 27.7582,
+      "step": 2439
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.307914741600186e-05,
+      "loss": 0.2222,
+      "step": 2440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.307259572581998e-05,
+      "loss": 26.8572,
+      "step": 2441
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.30660411668033e-05,
+      "loss": 1.0445,
+      "step": 2442
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.305948373938838e-05,
+      "loss": 24.757,
+      "step": 2443
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.305292344401193e-05,
+      "loss": 26.8719,
+      "step": 2444
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.304636028111094e-05,
+      "loss": 0.9639,
+      "step": 2445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.303979425112251e-05,
+      "loss": 0.2359,
+      "step": 2446
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3033225354484e-05,
+      "loss": 25.9984,
+      "step": 2447
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.302665359163292e-05,
+      "loss": 0.2299,
+      "step": 2448
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.302007896300698e-05,
+      "loss": 0.2232,
+      "step": 2449
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.301350146904406e-05,
+      "loss": 0.9787,
+      "step": 2450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.300692111018228e-05,
+      "loss": 0.2282,
+      "step": 2451
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.300033788685991e-05,
+      "loss": 25.9008,
+      "step": 2452
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.299375179951543e-05,
+      "loss": 24.3247,
+      "step": 2453
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.29871628485875e-05,
+      "loss": 0.9675,
+      "step": 2454
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.298057103451497e-05,
+      "loss": 24.4571,
+      "step": 2455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.297397635773688e-05,
+      "loss": 0.9516,
+      "step": 2456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.296737881869249e-05,
+      "loss": 0.9497,
+      "step": 2457
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.29607784178212e-05,
+      "loss": 0.9683,
+      "step": 2458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.295417515556265e-05,
+      "loss": 27.4886,
+      "step": 2459
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.294756903235664e-05,
+      "loss": 27.0124,
+      "step": 2460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.294096004864315e-05,
+      "loss": 0.9613,
+      "step": 2461
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.293434820486239e-05,
+      "loss": 0.2194,
+      "step": 2462
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.292773350145474e-05,
+      "loss": 0.2167,
+      "step": 2463
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.292111593886077e-05,
+      "loss": 0.9712,
+      "step": 2464
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.291449551752122e-05,
+      "loss": 0.9588,
+      "step": 2465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.290787223787706e-05,
+      "loss": 27.3847,
+      "step": 2466
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.290124610036944e-05,
+      "loss": 26.9466,
+      "step": 2467
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.289461710543965e-05,
+      "loss": 27.8116,
+      "step": 2468
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.288798525352925e-05,
+      "loss": 27.308,
+      "step": 2469
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.288135054507996e-05,
+      "loss": 26.0995,
+      "step": 2470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.287471298053365e-05,
+      "loss": 0.216,
+      "step": 2471
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.286807256033242e-05,
+      "loss": 25.4138,
+      "step": 2472
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.286142928491857e-05,
+      "loss": 0.978,
+      "step": 2473
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.285478315473456e-05,
+      "loss": 0.937,
+      "step": 2474
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.284813417022306e-05,
+      "loss": 0.9558,
+      "step": 2475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.284148233182692e-05,
+      "loss": 25.9889,
+      "step": 2476
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.283482763998916e-05,
+      "loss": 26.1304,
+      "step": 2477
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.282817009515306e-05,
+      "loss": 0.2566,
+      "step": 2478
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.282150969776202e-05,
+      "loss": 26.9094,
+      "step": 2479
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.281484644825964e-05,
+      "loss": 0.2165,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.280818034708973e-05,
+      "loss": 26.0873,
+      "step": 2481
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.28015113946963e-05,
+      "loss": 24.5579,
+      "step": 2482
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.279483959152354e-05,
+      "loss": 0.2519,
+      "step": 2483
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.278816493801578e-05,
+      "loss": 0.1946,
+      "step": 2484
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27814874346176e-05,
+      "loss": 0.2108,
+      "step": 2485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27748070817738e-05,
+      "loss": 0.9678,
+      "step": 2486
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.276812387992925e-05,
+      "loss": 23.3424,
+      "step": 2487
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.276143782952913e-05,
+      "loss": 0.2213,
+      "step": 2488
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.275474893101874e-05,
+      "loss": 26.1589,
+      "step": 2489
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.274805718484361e-05,
+      "loss": 24.389,
+      "step": 2490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.274136259144942e-05,
+      "loss": 26.4007,
+      "step": 2491
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.273466515128209e-05,
+      "loss": 26.1638,
+      "step": 2492
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.272796486478766e-05,
+      "loss": 25.483,
+      "step": 2493
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.272126173241243e-05,
+      "loss": 1.0178,
+      "step": 2494
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.271455575460287e-05,
+      "loss": 0.9704,
+      "step": 2495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.270784693180559e-05,
+      "loss": 27.0554,
+      "step": 2496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.270113526446744e-05,
+      "loss": 26.5493,
+      "step": 2497
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.269442075303548e-05,
+      "loss": 0.9647,
+      "step": 2498
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.268770339795691e-05,
+      "loss": 24.7555,
+      "step": 2499
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.268098319967912e-05,
+      "loss": 26.759,
+      "step": 2500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.267426015864972e-05,
+      "loss": 0.2049,
+      "step": 2501
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.266753427531649e-05,
+      "loss": 0.9356,
+      "step": 2502
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.266080555012743e-05,
+      "loss": 0.9357,
+      "step": 2503
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.265407398353067e-05,
+      "loss": 26.7307,
+      "step": 2504
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.264733957597458e-05,
+      "loss": 0.2342,
+      "step": 2505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.26406023279077e-05,
+      "loss": 24.9351,
+      "step": 2506
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.263386223977877e-05,
+      "loss": 0.9274,
+      "step": 2507
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.262711931203671e-05,
+      "loss": 0.9766,
+      "step": 2508
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.262037354513061e-05,
+      "loss": 0.2318,
+      "step": 2509
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.261362493950978e-05,
+      "loss": 0.9844,
+      "step": 2510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.260687349562373e-05,
+      "loss": 26.1847,
+      "step": 2511
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.260011921392211e-05,
+      "loss": 27.789,
+      "step": 2512
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25933620948548e-05,
+      "loss": 0.9528,
+      "step": 2513
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.258660213887184e-05,
+      "loss": 27.1435,
+      "step": 2514
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.257983934642351e-05,
+      "loss": 0.2168,
+      "step": 2515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25730737179602e-05,
+      "loss": 28.3427,
+      "step": 2516
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.256630525393257e-05,
+      "loss": 25.3812,
+      "step": 2517
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25595339547914e-05,
+      "loss": 24.8008,
+      "step": 2518
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25527598209877e-05,
+      "loss": 26.3758,
+      "step": 2519
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.254598285297266e-05,
+      "loss": 26.5564,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.253920305119767e-05,
+      "loss": 27.2614,
+      "step": 2521
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.253242041611429e-05,
+      "loss": 0.9613,
+      "step": 2522
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.252563494817425e-05,
+      "loss": 0.9892,
+      "step": 2523
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.251884664782953e-05,
+      "loss": 24.8868,
+      "step": 2524
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.251205551553226e-05,
+      "loss": 26.0858,
+      "step": 2525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.250526155173472e-05,
+      "loss": 27.8033,
+      "step": 2526
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.249846475688946e-05,
+      "loss": 1.0181,
+      "step": 2527
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.249166513144915e-05,
+      "loss": 25.5942,
+      "step": 2528
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.248486267586673e-05,
+      "loss": 25.433,
+      "step": 2529
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.247805739059521e-05,
+      "loss": 0.9759,
+      "step": 2530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.247124927608788e-05,
+      "loss": 0.9467,
+      "step": 2531
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.246443833279819e-05,
+      "loss": 0.9862,
+      "step": 2532
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.245762456117979e-05,
+      "loss": 0.9935,
+      "step": 2533
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24508079616865e-05,
+      "loss": 26.8222,
+      "step": 2534
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.244398853477235e-05,
+      "loss": 1.036,
+      "step": 2535
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.243716628089152e-05,
+      "loss": 27.8925,
+      "step": 2536
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24303412004984e-05,
+      "loss": 22.9925,
+      "step": 2537
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.242351329404763e-05,
+      "loss": 26.9526,
+      "step": 2538
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.241668256199392e-05,
+      "loss": 24.4151,
+      "step": 2539
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.240984900479225e-05,
+      "loss": 0.2428,
+      "step": 2540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.240301262289777e-05,
+      "loss": 28.2077,
+      "step": 2541
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.239617341676581e-05,
+      "loss": 27.5033,
+      "step": 2542
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.238933138685189e-05,
+      "loss": 23.8984,
+      "step": 2543
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.238248653361172e-05,
+      "loss": 0.1971,
+      "step": 2544
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23756388575012e-05,
+      "loss": 27.9709,
+      "step": 2545
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.236878835897644e-05,
+      "loss": 23.7544,
+      "step": 2546
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.236193503849369e-05,
+      "loss": 22.4157,
+      "step": 2547
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.235507889650941e-05,
+      "loss": 1.0479,
+      "step": 2548
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.234821993348025e-05,
+      "loss": 26.4925,
+      "step": 2549
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.234135814986307e-05,
+      "loss": 27.5096,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.233449354611488e-05,
+      "loss": 0.9693,
+      "step": 2551
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23276261226929e-05,
+      "loss": 0.9829,
+      "step": 2552
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.232075588005455e-05,
+      "loss": 24.9927,
+      "step": 2553
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.231388281865738e-05,
+      "loss": 0.9951,
+      "step": 2554
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23070069389592e-05,
+      "loss": 0.2129,
+      "step": 2555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.230012824141797e-05,
+      "loss": 26.305,
+      "step": 2556
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.22932467264918e-05,
+      "loss": 1.0028,
+      "step": 2557
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.228636239463911e-05,
+      "loss": 0.2403,
+      "step": 2558
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.227947524631838e-05,
+      "loss": 27.4211,
+      "step": 2559
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.227258528198831e-05,
+      "loss": 27.1062,
+      "step": 2560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.226569250210783e-05,
+      "loss": 26.7779,
+      "step": 2561
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.225879690713606e-05,
+      "loss": 0.9833,
+      "step": 2562
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.225189849753222e-05,
+      "loss": 0.9642,
+      "step": 2563
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.224499727375582e-05,
+      "loss": 0.9951,
+      "step": 2564
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.223809323626648e-05,
+      "loss": 26.424,
+      "step": 2565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.223118638552406e-05,
+      "loss": 24.5694,
+      "step": 2566
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.222427672198859e-05,
+      "loss": 0.9762,
+      "step": 2567
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.221736424612027e-05,
+      "loss": 26.7297,
+      "step": 2568
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.221044895837953e-05,
+      "loss": 26.8377,
+      "step": 2569
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.220353085922694e-05,
+      "loss": 25.9272,
+      "step": 2570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.219660994912328e-05,
+      "loss": 24.6319,
+      "step": 2571
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.218968622852952e-05,
+      "loss": 0.9225,
+      "step": 2572
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.218275969790679e-05,
+      "loss": 27.0308,
+      "step": 2573
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.217583035771647e-05,
+      "loss": 28.8346,
+      "step": 2574
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.216889820842005e-05,
+      "loss": 25.3196,
+      "step": 2575
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.216196325047927e-05,
+      "loss": 25.2414,
+      "step": 2576
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.215502548435601e-05,
+      "loss": 26.864,
+      "step": 2577
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.214808491051236e-05,
+      "loss": 0.2066,
+      "step": 2578
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.214114152941063e-05,
+      "loss": 26.7514,
+      "step": 2579
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.213419534151322e-05,
+      "loss": 0.9628,
+      "step": 2580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.21272463472828e-05,
+      "loss": 23.7153,
+      "step": 2581
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.212029454718224e-05,
+      "loss": 0.9513,
+      "step": 2582
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.211333994167453e-05,
+      "loss": 26.2074,
+      "step": 2583
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.210638253122287e-05,
+      "loss": 0.1883,
+      "step": 2584
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.209942231629069e-05,
+      "loss": 27.5802,
+      "step": 2585
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.209245929734156e-05,
+      "loss": 0.9816,
+      "step": 2586
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.208549347483922e-05,
+      "loss": 26.0438,
+      "step": 2587
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.207852484924766e-05,
+      "loss": 23.6181,
+      "step": 2588
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.207155342103101e-05,
+      "loss": 23.3538,
+      "step": 2589
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.206457919065361e-05,
+      "loss": 1.0019,
+      "step": 2590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.205760215857997e-05,
+      "loss": 27.0968,
+      "step": 2591
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.20506223252748e-05,
+      "loss": 25.2336,
+      "step": 2592
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.204363969120295e-05,
+      "loss": 26.5574,
+      "step": 2593
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.203665425682954e-05,
+      "loss": 21.306,
+      "step": 2594
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.202966602261982e-05,
+      "loss": 24.3713,
+      "step": 2595
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.202267498903924e-05,
+      "loss": 26.236,
+      "step": 2596
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.201568115655342e-05,
+      "loss": 24.86,
+      "step": 2597
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.200868452562821e-05,
+      "loss": 22.8015,
+      "step": 2598
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.20016850967296e-05,
+      "loss": 25.855,
+      "step": 2599
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.199468287032378e-05,
+      "loss": 0.9774,
+      "step": 2600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.198767784687712e-05,
+      "loss": 26.8337,
+      "step": 2601
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.198067002685623e-05,
+      "loss": 25.6659,
+      "step": 2602
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.197365941072784e-05,
+      "loss": 0.2413,
+      "step": 2603
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.196664599895888e-05,
+      "loss": 26.5199,
+      "step": 2604
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.195962979201646e-05,
+      "loss": 25.8771,
+      "step": 2605
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.195261079036794e-05,
+      "loss": 0.9687,
+      "step": 2606
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.194558899448081e-05,
+      "loss": 28.0488,
+      "step": 2607
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.193856440482271e-05,
+      "loss": 25.5524,
+      "step": 2608
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.193153702186155e-05,
+      "loss": 0.251,
+      "step": 2609
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.192450684606536e-05,
+      "loss": 0.9528,
+      "step": 2610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.191747387790242e-05,
+      "loss": 28.034,
+      "step": 2611
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.191043811784111e-05,
+      "loss": 25.244,
+      "step": 2612
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.190339956635008e-05,
+      "loss": 0.2084,
+      "step": 2613
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.189635822389812e-05,
+      "loss": 26.0348,
+      "step": 2614
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.188931409095421e-05,
+      "loss": 24.1933,
+      "step": 2615
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.188226716798753e-05,
+      "loss": 0.2836,
+      "step": 2616
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.187521745546745e-05,
+      "loss": 0.9337,
+      "step": 2617
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.186816495386348e-05,
+      "loss": 0.9596,
+      "step": 2618
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.186110966364537e-05,
+      "loss": 0.9234,
+      "step": 2619
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.185405158528305e-05,
+      "loss": 26.3235,
+      "step": 2620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.184699071924658e-05,
+      "loss": 27.4209,
+      "step": 2621
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.183992706600628e-05,
+      "loss": 26.4076,
+      "step": 2622
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.183286062603261e-05,
+      "loss": 26.3733,
+      "step": 2623
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.182579139979624e-05,
+      "loss": 0.2237,
+      "step": 2624
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.181871938776798e-05,
+      "loss": 0.9356,
+      "step": 2625
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.181164459041892e-05,
+      "loss": 25.153,
+      "step": 2626
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.18045670082202e-05,
+      "loss": 0.9306,
+      "step": 2627
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.179748664164327e-05,
+      "loss": 25.0907,
+      "step": 2628
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17904034911597e-05,
+      "loss": 24.4818,
+      "step": 2629
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.178331755724127e-05,
+      "loss": 0.9479,
+      "step": 2630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.177622884035992e-05,
+      "loss": 0.9096,
+      "step": 2631
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.176913734098781e-05,
+      "loss": 0.2423,
+      "step": 2632
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.176204305959726e-05,
+      "loss": 0.1887,
+      "step": 2633
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.175494599666077e-05,
+      "loss": 0.9704,
+      "step": 2634
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.174784615265104e-05,
+      "loss": 25.1945,
+      "step": 2635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.174074352804096e-05,
+      "loss": 0.9642,
+      "step": 2636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17336381233036e-05,
+      "loss": 26.0569,
+      "step": 2637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.172652993891222e-05,
+      "loss": 22.4532,
+      "step": 2638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.171941897534023e-05,
+      "loss": 0.213,
+      "step": 2639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.171230523306128e-05,
+      "loss": 0.9046,
+      "step": 2640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.170518871254916e-05,
+      "loss": 24.2482,
+      "step": 2641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.169806941427789e-05,
+      "loss": 0.9632,
+      "step": 2642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.169094733872161e-05,
+      "loss": 25.0991,
+      "step": 2643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.168382248635472e-05,
+      "loss": 26.9189,
+      "step": 2644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.167669485765174e-05,
+      "loss": 0.2189,
+      "step": 2645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.166956445308743e-05,
+      "loss": 25.3779,
+      "step": 2646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.166243127313667e-05,
+      "loss": 0.9834,
+      "step": 2647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.165529531827461e-05,
+      "loss": 27.987,
+      "step": 2648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.16481565889765e-05,
+      "loss": 22.4268,
+      "step": 2649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.164101508571783e-05,
+      "loss": 24.4454,
+      "step": 2650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.163387080897425e-05,
+      "loss": 22.7591,
+      "step": 2651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.162672375922162e-05,
+      "loss": 25.4569,
+      "step": 2652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.161957393693595e-05,
+      "loss": 0.9624,
+      "step": 2653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.161242134259344e-05,
+      "loss": 0.9168,
+      "step": 2654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.16052659766705e-05,
+      "loss": 0.9648,
+      "step": 2655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.159810783964371e-05,
+      "loss": 26.5612,
+      "step": 2656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.159094693198985e-05,
+      "loss": 24.7517,
+      "step": 2657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.158378325418584e-05,
+      "loss": 27.2584,
+      "step": 2658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.157661680670884e-05,
+      "loss": 0.9216,
+      "step": 2659
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.156944759003616e-05,
+      "loss": 24.9767,
+      "step": 2660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.156227560464528e-05,
+      "loss": 0.9715,
+      "step": 2661
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.155510085101393e-05,
+      "loss": 25.1157,
+      "step": 2662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.154792332961995e-05,
+      "loss": 25.2962,
+      "step": 2663
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.154074304094142e-05,
+      "loss": 0.227,
+      "step": 2664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.153355998545656e-05,
+      "loss": 23.418,
+      "step": 2665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.15263741636438e-05,
+      "loss": 0.9433,
+      "step": 2666
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.151918557598175e-05,
+      "loss": 0.2465,
+      "step": 2667
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.15119942229492e-05,
+      "loss": 26.1934,
+      "step": 2668
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.150480010502513e-05,
+      "loss": 27.5443,
+      "step": 2669
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.149760322268873e-05,
+      "loss": 0.985,
+      "step": 2670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.149040357641929e-05,
+      "loss": 0.953,
+      "step": 2671
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.148320116669637e-05,
+      "loss": 0.2188,
+      "step": 2672
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.14759959939997e-05,
+      "loss": 24.184,
+      "step": 2673
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.146878805880917e-05,
+      "loss": 0.1911,
+      "step": 2674
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.146157736160484e-05,
+      "loss": 0.9278,
+      "step": 2675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.145436390286698e-05,
+      "loss": 0.9399,
+      "step": 2676
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.144714768307605e-05,
+      "loss": 25.5801,
+      "step": 2677
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.143992870271269e-05,
+      "loss": 0.9322,
+      "step": 2678
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.143270696225771e-05,
+      "loss": 0.948,
+      "step": 2679
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.142548246219212e-05,
+      "loss": 1.0155,
+      "step": 2680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.141825520299707e-05,
+      "loss": 25.7194,
+      "step": 2681
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.141102518515398e-05,
+      "loss": 0.2014,
+      "step": 2682
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.140379240914437e-05,
+      "loss": 23.0165,
+      "step": 2683
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.139655687544999e-05,
+      "loss": 22.6706,
+      "step": 2684
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.138931858455276e-05,
+      "loss": 24.5755,
+      "step": 2685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.138207753693477e-05,
+      "loss": 27.2767,
+      "step": 2686
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.137483373307831e-05,
+      "loss": 24.0185,
+      "step": 2687
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.136758717346588e-05,
+      "loss": 26.2642,
+      "step": 2688
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.13603378585801e-05,
+      "loss": 25.648,
+      "step": 2689
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.13530857889038e-05,
+      "loss": 26.574,
+      "step": 2690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.134583096492004e-05,
+      "loss": 0.9346,
+      "step": 2691
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.1338573387112e-05,
+      "loss": 0.2283,
+      "step": 2692
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.133131305596308e-05,
+      "loss": 25.0227,
+      "step": 2693
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.132404997195685e-05,
+      "loss": 1.0043,
+      "step": 2694
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.131678413557704e-05,
+      "loss": 25.7322,
+      "step": 2695
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.130951554730765e-05,
+      "loss": 24.4956,
+      "step": 2696
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.130224420763274e-05,
+      "loss": 27.313,
+      "step": 2697
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.129497011703663e-05,
+      "loss": 23.2223,
+      "step": 2698
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.128769327600382e-05,
+      "loss": 0.9731,
+      "step": 2699
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.128041368501899e-05,
+      "loss": 23.7248,
+      "step": 2700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.127313134456697e-05,
+      "loss": 24.8708,
+      "step": 2701
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.12658462551328e-05,
+      "loss": 23.8482,
+      "step": 2702
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.12585584172017e-05,
+      "loss": 0.9942,
+      "step": 2703
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.125126783125911e-05,
+      "loss": 0.9823,
+      "step": 2704
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.124397449779059e-05,
+      "loss": 0.2579,
+      "step": 2705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.12366784172819e-05,
+      "loss": 27.259,
+      "step": 2706
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.122937959021899e-05,
+      "loss": 0.9916,
+      "step": 2707
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.122207801708802e-05,
+      "loss": 0.9547,
+      "step": 2708
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.12147736983753e-05,
+      "loss": 0.9972,
+      "step": 2709
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.120746663456731e-05,
+      "loss": 23.9987,
+      "step": 2710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.120015682615079e-05,
+      "loss": 26.5544,
+      "step": 2711
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.119284427361255e-05,
+      "loss": 0.9037,
+      "step": 2712
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.118552897743967e-05,
+      "loss": 24.2102,
+      "step": 2713
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.117821093811935e-05,
+      "loss": 0.9482,
+      "step": 2714
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.117089015613905e-05,
+      "loss": 25.5217,
+      "step": 2715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.116356663198633e-05,
+      "loss": 25.9892,
+      "step": 2716
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.115624036614899e-05,
+      "loss": 0.236,
+      "step": 2717
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.1148911359115e-05,
+      "loss": 0.2473,
+      "step": 2718
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.114157961137249e-05,
+      "loss": 0.2208,
+      "step": 2719
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.11342451234098e-05,
+      "loss": 26.0844,
+      "step": 2720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.112690789571543e-05,
+      "loss": 0.9592,
+      "step": 2721
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.111956792877807e-05,
+      "loss": 27.2875,
+      "step": 2722
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.111222522308662e-05,
+      "loss": 25.2629,
+      "step": 2723
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.110487977913011e-05,
+      "loss": 23.6628,
+      "step": 2724
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.109753159739781e-05,
+      "loss": 24.5827,
+      "step": 2725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.109018067837913e-05,
+      "loss": 23.5939,
+      "step": 2726
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.108282702256365e-05,
+      "loss": 26.8598,
+      "step": 2727
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.10754706304412e-05,
+      "loss": 0.9353,
+      "step": 2728
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.106811150250172e-05,
+      "loss": 23.7275,
+      "step": 2729
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.106074963923537e-05,
+      "loss": 0.2362,
+      "step": 2730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.105338504113248e-05,
+      "loss": 23.771,
+      "step": 2731
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.104601770868358e-05,
+      "loss": 24.906,
+      "step": 2732
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.103864764237937e-05,
+      "loss": 25.407,
+      "step": 2733
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.103127484271071e-05,
+      "loss": 0.9288,
+      "step": 2734
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.102389931016868e-05,
+      "loss": 24.0337,
+      "step": 2735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.101652104524452e-05,
+      "loss": 26.7767,
+      "step": 2736
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.100914004842965e-05,
+      "loss": 25.5235,
+      "step": 2737
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.100175632021568e-05,
+      "loss": 23.0435,
+      "step": 2738
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.099436986109442e-05,
+      "loss": 22.8773,
+      "step": 2739
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.098698067155782e-05,
+      "loss": 0.9773,
+      "step": 2740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.097958875209804e-05,
+      "loss": 26.8028,
+      "step": 2741
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.097219410320742e-05,
+      "loss": 23.9469,
+      "step": 2742
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.096479672537848e-05,
+      "loss": 0.2046,
+      "step": 2743
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.095739661910391e-05,
+      "loss": 0.9525,
+      "step": 2744
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.094999378487659e-05,
+      "loss": 21.9568,
+      "step": 2745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09425882231896e-05,
+      "loss": 0.9408,
+      "step": 2746
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.093517993453618e-05,
+      "loss": 0.1614,
+      "step": 2747
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.092776891940974e-05,
+      "loss": 22.173,
+      "step": 2748
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.092035517830391e-05,
+      "loss": 0.2586,
+      "step": 2749
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.091293871171245e-05,
+      "loss": 25.0089,
+      "step": 2750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.090551952012936e-05,
+      "loss": 23.0007,
+      "step": 2751
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.08980976040488e-05,
+      "loss": 25.0409,
+      "step": 2752
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.089067296396506e-05,
+      "loss": 26.4249,
+      "step": 2753
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.088324560037268e-05,
+      "loss": 0.9403,
+      "step": 2754
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.087581551376638e-05,
+      "loss": 0.2063,
+      "step": 2755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.086838270464102e-05,
+      "loss": 26.3819,
+      "step": 2756
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.086094717349164e-05,
+      "loss": 23.4793,
+      "step": 2757
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.08535089208135e-05,
+      "loss": 0.2166,
+      "step": 2758
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.084606794710202e-05,
+      "loss": 24.062,
+      "step": 2759
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.08386242528528e-05,
+      "loss": 0.1998,
+      "step": 2760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.083117783856165e-05,
+      "loss": 0.9174,
+      "step": 2761
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.082372870472451e-05,
+      "loss": 25.0195,
+      "step": 2762
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.081627685183751e-05,
+      "loss": 26.5521,
+      "step": 2763
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.080882228039701e-05,
+      "loss": 25.2947,
+      "step": 2764
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.080136499089951e-05,
+      "loss": 1.0185,
+      "step": 2765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.079390498384169e-05,
+      "loss": 24.3285,
+      "step": 2766
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.078644225972043e-05,
+      "loss": 22.7594,
+      "step": 2767
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.077897681903278e-05,
+      "loss": 25.7081,
+      "step": 2768
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.077150866227597e-05,
+      "loss": 23.301,
+      "step": 2769
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.076403778994743e-05,
+      "loss": 0.2114,
+      "step": 2770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.075656420254475e-05,
+      "loss": 0.2077,
+      "step": 2771
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.074908790056567e-05,
+      "loss": 0.9063,
+      "step": 2772
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.07416088845082e-05,
+      "loss": 25.1989,
+      "step": 2773
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.073412715487044e-05,
+      "loss": 26.3248,
+      "step": 2774
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.072664271215072e-05,
+      "loss": 24.6047,
+      "step": 2775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.071915555684754e-05,
+      "loss": 25.1194,
+      "step": 2776
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.071166568945956e-05,
+      "loss": 26.7606,
+      "step": 2777
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.07041731104857e-05,
+      "loss": 23.2082,
+      "step": 2778
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.069667782042492e-05,
+      "loss": 0.9162,
+      "step": 2779
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06891798197765e-05,
+      "loss": 0.2108,
+      "step": 2780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06816791090398e-05,
+      "loss": 28.5459,
+      "step": 2781
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.067417568871445e-05,
+      "loss": 24.095,
+      "step": 2782
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.066666955930016e-05,
+      "loss": 25.3582,
+      "step": 2783
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.065916072129692e-05,
+      "loss": 25.2376,
+      "step": 2784
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.065164917520483e-05,
+      "loss": 24.1045,
+      "step": 2785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.064413492152417e-05,
+      "loss": 0.9053,
+      "step": 2786
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.063661796075547e-05,
+      "loss": 25.5087,
+      "step": 2787
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.062909829339938e-05,
+      "loss": 21.2925,
+      "step": 2788
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.062157591995674e-05,
+      "loss": 23.0426,
+      "step": 2789
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.061405084092856e-05,
+      "loss": 0.2463,
+      "step": 2790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.060652305681607e-05,
+      "loss": 24.5194,
+      "step": 2791
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.059899256812065e-05,
+      "loss": 24.8137,
+      "step": 2792
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.059145937534385e-05,
+      "loss": 0.9408,
+      "step": 2793
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.058392347898744e-05,
+      "loss": 23.2488,
+      "step": 2794
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.057638487955331e-05,
+      "loss": 25.2415,
+      "step": 2795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.05688435775436e-05,
+      "loss": 0.9483,
+      "step": 2796
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.056129957346058e-05,
+      "loss": 26.7658,
+      "step": 2797
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.055375286780672e-05,
+      "loss": 22.7167,
+      "step": 2798
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.054620346108466e-05,
+      "loss": 24.3883,
+      "step": 2799
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.053865135379724e-05,
+      "loss": 22.9458,
+      "step": 2800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.053109654644745e-05,
+      "loss": 0.185,
+      "step": 2801
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.052353903953847e-05,
+      "loss": 0.949,
+      "step": 2802
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.05159788335737e-05,
+      "loss": 0.9412,
+      "step": 2803
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.050841592905664e-05,
+      "loss": 0.22,
+      "step": 2804
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.050085032649105e-05,
+      "loss": 25.7363,
+      "step": 2805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.049328202638081e-05,
+      "loss": 0.9831,
+      "step": 2806
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.048571102923e-05,
+      "loss": 24.3288,
+      "step": 2807
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.047813733554292e-05,
+      "loss": 0.929,
+      "step": 2808
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.047056094582398e-05,
+      "loss": 0.2434,
+      "step": 2809
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.046298186057782e-05,
+      "loss": 27.2571,
+      "step": 2810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.045540008030923e-05,
+      "loss": 24.4108,
+      "step": 2811
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.04478156055232e-05,
+      "loss": 0.976,
+      "step": 2812
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.044022843672488e-05,
+      "loss": 23.817,
+      "step": 2813
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.043263857441962e-05,
+      "loss": 24.3132,
+      "step": 2814
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.042504601911295e-05,
+      "loss": 0.2259,
+      "step": 2815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.041745077131054e-05,
+      "loss": 0.9593,
+      "step": 2816
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.04098528315183e-05,
+      "loss": 23.5259,
+      "step": 2817
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.040225220024228e-05,
+      "loss": 23.868,
+      "step": 2818
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.03946488779887e-05,
+      "loss": 23.8594,
+      "step": 2819
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.038704286526399e-05,
+      "loss": 23.7226,
+      "step": 2820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.037943416257474e-05,
+      "loss": 25.2437,
+      "step": 2821
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.037182277042775e-05,
+      "loss": 0.9429,
+      "step": 2822
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.036420868932993e-05,
+      "loss": 1.0354,
+      "step": 2823
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.035659191978845e-05,
+      "loss": 25.0854,
+      "step": 2824
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.034897246231061e-05,
+      "loss": 0.2084,
+      "step": 2825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.034135031740389e-05,
+      "loss": 24.0948,
+      "step": 2826
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.033372548557598e-05,
+      "loss": 26.2993,
+      "step": 2827
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.032609796733471e-05,
+      "loss": 0.2336,
+      "step": 2828
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.031846776318813e-05,
+      "loss": 23.2261,
+      "step": 2829
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.031083487364443e-05,
+      "loss": 23.9235,
+      "step": 2830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.0303199299212e-05,
+      "loss": 0.9228,
+      "step": 2831
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.02955610403994e-05,
+      "loss": 0.947,
+      "step": 2832
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.028792009771538e-05,
+      "loss": 0.1966,
+      "step": 2833
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.028027647166888e-05,
+      "loss": 0.9547,
+      "step": 2834
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.027263016276896e-05,
+      "loss": 0.9717,
+      "step": 2835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.026498117152493e-05,
+      "loss": 0.2311,
+      "step": 2836
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.025732949844624e-05,
+      "loss": 23.6711,
+      "step": 2837
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.024967514404253e-05,
+      "loss": 24.0399,
+      "step": 2838
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.02420181088236e-05,
+      "loss": 24.7882,
+      "step": 2839
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.023435839329947e-05,
+      "loss": 24.3999,
+      "step": 2840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.022669599798029e-05,
+      "loss": 0.952,
+      "step": 2841
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.021903092337642e-05,
+      "loss": 27.2723,
+      "step": 2842
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.02113631699984e-05,
+      "loss": 25.9687,
+      "step": 2843
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.020369273835692e-05,
+      "loss": 0.2524,
+      "step": 2844
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.019601962896287e-05,
+      "loss": 0.9887,
+      "step": 2845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.018834384232733e-05,
+      "loss": 0.9674,
+      "step": 2846
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.018066537896153e-05,
+      "loss": 0.9627,
+      "step": 2847
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.017298423937689e-05,
+      "loss": 25.5807,
+      "step": 2848
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.016530042408501e-05,
+      "loss": 0.9386,
+      "step": 2849
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.015761393359767e-05,
+      "loss": 24.3092,
+      "step": 2850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.014992476842683e-05,
+      "loss": 20.3788,
+      "step": 2851
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.014223292908462e-05,
+      "loss": 0.9453,
+      "step": 2852
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.013453841608335e-05,
+      "loss": 1.0153,
+      "step": 2853
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.012684122993552e-05,
+      "loss": 24.7325,
+      "step": 2854
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.011914137115379e-05,
+      "loss": 26.2545,
+      "step": 2855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.011143884025101e-05,
+      "loss": 0.207,
+      "step": 2856
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01037336377402e-05,
+      "loss": 0.9759,
+      "step": 2857
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.009602576413456e-05,
+      "loss": 27.5318,
+      "step": 2858
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.008831521994747e-05,
+      "loss": 25.1044,
+      "step": 2859
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.008060200569251e-05,
+      "loss": 0.9661,
+      "step": 2860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.007288612188338e-05,
+      "loss": 23.4159,
+      "step": 2861
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.006516756903402e-05,
+      "loss": 26.7874,
+      "step": 2862
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.005744634765851e-05,
+      "loss": 0.9148,
+      "step": 2863
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.004972245827112e-05,
+      "loss": 0.9508,
+      "step": 2864
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.004199590138632e-05,
+      "loss": 0.1935,
+      "step": 2865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.003426667751869e-05,
+      "loss": 24.4548,
+      "step": 2866
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.002653478718307e-05,
+      "loss": 24.643,
+      "step": 2867
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.001880023089441e-05,
+      "loss": 23.8604,
+      "step": 2868
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.00110630091679e-05,
+      "loss": 24.2386,
+      "step": 2869
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.000332312251884e-05,
+      "loss": 0.9528,
+      "step": 2870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.999558057146278e-05,
+      "loss": 0.9558,
+      "step": 2871
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.998783535651538e-05,
+      "loss": 0.9681,
+      "step": 2872
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.998008747819253e-05,
+      "loss": 0.1952,
+      "step": 2873
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.997233693701026e-05,
+      "loss": 0.2214,
+      "step": 2874
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.996458373348477e-05,
+      "loss": 0.2243,
+      "step": 2875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99568278681325e-05,
+      "loss": 26.5846,
+      "step": 2876
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.994906934147003e-05,
+      "loss": 0.2489,
+      "step": 2877
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.994130815401408e-05,
+      "loss": 0.9365,
+      "step": 2878
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.993354430628159e-05,
+      "loss": 0.9811,
+      "step": 2879
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.992577779878968e-05,
+      "loss": 23.4001,
+      "step": 2880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.991800863205563e-05,
+      "loss": 24.1785,
+      "step": 2881
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99102368065969e-05,
+      "loss": 23.9459,
+      "step": 2882
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.990246232293113e-05,
+      "loss": 22.8668,
+      "step": 2883
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.989468518157613e-05,
+      "loss": 23.2611,
+      "step": 2884
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.988690538304992e-05,
+      "loss": 22.274,
+      "step": 2885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.987912292787064e-05,
+      "loss": 25.5817,
+      "step": 2886
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.987133781655665e-05,
+      "loss": 0.946,
+      "step": 2887
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.986355004962648e-05,
+      "loss": 0.9738,
+      "step": 2888
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.985575962759882e-05,
+      "loss": 23.3613,
+      "step": 2889
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.984796655099256e-05,
+      "loss": 23.0857,
+      "step": 2890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.984017082032674e-05,
+      "loss": 25.4214,
+      "step": 2891
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.983237243612062e-05,
+      "loss": 0.939,
+      "step": 2892
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.982457139889357e-05,
+      "loss": 25.8541,
+      "step": 2893
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.981676770916521e-05,
+      "loss": 25.5043,
+      "step": 2894
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.980896136745528e-05,
+      "loss": 23.9161,
+      "step": 2895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.980115237428373e-05,
+      "loss": 24.653,
+      "step": 2896
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.979334073017067e-05,
+      "loss": 0.2154,
+      "step": 2897
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.978552643563638e-05,
+      "loss": 20.8083,
+      "step": 2898
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.977770949120136e-05,
+      "loss": 0.93,
+      "step": 2899
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.976988989738622e-05,
+      "loss": 24.4153,
+      "step": 2900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.976206765471181e-05,
+      "loss": 0.9433,
+      "step": 2901
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.975424276369911e-05,
+      "loss": 0.2308,
+      "step": 2902
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97464152248693e-05,
+      "loss": 0.9606,
+      "step": 2903
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.973858503874373e-05,
+      "loss": 23.4738,
+      "step": 2904
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.973075220584393e-05,
+      "loss": 25.1991,
+      "step": 2905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97229167266916e-05,
+      "loss": 23.3087,
+      "step": 2906
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97150786018086e-05,
+      "loss": 25.1236,
+      "step": 2907
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.970723783171702e-05,
+      "loss": 0.9752,
+      "step": 2908
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.969939441693907e-05,
+      "loss": 23.4257,
+      "step": 2909
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.969154835799717e-05,
+      "loss": 26.0011,
+      "step": 2910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.968369965541389e-05,
+      "loss": 25.1799,
+      "step": 2911
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.967584830971201e-05,
+      "loss": 0.2203,
+      "step": 2912
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.966799432141444e-05,
+      "loss": 0.2018,
+      "step": 2913
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.966013769104431e-05,
+      "loss": 0.2288,
+      "step": 2914
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.965227841912489e-05,
+      "loss": 0.2529,
+      "step": 2915
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.964441650617967e-05,
+      "loss": 0.9586,
+      "step": 2916
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.963655195273228e-05,
+      "loss": 26.6722,
+      "step": 2917
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.962868475930653e-05,
+      "loss": 26.398,
+      "step": 2918
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.962081492642642e-05,
+      "loss": 24.9402,
+      "step": 2919
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.961294245461611e-05,
+      "loss": 0.2312,
+      "step": 2920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.960506734439993e-05,
+      "loss": 24.0681,
+      "step": 2921
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.959718959630243e-05,
+      "loss": 0.9679,
+      "step": 2922
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.958930921084829e-05,
+      "loss": 27.1542,
+      "step": 2923
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.958142618856238e-05,
+      "loss": 0.2043,
+      "step": 2924
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.957354052996973e-05,
+      "loss": 0.2073,
+      "step": 2925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.956565223559558e-05,
+      "loss": 0.9872,
+      "step": 2926
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.955776130596534e-05,
+      "loss": 25.2438,
+      "step": 2927
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.954986774160457e-05,
+      "loss": 25.5855,
+      "step": 2928
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.954197154303899e-05,
+      "loss": 0.2115,
+      "step": 2929
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.953407271079455e-05,
+      "loss": 25.0608,
+      "step": 2930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.952617124539736e-05,
+      "loss": 0.9367,
+      "step": 2931
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.951826714737369e-05,
+      "loss": 25.1491,
+      "step": 2932
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.951036041724996e-05,
+      "loss": 0.9694,
+      "step": 2933
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.950245105555281e-05,
+      "loss": 25.8471,
+      "step": 2934
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.949453906280907e-05,
+      "loss": 0.9731,
+      "step": 2935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.948662443954567e-05,
+      "loss": 25.5451,
+      "step": 2936
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.947870718628979e-05,
+      "loss": 21.5567,
+      "step": 2937
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.947078730356874e-05,
+      "loss": 24.1778,
+      "step": 2938
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.946286479191004e-05,
+      "loss": 22.8758,
+      "step": 2939
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.945493965184135e-05,
+      "loss": 22.8441,
+      "step": 2940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.944701188389052e-05,
+      "loss": 0.9772,
+      "step": 2941
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.943908148858558e-05,
+      "loss": 0.1837,
+      "step": 2942
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.943114846645475e-05,
+      "loss": 0.9453,
+      "step": 2943
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.942321281802637e-05,
+      "loss": 21.7612,
+      "step": 2944
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.941527454382902e-05,
+      "loss": 0.1919,
+      "step": 2945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.940733364439142e-05,
+      "loss": 22.4922,
+      "step": 2946
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.939939012024246e-05,
+      "loss": 0.953,
+      "step": 2947
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.939144397191122e-05,
+      "loss": 22.9465,
+      "step": 2948
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.938349519992694e-05,
+      "loss": 23.6251,
+      "step": 2949
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.937554380481908e-05,
+      "loss": 25.6793,
+      "step": 2950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.936758978711722e-05,
+      "loss": 22.6955,
+      "step": 2951
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.935963314735111e-05,
+      "loss": 0.9583,
+      "step": 2952
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.935167388605074e-05,
+      "loss": 23.2679,
+      "step": 2953
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.93437120037462e-05,
+      "loss": 24.8508,
+      "step": 2954
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.933574750096783e-05,
+      "loss": 20.5275,
+      "step": 2955
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.932778037824605e-05,
+      "loss": 23.6144,
+      "step": 2956
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.931981063611154e-05,
+      "loss": 0.2189,
+      "step": 2957
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.931183827509512e-05,
+      "loss": 23.8364,
+      "step": 2958
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.930386329572777e-05,
+      "loss": 24.5109,
+      "step": 2959
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.929588569854068e-05,
+      "loss": 24.1737,
+      "step": 2960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.928790548406518e-05,
+      "loss": 26.0015,
+      "step": 2961
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.927992265283282e-05,
+      "loss": 0.2455,
+      "step": 2962
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.927193720537524e-05,
+      "loss": 25.9473,
+      "step": 2963
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.926394914222436e-05,
+      "loss": 22.9657,
+      "step": 2964
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.92559584639122e-05,
+      "loss": 0.2112,
+      "step": 2965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.924796517097096e-05,
+      "loss": 0.9807,
+      "step": 2966
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.923996926393305e-05,
+      "loss": 23.0818,
+      "step": 2967
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.923197074333104e-05,
+      "loss": 23.6358,
+      "step": 2968
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.922396960969765e-05,
+      "loss": 0.9436,
+      "step": 2969
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.921596586356582e-05,
+      "loss": 22.7578,
+      "step": 2970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.920795950546861e-05,
+      "loss": 23.4354,
+      "step": 2971
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.919995053593928e-05,
+      "loss": 24.6413,
+      "step": 2972
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.919193895551128e-05,
+      "loss": 0.1999,
+      "step": 2973
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.918392476471821e-05,
+      "loss": 0.9488,
+      "step": 2974
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.917590796409387e-05,
+      "loss": 24.53,
+      "step": 2975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.916788855417218e-05,
+      "loss": 24.8073,
+      "step": 2976
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.915986653548732e-05,
+      "loss": 25.1129,
+      "step": 2977
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.915184190857352e-05,
+      "loss": 25.5793,
+      "step": 2978
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.914381467396534e-05,
+      "loss": 21.8768,
+      "step": 2979
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.913578483219739e-05,
+      "loss": 25.1581,
+      "step": 2980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.912775238380449e-05,
+      "loss": 23.5167,
+      "step": 2981
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.911971732932165e-05,
+      "loss": 24.5642,
+      "step": 2982
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.911167966928402e-05,
+      "loss": 25.9342,
+      "step": 2983
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.910363940422698e-05,
+      "loss": 22.0109,
+      "step": 2984
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.909559653468604e-05,
+      "loss": 26.7412,
+      "step": 2985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.908755106119687e-05,
+      "loss": 23.5812,
+      "step": 2986
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.907950298429536e-05,
+      "loss": 0.2075,
+      "step": 2987
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.907145230451754e-05,
+      "loss": 23.4119,
+      "step": 2988
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.906339902239962e-05,
+      "loss": 0.9719,
+      "step": 2989
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.905534313847799e-05,
+      "loss": 0.973,
+      "step": 2990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.904728465328922e-05,
+      "loss": 25.0825,
+      "step": 2991
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.903922356737e-05,
+      "loss": 0.1948,
+      "step": 2992
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.90311598812573e-05,
+      "loss": 23.907,
+      "step": 2993
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.902309359548815e-05,
+      "loss": 25.7151,
+      "step": 2994
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.901502471059982e-05,
+      "loss": 0.9588,
+      "step": 2995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.900695322712972e-05,
+      "loss": 22.3418,
+      "step": 2996
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.899887914561548e-05,
+      "loss": 23.2408,
+      "step": 2997
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.899080246659484e-05,
+      "loss": 0.9466,
+      "step": 2998
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.898272319060577e-05,
+      "loss": 23.3752,
+      "step": 2999
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.897464131818636e-05,
+      "loss": 23.4032,
+      "step": 3000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.896655684987493e-05,
+      "loss": 0.9293,
+      "step": 3001
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.895846978620992e-05,
+      "loss": 25.102,
+      "step": 3002
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.895038012772996e-05,
+      "loss": 22.9732,
+      "step": 3003
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.894228787497389e-05,
+      "loss": 24.7137,
+      "step": 3004
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.893419302848066e-05,
+      "loss": 22.2852,
+      "step": 3005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.892609558878944e-05,
+      "loss": 0.2027,
+      "step": 3006
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.891799555643956e-05,
+      "loss": 0.2424,
+      "step": 3007
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.890989293197052e-05,
+      "loss": 0.2531,
+      "step": 3008
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.890178771592199e-05,
+      "loss": 23.1262,
+      "step": 3009
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.889367990883379e-05,
+      "loss": 0.9425,
+      "step": 3010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.888556951124598e-05,
+      "loss": 21.6765,
+      "step": 3011
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.887745652369872e-05,
+      "loss": 0.9772,
+      "step": 3012
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.886934094673239e-05,
+      "loss": 23.8671,
+      "step": 3013
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.88612227808875e-05,
+      "loss": 0.9406,
+      "step": 3014
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.88531020267048e-05,
+      "loss": 0.2759,
+      "step": 3015
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.884497868472513e-05,
+      "loss": 24.1218,
+      "step": 3016
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.883685275548957e-05,
+      "loss": 25.9088,
+      "step": 3017
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.882872423953932e-05,
+      "loss": 22.7806,
+      "step": 3018
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.88205931374158e-05,
+      "loss": 26.3707,
+      "step": 3019
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.881245944966054e-05,
+      "loss": 0.9226,
+      "step": 3020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.880432317681535e-05,
+      "loss": 24.9631,
+      "step": 3021
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.879618431942207e-05,
+      "loss": 23.9913,
+      "step": 3022
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.878804287802283e-05,
+      "loss": 25.2713,
+      "step": 3023
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.877989885315988e-05,
+      "loss": 0.9663,
+      "step": 3024
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.877175224537564e-05,
+      "loss": 0.2363,
+      "step": 3025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.876360305521273e-05,
+      "loss": 0.9876,
+      "step": 3026
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.87554512832139e-05,
+      "loss": 25.2019,
+      "step": 3027
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.874729692992212e-05,
+      "loss": 25.5641,
+      "step": 3028
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.873913999588048e-05,
+      "loss": 0.9008,
+      "step": 3029
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.87309804816323e-05,
+      "loss": 0.9656,
+      "step": 3030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.872281838772102e-05,
+      "loss": 0.2468,
+      "step": 3031
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.871465371469028e-05,
+      "loss": 21.3806,
+      "step": 3032
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.870648646308389e-05,
+      "loss": 0.2221,
+      "step": 3033
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86983166334458e-05,
+      "loss": 0.1856,
+      "step": 3034
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.869014422632021e-05,
+      "loss": 23.7796,
+      "step": 3035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86819692422514e-05,
+      "loss": 0.2387,
+      "step": 3036
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.867379168178387e-05,
+      "loss": 22.0749,
+      "step": 3037
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.866561154546229e-05,
+      "loss": 23.588,
+      "step": 3038
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86574288338315e-05,
+      "loss": 0.2257,
+      "step": 3039
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.864924354743649e-05,
+      "loss": 25.1268,
+      "step": 3040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.864105568682244e-05,
+      "loss": 22.6158,
+      "step": 3041
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.863286525253472e-05,
+      "loss": 26.4538,
+      "step": 3042
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.862467224511883e-05,
+      "loss": 0.9353,
+      "step": 3043
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.861647666512047e-05,
+      "loss": 0.239,
+      "step": 3044
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86082785130855e-05,
+      "loss": 23.3343,
+      "step": 3045
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.860007778955995e-05,
+      "loss": 23.0179,
+      "step": 3046
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.859187449509005e-05,
+      "loss": 24.753,
+      "step": 3047
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.858366863022214e-05,
+      "loss": 0.9804,
+      "step": 3048
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.85754601955028e-05,
+      "loss": 21.5483,
+      "step": 3049
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.856724919147875e-05,
+      "loss": 22.3393,
+      "step": 3050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.855903561869688e-05,
+      "loss": 0.9536,
+      "step": 3051
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.855081947770421e-05,
+      "loss": 0.9579,
+      "step": 3052
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.854260076904803e-05,
+      "loss": 0.9665,
+      "step": 3053
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.853437949327571e-05,
+      "loss": 23.7308,
+      "step": 3054
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.852615565093483e-05,
+      "loss": 0.9406,
+      "step": 3055
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.851792924257317e-05,
+      "loss": 22.9007,
+      "step": 3056
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.850970026873859e-05,
+      "loss": 24.6483,
+      "step": 3057
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.850146872997921e-05,
+      "loss": 0.1927,
+      "step": 3058
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.849323462684328e-05,
+      "loss": 24.8478,
+      "step": 3059
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.848499795987926e-05,
+      "loss": 25.1944,
+      "step": 3060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84767587296357e-05,
+      "loss": 23.9364,
+      "step": 3061
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84685169366614e-05,
+      "loss": 22.7894,
+      "step": 3062
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84602725815053e-05,
+      "loss": 0.203,
+      "step": 3063
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.845202566471653e-05,
+      "loss": 0.2163,
+      "step": 3064
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.844377618684434e-05,
+      "loss": 25.5041,
+      "step": 3065
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.843552414843818e-05,
+      "loss": 22.0453,
+      "step": 3066
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84272695500477e-05,
+      "loss": 0.2135,
+      "step": 3067
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.841901239222269e-05,
+      "loss": 0.1908,
+      "step": 3068
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84107526755131e-05,
+      "loss": 25.5915,
+      "step": 3069
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.840249040046908e-05,
+      "loss": 21.8048,
+      "step": 3070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.839422556764095e-05,
+      "loss": 0.9119,
+      "step": 3071
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.838595817757914e-05,
+      "loss": 0.9643,
+      "step": 3072
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.837768823083434e-05,
+      "loss": 22.9314,
+      "step": 3073
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.836941572795734e-05,
+      "loss": 0.2825,
+      "step": 3074
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.836114066949914e-05,
+      "loss": 0.1907,
+      "step": 3075
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.83528630560109e-05,
+      "loss": 23.5905,
+      "step": 3076
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.834458288804395e-05,
+      "loss": 0.1933,
+      "step": 3077
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.833630016614976e-05,
+      "loss": 0.9193,
+      "step": 3078
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.832801489088005e-05,
+      "loss": 22.9177,
+      "step": 3079
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.83197270627866e-05,
+      "loss": 0.9729,
+      "step": 3080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.831143668242144e-05,
+      "loss": 0.2259,
+      "step": 3081
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.830314375033677e-05,
+      "loss": 0.9398,
+      "step": 3082
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.829484826708489e-05,
+      "loss": 0.9505,
+      "step": 3083
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.828655023321839e-05,
+      "loss": 0.8912,
+      "step": 3084
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.827824964928989e-05,
+      "loss": 0.2272,
+      "step": 3085
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.826994651585227e-05,
+      "loss": 25.5148,
+      "step": 3086
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.826164083345856e-05,
+      "loss": 21.7549,
+      "step": 3087
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.825333260266195e-05,
+      "loss": 0.2317,
+      "step": 3088
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.824502182401584e-05,
+      "loss": 0.9259,
+      "step": 3089
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.82367084980737e-05,
+      "loss": 0.9659,
+      "step": 3090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.822839262538931e-05,
+      "loss": 24.7218,
+      "step": 3091
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.822007420651651e-05,
+      "loss": 0.2386,
+      "step": 3092
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.821175324200932e-05,
+      "loss": 0.2179,
+      "step": 3093
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.820342973242199e-05,
+      "loss": 0.9712,
+      "step": 3094
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.819510367830892e-05,
+      "loss": 24.738,
+      "step": 3095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.818677508022461e-05,
+      "loss": 23.2177,
+      "step": 3096
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.817844393872383e-05,
+      "loss": 0.999,
+      "step": 3097
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.817011025436144e-05,
+      "loss": 0.2235,
+      "step": 3098
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.816177402769255e-05,
+      "loss": 0.9462,
+      "step": 3099
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.815343525927233e-05,
+      "loss": 24.5216,
+      "step": 3100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.81450939496562e-05,
+      "loss": 0.9927,
+      "step": 3101
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.813675009939977e-05,
+      "loss": 23.8275,
+      "step": 3102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.812840370905873e-05,
+      "loss": 23.3466,
+      "step": 3103
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.812005477918901e-05,
+      "loss": 24.2164,
+      "step": 3104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.81117033103467e-05,
+      "loss": 22.5927,
+      "step": 3105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.810334930308801e-05,
+      "loss": 23.751,
+      "step": 3106
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.809499275796939e-05,
+      "loss": 0.2141,
+      "step": 3107
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80866336755474e-05,
+      "loss": 20.6519,
+      "step": 3108
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.807827205637881e-05,
+      "loss": 0.1958,
+      "step": 3109
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.806990790102056e-05,
+      "loss": 24.5667,
+      "step": 3110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80615412100297e-05,
+      "loss": 22.9302,
+      "step": 3111
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.805317198396351e-05,
+      "loss": 21.1415,
+      "step": 3112
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.804480022337941e-05,
+      "loss": 23.1791,
+      "step": 3113
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.803642592883502e-05,
+      "loss": 25.4163,
+      "step": 3114
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.802804910088809e-05,
+      "loss": 21.4219,
+      "step": 3115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.801966974009656e-05,
+      "loss": 23.0867,
+      "step": 3116
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.801128784701854e-05,
+      "loss": 24.3112,
+      "step": 3117
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.800290342221229e-05,
+      "loss": 0.9616,
+      "step": 3118
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.799451646623627e-05,
+      "loss": 0.197,
+      "step": 3119
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.798612697964907e-05,
+      "loss": 19.7743,
+      "step": 3120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.797773496300947e-05,
+      "loss": 21.7832,
+      "step": 3121
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.796934041687643e-05,
+      "loss": 0.9519,
+      "step": 3122
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.796094334180906e-05,
+      "loss": 0.9559,
+      "step": 3123
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.795254373836666e-05,
+      "loss": 0.2342,
+      "step": 3124
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.794414160710867e-05,
+      "loss": 1.0104,
+      "step": 3125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79357369485947e-05,
+      "loss": 24.8752,
+      "step": 3126
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.792732976338454e-05,
+      "loss": 0.9969,
+      "step": 3127
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.791892005203814e-05,
+      "loss": 0.9444,
+      "step": 3128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.791050781511567e-05,
+      "loss": 26.5002,
+      "step": 3129
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.790209305317739e-05,
+      "loss": 0.9476,
+      "step": 3130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.789367576678375e-05,
+      "loss": 20.8215,
+      "step": 3131
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.788525595649542e-05,
+      "loss": 23.0954,
+      "step": 3132
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.787683362287317e-05,
+      "loss": 22.1524,
+      "step": 3133
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.786840876647798e-05,
+      "loss": 22.9883,
+      "step": 3134
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.785998138787097e-05,
+      "loss": 0.9352,
+      "step": 3135
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.785155148761344e-05,
+      "loss": 24.641,
+      "step": 3136
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.784311906626688e-05,
+      "loss": 25.3774,
+      "step": 3137
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.78346841243929e-05,
+      "loss": 26.2168,
+      "step": 3138
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.782624666255335e-05,
+      "loss": 0.2249,
+      "step": 3139
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.781780668131017e-05,
+      "loss": 25.0881,
+      "step": 3140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.780936418122551e-05,
+      "loss": 24.0699,
+      "step": 3141
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.780091916286169e-05,
+      "loss": 0.9776,
+      "step": 3142
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.779247162678117e-05,
+      "loss": 24.076,
+      "step": 3143
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.778402157354659e-05,
+      "loss": 0.9602,
+      "step": 3144
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.77755690037208e-05,
+      "loss": 0.2408,
+      "step": 3145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.776711391786675e-05,
+      "loss": 25.5372,
+      "step": 3146
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.77586563165476e-05,
+      "loss": 22.879,
+      "step": 3147
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.775019620032665e-05,
+      "loss": 24.4386,
+      "step": 3148
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.774173356976738e-05,
+      "loss": 23.1291,
+      "step": 3149
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.773326842543347e-05,
+      "loss": 22.2411,
+      "step": 3150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.772480076788871e-05,
+      "loss": 22.5117,
+      "step": 3151
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.771633059769711e-05,
+      "loss": 23.132,
+      "step": 3152
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.770785791542282e-05,
+      "loss": 22.7085,
+      "step": 3153
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.769938272163013e-05,
+      "loss": 21.6765,
+      "step": 3154
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.769090501688355e-05,
+      "loss": 0.9735,
+      "step": 3155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.768242480174772e-05,
+      "loss": 0.971,
+      "step": 3156
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.76739420767875e-05,
+      "loss": 22.8683,
+      "step": 3157
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.766545684256784e-05,
+      "loss": 0.9428,
+      "step": 3158
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.76569690996539e-05,
+      "loss": 23.2341,
+      "step": 3159
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.7648478848611e-05,
+      "loss": 23.6283,
+      "step": 3160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.763998609000466e-05,
+      "loss": 22.8357,
+      "step": 3161
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.763149082440052e-05,
+      "loss": 0.9645,
+      "step": 3162
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.762299305236439e-05,
+      "loss": 24.7289,
+      "step": 3163
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.761449277446227e-05,
+      "loss": 22.3829,
+      "step": 3164
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.760598999126034e-05,
+      "loss": 22.9709,
+      "step": 3165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.759748470332489e-05,
+      "loss": 0.2299,
+      "step": 3166
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.758897691122244e-05,
+      "loss": 0.9777,
+      "step": 3167
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.758046661551963e-05,
+      "loss": 25.0907,
+      "step": 3168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.757195381678329e-05,
+      "loss": 22.3981,
+      "step": 3169
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.75634385155804e-05,
+      "loss": 0.2316,
+      "step": 3170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.755492071247815e-05,
+      "loss": 23.8365,
+      "step": 3171
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.754640040804384e-05,
+      "loss": 0.2471,
+      "step": 3172
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.753787760284498e-05,
+      "loss": 0.9429,
+      "step": 3173
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.752935229744921e-05,
+      "loss": 0.2163,
+      "step": 3174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.752082449242436e-05,
+      "loss": 21.0918,
+      "step": 3175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.751229418833842e-05,
+      "loss": 0.2281,
+      "step": 3176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.750376138575955e-05,
+      "loss": 0.8897,
+      "step": 3177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.749522608525608e-05,
+      "loss": 0.9492,
+      "step": 3178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.74866882873965e-05,
+      "loss": 0.2201,
+      "step": 3179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.747814799274948e-05,
+      "loss": 0.1967,
+      "step": 3180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.746960520188379e-05,
+      "loss": 0.9989,
+      "step": 3181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.746105991536847e-05,
+      "loss": 0.273,
+      "step": 3182
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.745251213377266e-05,
+      "loss": 22.3515,
+      "step": 3183
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.74439618576657e-05,
+      "loss": 0.2173,
+      "step": 3184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.743540908761705e-05,
+      "loss": 0.9012,
+      "step": 3185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.742685382419638e-05,
+      "loss": 0.9364,
+      "step": 3186
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.741829606797349e-05,
+      "loss": 24.2834,
+      "step": 3187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.74097358195184e-05,
+      "loss": 0.2107,
+      "step": 3188
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.740117307940123e-05,
+      "loss": 23.2066,
+      "step": 3189
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.739260784819233e-05,
+      "loss": 0.9318,
+      "step": 3190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.738404012646215e-05,
+      "loss": 23.6982,
+      "step": 3191
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.737546991478139e-05,
+      "loss": 0.2021,
+      "step": 3192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.73668972137208e-05,
+      "loss": 0.9324,
+      "step": 3193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.73583220238514e-05,
+      "loss": 20.4601,
+      "step": 3194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.734974434574435e-05,
+      "loss": 25.5298,
+      "step": 3195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.734116417997092e-05,
+      "loss": 0.98,
+      "step": 3196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.733258152710262e-05,
+      "loss": 0.9611,
+      "step": 3197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.73239963877111e-05,
+      "loss": 19.7803,
+      "step": 3198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.731540876236815e-05,
+      "loss": 0.9385,
+      "step": 3199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.730681865164576e-05,
+      "loss": 22.9158,
+      "step": 3200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.729822605611608e-05,
+      "loss": 0.9162,
+      "step": 3201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.728963097635138e-05,
+      "loss": 0.9356,
+      "step": 3202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.728103341292417e-05,
+      "loss": 0.9567,
+      "step": 3203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.727243336640705e-05,
+      "loss": 24.3855,
+      "step": 3204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.726383083737287e-05,
+      "loss": 0.968,
+      "step": 3205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.725522582639456e-05,
+      "loss": 22.7635,
+      "step": 3206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.724661833404527e-05,
+      "loss": 24.96,
+      "step": 3207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.72380083608983e-05,
+      "loss": 0.2603,
+      "step": 3208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.722939590752711e-05,
+      "loss": 22.6261,
+      "step": 3209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.722078097450533e-05,
+      "loss": 26.4881,
+      "step": 3210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.721216356240676e-05,
+      "loss": 1.0091,
+      "step": 3211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.720354367180534e-05,
+      "loss": 1.0185,
+      "step": 3212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.719492130327523e-05,
+      "loss": 25.1844,
+      "step": 3213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.718629645739067e-05,
+      "loss": 26.6365,
+      "step": 3214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.717766913472615e-05,
+      "loss": 0.9656,
+      "step": 3215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.71690393358563e-05,
+      "loss": 23.725,
+      "step": 3216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.716040706135588e-05,
+      "loss": 0.21,
+      "step": 3217
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.715177231179985e-05,
+      "loss": 24.128,
+      "step": 3218
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.714313508776332e-05,
+      "loss": 23.5424,
+      "step": 3219
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.713449538982158e-05,
+      "loss": 0.9529,
+      "step": 3220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.712585321855006e-05,
+      "loss": 25.9223,
+      "step": 3221
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.711720857452436e-05,
+      "loss": 0.941,
+      "step": 3222
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.710856145832029e-05,
+      "loss": 23.6967,
+      "step": 3223
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.709991187051374e-05,
+      "loss": 22.5473,
+      "step": 3224
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.709125981168086e-05,
+      "loss": 0.2083,
+      "step": 3225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.708260528239788e-05,
+      "loss": 21.6956,
+      "step": 3226
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.707394828324126e-05,
+      "loss": 23.0418,
+      "step": 3227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.706528881478758e-05,
+      "loss": 0.2258,
+      "step": 3228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.70566268776136e-05,
+      "loss": 0.9412,
+      "step": 3229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.704796247229623e-05,
+      "loss": 23.9886,
+      "step": 3230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.703929559941261e-05,
+      "loss": 0.2394,
+      "step": 3231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.703062625953995e-05,
+      "loss": 22.0633,
+      "step": 3232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.70219544532557e-05,
+      "loss": 0.9492,
+      "step": 3233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.70132801811374e-05,
+      "loss": 20.2631,
+      "step": 3234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.700460344376282e-05,
+      "loss": 0.1775,
+      "step": 3235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.699592424170987e-05,
+      "loss": 25.5465,
+      "step": 3236
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.698724257555666e-05,
+      "loss": 0.2161,
+      "step": 3237
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.697855844588137e-05,
+      "loss": 1.0177,
+      "step": 3238
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.696987185326242e-05,
+      "loss": 0.977,
+      "step": 3239
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.696118279827841e-05,
+      "loss": 21.998,
+      "step": 3240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.695249128150805e-05,
+      "loss": 0.9427,
+      "step": 3241
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.69437973035302e-05,
+      "loss": 0.2131,
+      "step": 3242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.693510086492398e-05,
+      "loss": 0.192,
+      "step": 3243
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.692640196626858e-05,
+      "loss": 0.9559,
+      "step": 3244
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.69177006081434e-05,
+      "loss": 21.6411,
+      "step": 3245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.690899679112798e-05,
+      "loss": 0.2046,
+      "step": 3246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.690029051580203e-05,
+      "loss": 23.4523,
+      "step": 3247
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.689158178274544e-05,
+      "loss": 1.0072,
+      "step": 3248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.688287059253826e-05,
+      "loss": 25.5598,
+      "step": 3249
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.687415694576066e-05,
+      "loss": 24.5666,
+      "step": 3250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.686544084299307e-05,
+      "loss": 22.8546,
+      "step": 3251
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.685672228481597e-05,
+      "loss": 0.9487,
+      "step": 3252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.684800127181006e-05,
+      "loss": 0.9243,
+      "step": 3253
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.683927780455623e-05,
+      "loss": 25.8824,
+      "step": 3254
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.683055188363546e-05,
+      "loss": 0.2089,
+      "step": 3255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.682182350962899e-05,
+      "loss": 21.7469,
+      "step": 3256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.681309268311815e-05,
+      "loss": 23.5889,
+      "step": 3257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.680435940468443e-05,
+      "loss": 22.9248,
+      "step": 3258
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.679562367490953e-05,
+      "loss": 21.7018,
+      "step": 3259
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.678688549437529e-05,
+      "loss": 0.2348,
+      "step": 3260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.67781448636637e-05,
+      "loss": 0.9467,
+      "step": 3261
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.676940178335693e-05,
+      "loss": 0.9588,
+      "step": 3262
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.676065625403733e-05,
+      "loss": 23.6455,
+      "step": 3263
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.675190827628736e-05,
+      "loss": 21.7754,
+      "step": 3264
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.674315785068969e-05,
+      "loss": 1.0,
+      "step": 3265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.673440497782717e-05,
+      "loss": 24.4993,
+      "step": 3266
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.672564965828273e-05,
+      "loss": 22.7885,
+      "step": 3267
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.671689189263954e-05,
+      "loss": 0.2186,
+      "step": 3268
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.67081316814809e-05,
+      "loss": 1.0174,
+      "step": 3269
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.669936902539031e-05,
+      "loss": 0.2019,
+      "step": 3270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.669060392495136e-05,
+      "loss": 23.5741,
+      "step": 3271
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.66818363807479e-05,
+      "loss": 22.1753,
+      "step": 3272
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.667306639336382e-05,
+      "loss": 0.215,
+      "step": 3273
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.66642939633833e-05,
+      "loss": 23.6334,
+      "step": 3274
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.665551909139058e-05,
+      "loss": 0.942,
+      "step": 3275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.664674177797015e-05,
+      "loss": 0.2415,
+      "step": 3276
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.66379620237066e-05,
+      "loss": 0.9291,
+      "step": 3277
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.662917982918469e-05,
+      "loss": 22.1241,
+      "step": 3278
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.662039519498937e-05,
+      "loss": 0.2325,
+      "step": 3279
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.661160812170574e-05,
+      "loss": 24.9895,
+      "step": 3280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.660281860991905e-05,
+      "loss": 24.2498,
+      "step": 3281
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.659402666021473e-05,
+      "loss": 0.9517,
+      "step": 3282
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.658523227317836e-05,
+      "loss": 24.5222,
+      "step": 3283
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.657643544939569e-05,
+      "loss": 0.2448,
+      "step": 3284
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.656763618945263e-05,
+      "loss": 25.3646,
+      "step": 3285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.655883449393525e-05,
+      "loss": 23.2157,
+      "step": 3286
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.655003036342976e-05,
+      "loss": 21.6449,
+      "step": 3287
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.654122379852261e-05,
+      "loss": 23.3967,
+      "step": 3288
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.653241479980032e-05,
+      "loss": 0.9135,
+      "step": 3289
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.65236033678496e-05,
+      "loss": 22.7502,
+      "step": 3290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.651478950325737e-05,
+      "loss": 23.1772,
+      "step": 3291
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.650597320661065e-05,
+      "loss": 24.879,
+      "step": 3292
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.649715447849664e-05,
+      "loss": 23.8012,
+      "step": 3293
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.648833331950273e-05,
+      "loss": 23.4158,
+      "step": 3294
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.647950973021642e-05,
+      "loss": 23.4501,
+      "step": 3295
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.647068371122544e-05,
+      "loss": 0.9602,
+      "step": 3296
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.646185526311761e-05,
+      "loss": 0.9656,
+      "step": 3297
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.645302438648097e-05,
+      "loss": 23.9342,
+      "step": 3298
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.644419108190369e-05,
+      "loss": 0.2007,
+      "step": 3299
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.64353553499741e-05,
+      "loss": 0.9842,
+      "step": 3300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.64265171912807e-05,
+      "loss": 24.9557,
+      "step": 3301
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.641767660641218e-05,
+      "loss": 1.0068,
+      "step": 3302
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.640883359595733e-05,
+      "loss": 0.9906,
+      "step": 3303
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.639998816050515e-05,
+      "loss": 25.7727,
+      "step": 3304
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.639114030064479e-05,
+      "loss": 23.6079,
+      "step": 3305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.638229001696555e-05,
+      "loss": 0.9696,
+      "step": 3306
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.637343731005691e-05,
+      "loss": 21.2423,
+      "step": 3307
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.636458218050849e-05,
+      "loss": 24.1218,
+      "step": 3308
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.63557246289101e-05,
+      "loss": 0.9387,
+      "step": 3309
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.634686465585167e-05,
+      "loss": 22.4642,
+      "step": 3310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.633800226192333e-05,
+      "loss": 0.9319,
+      "step": 3311
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.632913744771535e-05,
+      "loss": 0.2211,
+      "step": 3312
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.632027021381819e-05,
+      "loss": 22.3427,
+      "step": 3313
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.631140056082243e-05,
+      "loss": 22.5824,
+      "step": 3314
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.630252848931881e-05,
+      "loss": 21.124,
+      "step": 3315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.62936539998983e-05,
+      "loss": 21.4999,
+      "step": 3316
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.628477709315194e-05,
+      "loss": 24.7797,
+      "step": 3317
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.627589776967098e-05,
+      "loss": 21.3517,
+      "step": 3318
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.626701603004684e-05,
+      "loss": 0.9864,
+      "step": 3319
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.625813187487109e-05,
+      "loss": 23.5035,
+      "step": 3320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.624924530473542e-05,
+      "loss": 0.9373,
+      "step": 3321
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.624035632023176e-05,
+      "loss": 0.9114,
+      "step": 3322
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.623146492195213e-05,
+      "loss": 0.9822,
+      "step": 3323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.622257111048875e-05,
+      "loss": 0.9158,
+      "step": 3324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.621367488643398e-05,
+      "loss": 24.2184,
+      "step": 3325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.620477625038034e-05,
+      "loss": 1.0302,
+      "step": 3326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.619587520292056e-05,
+      "loss": 23.143,
+      "step": 3327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.618697174464747e-05,
+      "loss": 0.183,
+      "step": 3328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.617806587615407e-05,
+      "loss": 22.3402,
+      "step": 3329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.616915759803354e-05,
+      "loss": 22.7114,
+      "step": 3330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.616024691087921e-05,
+      "loss": 0.9787,
+      "step": 3331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.615133381528457e-05,
+      "loss": 22.4246,
+      "step": 3332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.614241831184328e-05,
+      "loss": 0.2079,
+      "step": 3333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.613350040114918e-05,
+      "loss": 21.1239,
+      "step": 3334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.61245800837962e-05,
+      "loss": 20.8983,
+      "step": 3335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.611565736037848e-05,
+      "loss": 23.6607,
+      "step": 3336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.610673223149034e-05,
+      "loss": 0.2539,
+      "step": 3337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.609780469772623e-05,
+      "loss": 21.1018,
+      "step": 3338
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.608887475968075e-05,
+      "loss": 0.2567,
+      "step": 3339
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.607994241794868e-05,
+      "loss": 0.193,
+      "step": 3340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.607100767312498e-05,
+      "loss": 0.957,
+      "step": 3341
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.606207052580468e-05,
+      "loss": 23.365,
+      "step": 3342
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.605313097658312e-05,
+      "loss": 1.0009,
+      "step": 3343
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.604418902605565e-05,
+      "loss": 0.9584,
+      "step": 3344
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.603524467481787e-05,
+      "loss": 21.364,
+      "step": 3345
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.602629792346554e-05,
+      "loss": 21.6422,
+      "step": 3346
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.60173487725945e-05,
+      "loss": 22.6305,
+      "step": 3347
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.600839722280086e-05,
+      "loss": 22.665,
+      "step": 3348
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.59994432746808e-05,
+      "loss": 22.768,
+      "step": 3349
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.599048692883072e-05,
+      "loss": 0.9639,
+      "step": 3350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.598152818584713e-05,
+      "loss": 0.9386,
+      "step": 3351
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.597256704632672e-05,
+      "loss": 23.6713,
+      "step": 3352
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.596360351086637e-05,
+      "loss": 0.2132,
+      "step": 3353
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.595463758006307e-05,
+      "loss": 22.7102,
+      "step": 3354
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.594566925451401e-05,
+      "loss": 26.1552,
+      "step": 3355
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.593669853481651e-05,
+      "loss": 20.5862,
+      "step": 3356
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.592772542156806e-05,
+      "loss": 0.2378,
+      "step": 3357
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.591874991536632e-05,
+      "loss": 21.5313,
+      "step": 3358
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.590977201680912e-05,
+      "loss": 21.4396,
+      "step": 3359
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.590079172649438e-05,
+      "loss": 0.2113,
+      "step": 3360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.589180904502028e-05,
+      "loss": 0.2303,
+      "step": 3361
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.588282397298506e-05,
+      "loss": 0.1949,
+      "step": 3362
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.58738365109872e-05,
+      "loss": 23.1454,
+      "step": 3363
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.586484665962529e-05,
+      "loss": 0.2147,
+      "step": 3364
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.585585441949812e-05,
+      "loss": 24.2793,
+      "step": 3365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.584685979120459e-05,
+      "loss": 23.7764,
+      "step": 3366
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.58378627753438e-05,
+      "loss": 22.3594,
+      "step": 3367
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.582886337251499e-05,
+      "loss": 0.9488,
+      "step": 3368
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.581986158331755e-05,
+      "loss": 23.1302,
+      "step": 3369
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.581085740835104e-05,
+      "loss": 25.776,
+      "step": 3370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.580185084821522e-05,
+      "loss": 0.9399,
+      "step": 3371
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.579284190350991e-05,
+      "loss": 0.9308,
+      "step": 3372
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.578383057483517e-05,
+      "loss": 21.4289,
+      "step": 3373
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.577481686279123e-05,
+      "loss": 22.6766,
+      "step": 3374
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.57658007679784e-05,
+      "loss": 0.9318,
+      "step": 3375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.575678229099722e-05,
+      "loss": 0.2387,
+      "step": 3376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.574776143244833e-05,
+      "loss": 24.1902,
+      "step": 3377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.57387381929326e-05,
+      "loss": 0.9704,
+      "step": 3378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.5729712573051e-05,
+      "loss": 25.3967,
+      "step": 3379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.57206845734047e-05,
+      "loss": 22.3213,
+      "step": 3380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.571165419459497e-05,
+      "loss": 24.0646,
+      "step": 3381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.57026214372233e-05,
+      "loss": 22.1346,
+      "step": 3382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.56935863018913e-05,
+      "loss": 23.9788,
+      "step": 3383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.568454878920076e-05,
+      "loss": 0.201,
+      "step": 3384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.567550889975362e-05,
+      "loss": 22.8307,
+      "step": 3385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.566646663415199e-05,
+      "loss": 0.9233,
+      "step": 3386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.56574219929981e-05,
+      "loss": 0.169,
+      "step": 3387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.564837497689439e-05,
+      "loss": 0.202,
+      "step": 3388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.563932558644341e-05,
+      "loss": 0.2337,
+      "step": 3389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.563027382224792e-05,
+      "loss": 0.2338,
+      "step": 3390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.562121968491079e-05,
+      "loss": 22.4201,
+      "step": 3391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.561216317503507e-05,
+      "loss": 22.9362,
+      "step": 3392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.560310429322396e-05,
+      "loss": 23.0613,
+      "step": 3393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.559404304008084e-05,
+      "loss": 0.1813,
+      "step": 3394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.558497941620923e-05,
+      "loss": 21.2646,
+      "step": 3395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.557591342221278e-05,
+      "loss": 23.8713,
+      "step": 3396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.556684505869537e-05,
+      "loss": 24.3024,
+      "step": 3397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.555777432626097e-05,
+      "loss": 0.2348,
+      "step": 3398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.554870122551373e-05,
+      "loss": 0.9694,
+      "step": 3399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.553962575705796e-05,
+      "loss": 23.0622,
+      "step": 3400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.553054792149816e-05,
+      "loss": 0.9533,
+      "step": 3401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.552146771943891e-05,
+      "loss": 0.9976,
+      "step": 3402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.551238515148504e-05,
+      "loss": 23.404,
+      "step": 3403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.550330021824145e-05,
+      "loss": 26.0709,
+      "step": 3404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.549421292031325e-05,
+      "loss": 23.6677,
+      "step": 3405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.548512325830571e-05,
+      "loss": 0.2383,
+      "step": 3406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.547603123282424e-05,
+      "loss": 0.9358,
+      "step": 3407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.54669368444744e-05,
+      "loss": 0.9505,
+      "step": 3408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.545784009386193e-05,
+      "loss": 24.615,
+      "step": 3409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.54487409815927e-05,
+      "loss": 23.9953,
+      "step": 3410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.543963950827279e-05,
+      "loss": 22.735,
+      "step": 3411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.543053567450836e-05,
+      "loss": 24.7918,
+      "step": 3412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.542142948090576e-05,
+      "loss": 0.2189,
+      "step": 3413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.541232092807155e-05,
+      "loss": 21.5658,
+      "step": 3414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.540321001661237e-05,
+      "loss": 24.3398,
+      "step": 3415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.539409674713507e-05,
+      "loss": 0.9677,
+      "step": 3416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.538498112024662e-05,
+      "loss": 0.2085,
+      "step": 3417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.537586313655418e-05,
+      "loss": 23.8474,
+      "step": 3418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.5366742796665e-05,
+      "loss": 0.1914,
+      "step": 3419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.53576201011866e-05,
+      "loss": 0.9409,
+      "step": 3420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.534849505072656e-05,
+      "loss": 0.9588,
+      "step": 3421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.533936764589267e-05,
+      "loss": 0.9532,
+      "step": 3422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.533023788729283e-05,
+      "loss": 23.4897,
+      "step": 3423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.532110577553515e-05,
+      "loss": 0.8988,
+      "step": 3424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.531197131122786e-05,
+      "loss": 23.0401,
+      "step": 3425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.530283449497936e-05,
+      "loss": 21.8831,
+      "step": 3426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.52936953273982e-05,
+      "loss": 23.948,
+      "step": 3427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.528455380909309e-05,
+      "loss": 0.2464,
+      "step": 3428
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.527540994067292e-05,
+      "loss": 0.9496,
+      "step": 3429
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.526626372274666e-05,
+      "loss": 23.2523,
+      "step": 3430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.525711515592356e-05,
+      "loss": 23.2401,
+      "step": 3431
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.524796424081292e-05,
+      "loss": 22.4645,
+      "step": 3432
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.523881097802424e-05,
+      "loss": 23.0625,
+      "step": 3433
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.522965536816714e-05,
+      "loss": 0.9686,
+      "step": 3434
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.522049741185147e-05,
+      "loss": 0.2181,
+      "step": 3435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.521133710968718e-05,
+      "loss": 0.2545,
+      "step": 3436
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.520217446228438e-05,
+      "loss": 0.1975,
+      "step": 3437
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.519300947025335e-05,
+      "loss": 0.9368,
+      "step": 3438
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.518384213420452e-05,
+      "loss": 0.2626,
+      "step": 3439
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.517467245474847e-05,
+      "loss": 0.9948,
+      "step": 3440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.516550043249595e-05,
+      "loss": 0.2413,
+      "step": 3441
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.515632606805789e-05,
+      "loss": 0.2294,
+      "step": 3442
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.514714936204528e-05,
+      "loss": 0.1992,
+      "step": 3443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.513797031506938e-05,
+      "loss": 0.9455,
+      "step": 3444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.512878892774154e-05,
+      "loss": 26.681,
+      "step": 3445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.511960520067328e-05,
+      "loss": 0.2198,
+      "step": 3446
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.511041913447629e-05,
+      "loss": 0.9648,
+      "step": 3447
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.510123072976239e-05,
+      "loss": 0.2427,
+      "step": 3448
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.50920399871436e-05,
+      "loss": 0.2169,
+      "step": 3449
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.508284690723203e-05,
+      "loss": 20.9484,
+      "step": 3450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.507365149064002e-05,
+      "loss": 0.2207,
+      "step": 3451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.506445373797998e-05,
+      "loss": 0.1978,
+      "step": 3452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.505525364986456e-05,
+      "loss": 0.8955,
+      "step": 3453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.504605122690652e-05,
+      "loss": 24.3433,
+      "step": 3454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.503684646971877e-05,
+      "loss": 22.6903,
+      "step": 3455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.502763937891441e-05,
+      "loss": 23.5429,
+      "step": 3456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.501842995510667e-05,
+      "loss": 0.9785,
+      "step": 3457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.500921819890892e-05,
+      "loss": 0.2159,
+      "step": 3458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.500000411093474e-05,
+      "loss": 0.2245,
+      "step": 3459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.49907876917978e-05,
+      "loss": 23.633,
+      "step": 3460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.498156894211196e-05,
+      "loss": 0.9389,
+      "step": 3461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.497234786249127e-05,
+      "loss": 0.223,
+      "step": 3462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.496312445354985e-05,
+      "loss": 23.6696,
+      "step": 3463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.495389871590203e-05,
+      "loss": 0.2132,
+      "step": 3464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.494467065016232e-05,
+      "loss": 0.2073,
+      "step": 3465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.49354402569453e-05,
+      "loss": 1.0078,
+      "step": 3466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.492620753686579e-05,
+      "loss": 23.8708,
+      "step": 3467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.491697249053875e-05,
+      "loss": 23.3269,
+      "step": 3468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.490773511857923e-05,
+      "loss": 23.9032,
+      "step": 3469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.489849542160253e-05,
+      "loss": 0.9176,
+      "step": 3470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.488925340022403e-05,
+      "loss": 22.943,
+      "step": 3471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.488000905505926e-05,
+      "loss": 22.2129,
+      "step": 3472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4870762386724e-05,
+      "loss": 22.3547,
+      "step": 3473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.486151339583409e-05,
+      "loss": 0.2344,
+      "step": 3474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.485226208300555e-05,
+      "loss": 23.8422,
+      "step": 3475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.484300844885457e-05,
+      "loss": 0.2133,
+      "step": 3476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.483375249399748e-05,
+      "loss": 0.2169,
+      "step": 3477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.482449421905079e-05,
+      "loss": 0.9679,
+      "step": 3478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.481523362463111e-05,
+      "loss": 22.5247,
+      "step": 3479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.480597071135525e-05,
+      "loss": 22.3806,
+      "step": 3480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.479670547984017e-05,
+      "loss": 22.6797,
+      "step": 3481
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.478743793070299e-05,
+      "loss": 22.5641,
+      "step": 3482
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.477816806456093e-05,
+      "loss": 0.9308,
+      "step": 3483
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.476889588203147e-05,
+      "loss": 0.2242,
+      "step": 3484
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.475962138373213e-05,
+      "loss": 23.3565,
+      "step": 3485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.475034457028063e-05,
+      "loss": 0.2384,
+      "step": 3486
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.474106544229487e-05,
+      "loss": 23.4495,
+      "step": 3487
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.47317840003929e-05,
+      "loss": 20.2379,
+      "step": 3488
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.472250024519288e-05,
+      "loss": 21.2818,
+      "step": 3489
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.471321417731315e-05,
+      "loss": 21.0451,
+      "step": 3490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.470392579737221e-05,
+      "loss": 24.2754,
+      "step": 3491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.469463510598874e-05,
+      "loss": 0.2171,
+      "step": 3492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.46853421037815e-05,
+      "loss": 0.2223,
+      "step": 3493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.467604679136944e-05,
+      "loss": 0.9828,
+      "step": 3494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.466674916937173e-05,
+      "loss": 0.2132,
+      "step": 3495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.46574492384076e-05,
+      "loss": 23.8153,
+      "step": 3496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.464814699909643e-05,
+      "loss": 0.9509,
+      "step": 3497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.463884245205786e-05,
+      "loss": 0.9468,
+      "step": 3498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.462953559791158e-05,
+      "loss": 23.4654,
+      "step": 3499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.462022643727747e-05,
+      "loss": 0.219,
+      "step": 3500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.461091497077558e-05,
+      "loss": 23.8377,
+      "step": 3501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.460160119902607e-05,
+      "loss": 24.4471,
+      "step": 3502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.459228512264929e-05,
+      "loss": 21.8007,
+      "step": 3503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.458296674226577e-05,
+      "loss": 25.6815,
+      "step": 3504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.457364605849611e-05,
+      "loss": 23.631,
+      "step": 3505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.456432307196112e-05,
+      "loss": 22.8179,
+      "step": 3506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.455499778328177e-05,
+      "loss": 23.5737,
+      "step": 3507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.454567019307917e-05,
+      "loss": 24.4312,
+      "step": 3508
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.453634030197457e-05,
+      "loss": 22.3171,
+      "step": 3509
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.452700811058937e-05,
+      "loss": 22.9971,
+      "step": 3510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.451767361954516e-05,
+      "loss": 23.348,
+      "step": 3511
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.450833682946366e-05,
+      "loss": 0.2565,
+      "step": 3512
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.449899774096673e-05,
+      "loss": 0.2124,
+      "step": 3513
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.448965635467639e-05,
+      "loss": 20.8482,
+      "step": 3514
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.448031267121485e-05,
+      "loss": 0.9933,
+      "step": 3515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.447096669120442e-05,
+      "loss": 22.1662,
+      "step": 3516
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.446161841526758e-05,
+      "loss": 22.572,
+      "step": 3517
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4452267844027e-05,
+      "loss": 24.2424,
+      "step": 3518
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.444291497810544e-05,
+      "loss": 0.2191,
+      "step": 3519
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.443355981812584e-05,
+      "loss": 22.1783,
+      "step": 3520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.442420236471134e-05,
+      "loss": 0.1855,
+      "step": 3521
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.441484261848514e-05,
+      "loss": 0.2307,
+      "step": 3522
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.440548058007067e-05,
+      "loss": 0.2203,
+      "step": 3523
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.439611625009149e-05,
+      "loss": 23.2446,
+      "step": 3524
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.438674962917129e-05,
+      "loss": 24.012,
+      "step": 3525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.437738071793394e-05,
+      "loss": 0.216,
+      "step": 3526
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.436800951700345e-05,
+      "loss": 24.6977,
+      "step": 3527
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.435863602700398e-05,
+      "loss": 0.9811,
+      "step": 3528
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.434926024855987e-05,
+      "loss": 23.4216,
+      "step": 3529
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.433988218229556e-05,
+      "loss": 22.5637,
+      "step": 3530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43305018288357e-05,
+      "loss": 0.9384,
+      "step": 3531
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.432111918880505e-05,
+      "loss": 24.2744,
+      "step": 3532
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.431173426282852e-05,
+      "loss": 0.9737,
+      "step": 3533
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.430234705153124e-05,
+      "loss": 0.1903,
+      "step": 3534
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.429295755553839e-05,
+      "loss": 0.9778,
+      "step": 3535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.428356577547539e-05,
+      "loss": 22.8264,
+      "step": 3536
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.427417171196775e-05,
+      "loss": 23.2787,
+      "step": 3537
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.426477536564116e-05,
+      "loss": 20.7696,
+      "step": 3538
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.425537673712149e-05,
+      "loss": 21.8976,
+      "step": 3539
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.424597582703471e-05,
+      "loss": 23.343,
+      "step": 3540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.423657263600697e-05,
+      "loss": 0.1994,
+      "step": 3541
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.422716716466456e-05,
+      "loss": 1.0301,
+      "step": 3542
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.421775941363393e-05,
+      "loss": 22.7317,
+      "step": 3543
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.420834938354167e-05,
+      "loss": 0.983,
+      "step": 3544
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.419893707501456e-05,
+      "loss": 21.7646,
+      "step": 3545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.418952248867947e-05,
+      "loss": 1.0043,
+      "step": 3546
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.418010562516348e-05,
+      "loss": 0.9141,
+      "step": 3547
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.41706864850938e-05,
+      "loss": 25.6483,
+      "step": 3548
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.416126506909776e-05,
+      "loss": 23.464,
+      "step": 3549
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.415184137780288e-05,
+      "loss": 23.8897,
+      "step": 3550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.414241541183686e-05,
+      "loss": 0.265,
+      "step": 3551
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.413298717182744e-05,
+      "loss": 23.3072,
+      "step": 3552
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.412355665840265e-05,
+      "loss": 22.8859,
+      "step": 3553
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.411412387219057e-05,
+      "loss": 22.2918,
+      "step": 3554
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.410468881381948e-05,
+      "loss": 23.1059,
+      "step": 3555
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.409525148391781e-05,
+      "loss": 0.2069,
+      "step": 3556
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.40858118831141e-05,
+      "loss": 22.6131,
+      "step": 3557
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.407637001203711e-05,
+      "loss": 1.0373,
+      "step": 3558
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.406692587131568e-05,
+      "loss": 22.7429,
+      "step": 3559
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.405747946157885e-05,
+      "loss": 21.835,
+      "step": 3560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.404803078345577e-05,
+      "loss": 24.0584,
+      "step": 3561
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.403857983757582e-05,
+      "loss": 21.305,
+      "step": 3562
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.40291266245684e-05,
+      "loss": 25.08,
+      "step": 3563
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.401967114506323e-05,
+      "loss": 20.1675,
+      "step": 3564
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.401021339969e-05,
+      "loss": 21.6749,
+      "step": 3565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.400075338907868e-05,
+      "loss": 0.9725,
+      "step": 3566
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.399129111385938e-05,
+      "loss": 23.1019,
+      "step": 3567
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.398182657466228e-05,
+      "loss": 22.5439,
+      "step": 3568
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.397235977211779e-05,
+      "loss": 0.9852,
+      "step": 3569
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.396289070685644e-05,
+      "loss": 0.963,
+      "step": 3570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.395341937950889e-05,
+      "loss": 22.7003,
+      "step": 3571
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.394394579070601e-05,
+      "loss": 24.2865,
+      "step": 3572
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.393446994107877e-05,
+      "loss": 22.8665,
+      "step": 3573
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.39249918312583e-05,
+      "loss": 21.7255,
+      "step": 3574
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.391551146187589e-05,
+      "loss": 23.1392,
+      "step": 3575
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.390602883356297e-05,
+      "loss": 23.5218,
+      "step": 3576
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.389654394695115e-05,
+      "loss": 1.0077,
+      "step": 3577
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.388705680267215e-05,
+      "loss": 21.8474,
+      "step": 3578
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.387756740135784e-05,
+      "loss": 0.1999,
+      "step": 3579
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.38680757436403e-05,
+      "loss": 22.3457,
+      "step": 3580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.38585818301517e-05,
+      "loss": 0.2037,
+      "step": 3581
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.384908566152435e-05,
+      "loss": 21.7795,
+      "step": 3582
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.383958723839078e-05,
+      "loss": 0.9615,
+      "step": 3583
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.383008656138362e-05,
+      "loss": 0.1786,
+      "step": 3584
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.382058363113564e-05,
+      "loss": 21.6802,
+      "step": 3585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.38110784482798e-05,
+      "loss": 0.2609,
+      "step": 3586
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.380157101344916e-05,
+      "loss": 22.2802,
+      "step": 3587
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.379206132727701e-05,
+      "loss": 21.7822,
+      "step": 3588
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.378254939039672e-05,
+      "loss": 0.8629,
+      "step": 3589
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.377303520344178e-05,
+      "loss": 22.9703,
+      "step": 3590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.376351876704594e-05,
+      "loss": 0.9622,
+      "step": 3591
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.3754000081843e-05,
+      "loss": 23.8396,
+      "step": 3592
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.374447914846699e-05,
+      "loss": 24.1204,
+      "step": 3593
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.373495596755201e-05,
+      "loss": 21.8952,
+      "step": 3594
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.372543053973235e-05,
+      "loss": 21.6756,
+      "step": 3595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.371590286564247e-05,
+      "loss": 22.8885,
+      "step": 3596
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.370637294591695e-05,
+      "loss": 0.2199,
+      "step": 3597
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.369684078119051e-05,
+      "loss": 23.2568,
+      "step": 3598
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.368730637209807e-05,
+      "loss": 1.0028,
+      "step": 3599
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.367776971927464e-05,
+      "loss": 21.4466,
+      "step": 3600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.36682308233554e-05,
+      "loss": 0.9587,
+      "step": 3601
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.36586896849757e-05,
+      "loss": 23.4592,
+      "step": 3602
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.364914630477103e-05,
+      "loss": 23.2843,
+      "step": 3603
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.363960068337701e-05,
+      "loss": 21.6204,
+      "step": 3604
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.363005282142942e-05,
+      "loss": 0.1879,
+      "step": 3605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.362050271956421e-05,
+      "loss": 0.9398,
+      "step": 3606
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.361095037841744e-05,
+      "loss": 23.2066,
+      "step": 3607
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.360139579862535e-05,
+      "loss": 0.9547,
+      "step": 3608
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.359183898082431e-05,
+      "loss": 21.7966,
+      "step": 3609
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.358227992565087e-05,
+      "loss": 21.9746,
+      "step": 3610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.35727186337417e-05,
+      "loss": 1.0101,
+      "step": 3611
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.356315510573361e-05,
+      "loss": 22.0277,
+      "step": 3612
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.355358934226358e-05,
+      "loss": 21.8401,
+      "step": 3613
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.354402134396876e-05,
+      "loss": 0.193,
+      "step": 3614
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.353445111148639e-05,
+      "loss": 24.5498,
+      "step": 3615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.35248786454539e-05,
+      "loss": 23.264,
+      "step": 3616
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.351530394650889e-05,
+      "loss": 0.1971,
+      "step": 3617
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.350572701528903e-05,
+      "loss": 19.441,
+      "step": 3618
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.349614785243223e-05,
+      "loss": 19.7439,
+      "step": 3619
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.348656645857649e-05,
+      "loss": 22.5335,
+      "step": 3620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.347698283435996e-05,
+      "loss": 0.2229,
+      "step": 3621
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.346739698042098e-05,
+      "loss": 0.2156,
+      "step": 3622
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.345780889739801e-05,
+      "loss": 0.2066,
+      "step": 3623
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.344821858592966e-05,
+      "loss": 0.1992,
+      "step": 3624
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.343862604665467e-05,
+      "loss": 24.9561,
+      "step": 3625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.342903128021196e-05,
+      "loss": 21.6849,
+      "step": 3626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.341943428724058e-05,
+      "loss": 25.6382,
+      "step": 3627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.340983506837973e-05,
+      "loss": 20.8115,
+      "step": 3628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.340023362426881e-05,
+      "loss": 23.2629,
+      "step": 3629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.339062995554725e-05,
+      "loss": 0.9714,
+      "step": 3630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.338102406285476e-05,
+      "loss": 23.0306,
+      "step": 3631
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.33714159468311e-05,
+      "loss": 25.0707,
+      "step": 3632
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.336180560811619e-05,
+      "loss": 23.27,
+      "step": 3633
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.33521930473502e-05,
+      "loss": 21.4808,
+      "step": 3634
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.334257826517331e-05,
+      "loss": 22.7397,
+      "step": 3635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.333296126222594e-05,
+      "loss": 0.9639,
+      "step": 3636
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.33233420391486e-05,
+      "loss": 0.9796,
+      "step": 3637
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.3313720596582e-05,
+      "loss": 24.6103,
+      "step": 3638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.330409693516697e-05,
+      "loss": 0.9531,
+      "step": 3639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.329447105554447e-05,
+      "loss": 24.1098,
+      "step": 3640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.328484295835565e-05,
+      "loss": 19.6983,
+      "step": 3641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.327521264424177e-05,
+      "loss": 0.1839,
+      "step": 3642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.326558011384426e-05,
+      "loss": 22.5306,
+      "step": 3643
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.32559453678047e-05,
+      "loss": 0.9238,
+      "step": 3644
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.324630840676479e-05,
+      "loss": 0.2184,
+      "step": 3645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.323666923136641e-05,
+      "loss": 21.8798,
+      "step": 3646
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.32270278422516e-05,
+      "loss": 0.2173,
+      "step": 3647
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.321738424006246e-05,
+      "loss": 21.4515,
+      "step": 3648
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.320773842544134e-05,
+      "loss": 0.9961,
+      "step": 3649
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.319809039903068e-05,
+      "loss": 0.2229,
+      "step": 3650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.318844016147311e-05,
+      "loss": 0.917,
+      "step": 3651
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.317878771341134e-05,
+      "loss": 23.9179,
+      "step": 3652
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.31691330554883e-05,
+      "loss": 0.9619,
+      "step": 3653
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.315947618834704e-05,
+      "loss": 0.9389,
+      "step": 3654
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.314981711263072e-05,
+      "loss": 0.1929,
+      "step": 3655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.314015582898269e-05,
+      "loss": 19.6417,
+      "step": 3656
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.313049233804644e-05,
+      "loss": 24.2137,
+      "step": 3657
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.312082664046561e-05,
+      "loss": 22.821,
+      "step": 3658
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.311115873688397e-05,
+      "loss": 0.1736,
+      "step": 3659
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.310148862794547e-05,
+      "loss": 0.2213,
+      "step": 3660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.309181631429416e-05,
+      "loss": 22.5116,
+      "step": 3661
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.308214179657428e-05,
+      "loss": 23.0599,
+      "step": 3662
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.307246507543016e-05,
+      "loss": 0.2204,
+      "step": 3663
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.306278615150636e-05,
+      "loss": 25.155,
+      "step": 3664
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.305310502544749e-05,
+      "loss": 0.9339,
+      "step": 3665
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.304342169789843e-05,
+      "loss": 1.0122,
+      "step": 3666
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.303373616950408e-05,
+      "loss": 0.9833,
+      "step": 3667
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.302404844090955e-05,
+      "loss": 21.9484,
+      "step": 3668
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.301435851276009e-05,
+      "loss": 22.9132,
+      "step": 3669
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.30046663857011e-05,
+      "loss": 23.0819,
+      "step": 3670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.299497206037814e-05,
+      "loss": 23.45,
+      "step": 3671
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.298527553743684e-05,
+      "loss": 0.2007,
+      "step": 3672
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.297557681752309e-05,
+      "loss": 0.9565,
+      "step": 3673
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.296587590128284e-05,
+      "loss": 25.8273,
+      "step": 3674
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.295617278936223e-05,
+      "loss": 21.2306,
+      "step": 3675
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.294646748240754e-05,
+      "loss": 21.6884,
+      "step": 3676
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.293675998106515e-05,
+      "loss": 0.2131,
+      "step": 3677
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.292705028598165e-05,
+      "loss": 21.7506,
+      "step": 3678
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.291733839780376e-05,
+      "loss": 22.625,
+      "step": 3679
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.290762431717833e-05,
+      "loss": 0.9836,
+      "step": 3680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.289790804475235e-05,
+      "loss": 0.9522,
+      "step": 3681
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.2888189581173e-05,
+      "loss": 23.9135,
+      "step": 3682
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.287846892708752e-05,
+      "loss": 20.9358,
+      "step": 3683
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.28687460831434e-05,
+      "loss": 21.9952,
+      "step": 3684
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.285902104998822e-05,
+      "loss": 22.6277,
+      "step": 3685
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.284929382826969e-05,
+      "loss": 23.4297,
+      "step": 3686
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.283956441863571e-05,
+      "loss": 0.9055,
+      "step": 3687
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.28298328217343e-05,
+      "loss": 22.5161,
+      "step": 3688
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.282009903821362e-05,
+      "loss": 0.9431,
+      "step": 3689
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.281036306872198e-05,
+      "loss": 0.9582,
+      "step": 3690
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.280062491390786e-05,
+      "loss": 22.9411,
+      "step": 3691
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.279088457441986e-05,
+      "loss": 22.2989,
+      "step": 3692
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.278114205090673e-05,
+      "loss": 0.2176,
+      "step": 3693
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.277139734401736e-05,
+      "loss": 0.9266,
+      "step": 3694
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.276165045440078e-05,
+      "loss": 22.3253,
+      "step": 3695
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.275190138270624e-05,
+      "loss": 0.1965,
+      "step": 3696
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.2742150129583e-05,
+      "loss": 21.4978,
+      "step": 3697
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.273239669568058e-05,
+      "loss": 1.0084,
+      "step": 3698
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.272264108164858e-05,
+      "loss": 21.0221,
+      "step": 3699
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.27128832881368e-05,
+      "loss": 0.9708,
+      "step": 3700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.270312331579513e-05,
+      "loss": 0.9619,
+      "step": 3701
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.269336116527363e-05,
+      "loss": 21.8298,
+      "step": 3702
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.268359683722251e-05,
+      "loss": 23.2844,
+      "step": 3703
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.267383033229214e-05,
+      "loss": 23.3067,
+      "step": 3704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.266406165113296e-05,
+      "loss": 0.9619,
+      "step": 3705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.265429079439567e-05,
+      "loss": 22.8941,
+      "step": 3706
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.264451776273104e-05,
+      "loss": 0.2038,
+      "step": 3707
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.263474255678995e-05,
+      "loss": 23.5378,
+      "step": 3708
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.262496517722354e-05,
+      "loss": 25.3386,
+      "step": 3709
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.2615185624683e-05,
+      "loss": 0.9442,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.260540389981968e-05,
+      "loss": 21.7587,
+      "step": 3711
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.259562000328511e-05,
+      "loss": 23.6972,
+      "step": 3712
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.258583393573093e-05,
+      "loss": 21.7872,
+      "step": 3713
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.257604569780897e-05,
+      "loss": 0.9468,
+      "step": 3714
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.256625529017112e-05,
+      "loss": 23.4845,
+      "step": 3715
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.25564627134695e-05,
+      "loss": 24.4909,
+      "step": 3716
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.254666796835633e-05,
+      "loss": 0.9685,
+      "step": 3717
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.253687105548399e-05,
+      "loss": 22.243,
+      "step": 3718
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.2527071975505e-05,
+      "loss": 0.9315,
+      "step": 3719
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.251727072907201e-05,
+      "loss": 0.972,
+      "step": 3720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.250746731683786e-05,
+      "loss": 0.9576,
+      "step": 3721
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.249766173945547e-05,
+      "loss": 21.49,
+      "step": 3722
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.248785399757797e-05,
+      "loss": 0.9171,
+      "step": 3723
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.247804409185855e-05,
+      "loss": 0.9251,
+      "step": 3724
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.246823202295065e-05,
+      "loss": 0.9344,
+      "step": 3725
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.245841779150779e-05,
+      "loss": 22.7406,
+      "step": 3726
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.244860139818361e-05,
+      "loss": 0.9469,
+      "step": 3727
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.243878284363193e-05,
+      "loss": 0.9748,
+      "step": 3728
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.242896212850677e-05,
+      "loss": 21.857,
+      "step": 3729
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.241913925346217e-05,
+      "loss": 23.3685,
+      "step": 3730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.240931421915238e-05,
+      "loss": 21.4318,
+      "step": 3731
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.239948702623185e-05,
+      "loss": 23.5592,
+      "step": 3732
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.238965767535507e-05,
+      "loss": 20.2925,
+      "step": 3733
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.237982616717675e-05,
+      "loss": 1.0178,
+      "step": 3734
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.236999250235166e-05,
+      "loss": 22.283,
+      "step": 3735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.236015668153486e-05,
+      "loss": 24.5951,
+      "step": 3736
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.235031870538136e-05,
+      "loss": 22.7202,
+      "step": 3737
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.234047857454646e-05,
+      "loss": 0.2385,
+      "step": 3738
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.233063628968558e-05,
+      "loss": 0.9809,
+      "step": 3739
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.232079185145425e-05,
+      "loss": 0.9485,
+      "step": 3740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.231094526050812e-05,
+      "loss": 0.2156,
+      "step": 3741
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.230109651750307e-05,
+      "loss": 20.8959,
+      "step": 3742
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.229124562309503e-05,
+      "loss": 21.7425,
+      "step": 3743
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.228139257794012e-05,
+      "loss": 22.886,
+      "step": 3744
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.227153738269464e-05,
+      "loss": 21.2541,
+      "step": 3745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.226168003801493e-05,
+      "loss": 23.1789,
+      "step": 3746
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.22518205445576e-05,
+      "loss": 21.296,
+      "step": 3747
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.22419589029793e-05,
+      "loss": 0.9854,
+      "step": 3748
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.223209511393685e-05,
+      "loss": 19.946,
+      "step": 3749
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.222222917808725e-05,
+      "loss": 23.8696,
+      "step": 3750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.22123610960876e-05,
+      "loss": 0.9092,
+      "step": 3751
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.22024908685952e-05,
+      "loss": 0.9691,
+      "step": 3752
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.219261849626739e-05,
+      "loss": 23.3416,
+      "step": 3753
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.218274397976176e-05,
+      "loss": 21.5285,
+      "step": 3754
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.217286731973597e-05,
+      "loss": 0.932,
+      "step": 3755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.216298851684788e-05,
+      "loss": 21.8367,
+      "step": 3756
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.215310757175546e-05,
+      "loss": 21.8952,
+      "step": 3757
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.214322448511679e-05,
+      "loss": 0.972,
+      "step": 3758
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.213333925759017e-05,
+      "loss": 20.8295,
+      "step": 3759
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.2123451889834e-05,
+      "loss": 23.3811,
+      "step": 3760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.21135623825068e-05,
+      "loss": 0.9903,
+      "step": 3761
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.210367073626728e-05,
+      "loss": 0.1959,
+      "step": 3762
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.209377695177425e-05,
+      "loss": 0.1972,
+      "step": 3763
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.208388102968671e-05,
+      "loss": 0.9718,
+      "step": 3764
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.207398297066375e-05,
+      "loss": 0.2133,
+      "step": 3765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.206408277536464e-05,
+      "loss": 0.2015,
+      "step": 3766
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.205418044444874e-05,
+      "loss": 23.2076,
+      "step": 3767
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.204427597857567e-05,
+      "loss": 0.2433,
+      "step": 3768
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.203436937840503e-05,
+      "loss": 22.7395,
+      "step": 3769
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.20244606445967e-05,
+      "loss": 0.2507,
+      "step": 3770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.201454977781063e-05,
+      "loss": 0.9276,
+      "step": 3771
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.200463677870694e-05,
+      "loss": 0.2187,
+      "step": 3772
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.199472164794586e-05,
+      "loss": 23.7298,
+      "step": 3773
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.198480438618779e-05,
+      "loss": 0.9636,
+      "step": 3774
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.197488499409327e-05,
+      "loss": 0.2034,
+      "step": 3775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.196496347232299e-05,
+      "loss": 23.2668,
+      "step": 3776
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.195503982153776e-05,
+      "loss": 23.9613,
+      "step": 3777
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.194511404239853e-05,
+      "loss": 0.238,
+      "step": 3778
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.193518613556641e-05,
+      "loss": 20.6928,
+      "step": 3779
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.192525610170266e-05,
+      "loss": 20.9783,
+      "step": 3780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.191532394146865e-05,
+      "loss": 0.9396,
+      "step": 3781
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.19053896555259e-05,
+      "loss": 0.9997,
+      "step": 3782
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.18954532445361e-05,
+      "loss": 23.7607,
+      "step": 3783
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.188551470916105e-05,
+      "loss": 22.0005,
+      "step": 3784
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.18755740500627e-05,
+      "loss": 25.3579,
+      "step": 3785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.186563126790316e-05,
+      "loss": 22.7992,
+      "step": 3786
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.185568636334465e-05,
+      "loss": 0.9503,
+      "step": 3787
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.184573933704955e-05,
+      "loss": 0.9393,
+      "step": 3788
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.183579018968039e-05,
+      "loss": 20.94,
+      "step": 3789
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.182583892189979e-05,
+      "loss": 23.7936,
+      "step": 3790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.18158855343706e-05,
+      "loss": 21.6533,
+      "step": 3791
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.180593002775576e-05,
+      "loss": 21.0798,
+      "step": 3792
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.17959724027183e-05,
+      "loss": 1.0135,
+      "step": 3793
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.178601265992148e-05,
+      "loss": 22.2534,
+      "step": 3794
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.177605080002868e-05,
+      "loss": 0.2051,
+      "step": 3795
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.176608682370338e-05,
+      "loss": 20.787,
+      "step": 3796
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.175612073160925e-05,
+      "loss": 0.2347,
+      "step": 3797
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.174615252441005e-05,
+      "loss": 23.4683,
+      "step": 3798
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.173618220276972e-05,
+      "loss": 24.6637,
+      "step": 3799
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.172620976735234e-05,
+      "loss": 21.7774,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.17162352188221e-05,
+      "loss": 21.0172,
+      "step": 3801
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.170625855784338e-05,
+      "loss": 0.1944,
+      "step": 3802
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.169627978508065e-05,
+      "loss": 0.9687,
+      "step": 3803
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.168629890119857e-05,
+      "loss": 0.9976,
+      "step": 3804
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.167631590686186e-05,
+      "loss": 23.8121,
+      "step": 3805
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.166633080273549e-05,
+      "loss": 0.2153,
+      "step": 3806
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.165634358948448e-05,
+      "loss": 25.5234,
+      "step": 3807
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.164635426777404e-05,
+      "loss": 0.95,
+      "step": 3808
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.163636283826948e-05,
+      "loss": 21.8026,
+      "step": 3809
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.162636930163632e-05,
+      "loss": 23.2012,
+      "step": 3810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.161637365854016e-05,
+      "loss": 20.6613,
+      "step": 3811
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.160637590964672e-05,
+      "loss": 0.9598,
+      "step": 3812
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.159637605562194e-05,
+      "loss": 21.9445,
+      "step": 3813
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.158637409713184e-05,
+      "loss": 22.0252,
+      "step": 3814
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.157637003484259e-05,
+      "loss": 22.3398,
+      "step": 3815
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.156636386942051e-05,
+      "loss": 21.6338,
+      "step": 3816
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.15563556015321e-05,
+      "loss": 0.9795,
+      "step": 3817
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.154634523184388e-05,
+      "loss": 20.5609,
+      "step": 3818
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.153633276102263e-05,
+      "loss": 21.9638,
+      "step": 3819
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.152631818973524e-05,
+      "loss": 0.975,
+      "step": 3820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.151630151864871e-05,
+      "loss": 0.9732,
+      "step": 3821
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.150628274843018e-05,
+      "loss": 21.92,
+      "step": 3822
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.149626187974697e-05,
+      "loss": 20.9268,
+      "step": 3823
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.148623891326652e-05,
+      "loss": 23.1894,
+      "step": 3824
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.147621384965638e-05,
+      "loss": 0.9821,
+      "step": 3825
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.14661866895843e-05,
+      "loss": 0.1926,
+      "step": 3826
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.145615743371811e-05,
+      "loss": 20.7358,
+      "step": 3827
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.144612608272581e-05,
+      "loss": 23.7556,
+      "step": 3828
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.143609263727555e-05,
+      "loss": 22.5329,
+      "step": 3829
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.142605709803557e-05,
+      "loss": 22.2804,
+      "step": 3830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.14160194656743e-05,
+      "loss": 0.9768,
+      "step": 3831
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.140597974086032e-05,
+      "loss": 20.8251,
+      "step": 3832
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.13959379242623e-05,
+      "loss": 23.2799,
+      "step": 3833
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.138589401654906e-05,
+      "loss": 0.2404,
+      "step": 3834
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.137584801838957e-05,
+      "loss": 24.775,
+      "step": 3835
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.136579993045298e-05,
+      "loss": 0.2176,
+      "step": 3836
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.13557497534085e-05,
+      "loss": 24.5729,
+      "step": 3837
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.134569748792553e-05,
+      "loss": 22.3933,
+      "step": 3838
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.133564313467359e-05,
+      "loss": 22.6126,
+      "step": 3839
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.132558669432237e-05,
+      "loss": 0.9484,
+      "step": 3840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.131552816754165e-05,
+      "loss": 20.4996,
+      "step": 3841
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.130546755500137e-05,
+      "loss": 21.5626,
+      "step": 3842
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.129540485737165e-05,
+      "loss": 0.9261,
+      "step": 3843
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.128534007532268e-05,
+      "loss": 20.7013,
+      "step": 3844
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.127527320952484e-05,
+      "loss": 0.9888,
+      "step": 3845
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.126520426064861e-05,
+      "loss": 24.7798,
+      "step": 3846
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.125513322936464e-05,
+      "loss": 0.1968,
+      "step": 3847
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.124506011634372e-05,
+      "loss": 20.9663,
+      "step": 3848
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.123498492225674e-05,
+      "loss": 0.2161,
+      "step": 3849
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.122490764777478e-05,
+      "loss": 21.5338,
+      "step": 3850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.121482829356901e-05,
+      "loss": 22.3977,
+      "step": 3851
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.120474686031078e-05,
+      "loss": 0.1965,
+      "step": 3852
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.119466334867153e-05,
+      "loss": 21.0943,
+      "step": 3853
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.118457775932293e-05,
+      "loss": 22.6964,
+      "step": 3854
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.117449009293668e-05,
+      "loss": 0.2254,
+      "step": 3855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.116440035018467e-05,
+      "loss": 21.8929,
+      "step": 3856
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.115430853173894e-05,
+      "loss": 22.2661,
+      "step": 3857
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.114421463827161e-05,
+      "loss": 0.9689,
+      "step": 3858
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.113411867045504e-05,
+      "loss": 22.9554,
+      "step": 3859
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.112402062896162e-05,
+      "loss": 0.9666,
+      "step": 3860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.111392051446396e-05,
+      "loss": 19.9993,
+      "step": 3861
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.110381832763475e-05,
+      "loss": 0.9789,
+      "step": 3862
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.109371406914685e-05,
+      "loss": 22.3632,
+      "step": 3863
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.108360773967327e-05,
+      "loss": 0.1878,
+      "step": 3864
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.107349933988711e-05,
+      "loss": 23.1979,
+      "step": 3865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.106338887046165e-05,
+      "loss": 20.6324,
+      "step": 3866
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.105327633207027e-05,
+      "loss": 22.9263,
+      "step": 3867
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.104316172538657e-05,
+      "loss": 24.9495,
+      "step": 3868
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.103304505108418e-05,
+      "loss": 22.1541,
+      "step": 3869
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.102292630983692e-05,
+      "loss": 22.2625,
+      "step": 3870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.101280550231877e-05,
+      "loss": 22.4517,
+      "step": 3871
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.100268262920381e-05,
+      "loss": 0.9441,
+      "step": 3872
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.099255769116625e-05,
+      "loss": 22.59,
+      "step": 3873
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.098243068888049e-05,
+      "loss": 21.6066,
+      "step": 3874
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.097230162302103e-05,
+      "loss": 22.4522,
+      "step": 3875
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.09621704942625e-05,
+      "loss": 23.5419,
+      "step": 3876
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.09520373032797e-05,
+      "loss": 20.5919,
+      "step": 3877
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.09419020507475e-05,
+      "loss": 22.3707,
+      "step": 3878
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.093176473734101e-05,
+      "loss": 22.7828,
+      "step": 3879
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.09216253637354e-05,
+      "loss": 0.2147,
+      "step": 3880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.0911483930606e-05,
+      "loss": 23.3513,
+      "step": 3881
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.090134043862826e-05,
+      "loss": 0.2247,
+      "step": 3882
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.089119488847781e-05,
+      "loss": 20.1184,
+      "step": 3883
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.088104728083039e-05,
+      "loss": 0.2301,
+      "step": 3884
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.087089761636186e-05,
+      "loss": 21.177,
+      "step": 3885
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.086074589574824e-05,
+      "loss": 21.0731,
+      "step": 3886
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.085059211966568e-05,
+      "loss": 20.958,
+      "step": 3887
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.084043628879049e-05,
+      "loss": 22.4167,
+      "step": 3888
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.083027840379906e-05,
+      "loss": 22.0806,
+      "step": 3889
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.082011846536798e-05,
+      "loss": 20.7688,
+      "step": 3890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.080995647417396e-05,
+      "loss": 0.9306,
+      "step": 3891
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.07997924308938e-05,
+      "loss": 21.5807,
+      "step": 3892
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.078962633620449e-05,
+      "loss": 21.5397,
+      "step": 3893
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.077945819078312e-05,
+      "loss": 0.9755,
+      "step": 3894
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.076928799530696e-05,
+      "loss": 20.255,
+      "step": 3895
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.07591157504534e-05,
+      "loss": 0.2632,
+      "step": 3896
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.074894145689993e-05,
+      "loss": 22.6514,
+      "step": 3897
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.073876511532423e-05,
+      "loss": 0.9601,
+      "step": 3898
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.072858672640409e-05,
+      "loss": 24.647,
+      "step": 3899
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.07184062908174e-05,
+      "loss": 0.9864,
+      "step": 3900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.070822380924228e-05,
+      "loss": 20.5898,
+      "step": 3901
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.069803928235689e-05,
+      "loss": 0.1921,
+      "step": 3902
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.068785271083958e-05,
+      "loss": 22.5645,
+      "step": 3903
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.067766409536884e-05,
+      "loss": 0.9569,
+      "step": 3904
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.066747343662324e-05,
+      "loss": 22.6468,
+      "step": 3905
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.065728073528156e-05,
+      "loss": 18.327,
+      "step": 3906
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.064708599202267e-05,
+      "loss": 0.2085,
+      "step": 3907
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.063688920752558e-05,
+      "loss": 22.1137,
+      "step": 3908
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.062669038246945e-05,
+      "loss": 23.585,
+      "step": 3909
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.061648951753358e-05,
+      "loss": 21.0358,
+      "step": 3910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.060628661339737e-05,
+      "loss": 22.475,
+      "step": 3911
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.05960816707404e-05,
+      "loss": 23.222,
+      "step": 3912
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.058587469024234e-05,
+      "loss": 20.8618,
+      "step": 3913
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.057566567258306e-05,
+      "loss": 0.2088,
+      "step": 3914
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.056545461844253e-05,
+      "loss": 21.6989,
+      "step": 3915
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.05552415285008e-05,
+      "loss": 22.1311,
+      "step": 3916
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.054502640343817e-05,
+      "loss": 23.514,
+      "step": 3917
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.053480924393498e-05,
+      "loss": 20.9434,
+      "step": 3918
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.052459005067174e-05,
+      "loss": 0.9182,
+      "step": 3919
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.051436882432912e-05,
+      "loss": 20.2258,
+      "step": 3920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.050414556558789e-05,
+      "loss": 0.1883,
+      "step": 3921
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.049392027512895e-05,
+      "loss": 0.2265,
+      "step": 3922
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.048369295363337e-05,
+      "loss": 0.8905,
+      "step": 3923
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.047346360178232e-05,
+      "loss": 23.052,
+      "step": 3924
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.046323222025713e-05,
+      "loss": 21.4022,
+      "step": 3925
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.045299880973927e-05,
+      "loss": 0.9485,
+      "step": 3926
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.044276337091032e-05,
+      "loss": 24.1587,
+      "step": 3927
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.043252590445203e-05,
+      "loss": 0.9485,
+      "step": 3928
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.042228641104622e-05,
+      "loss": 0.9798,
+      "step": 3929
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.041204489137492e-05,
+      "loss": 20.0279,
+      "step": 3930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.040180134612026e-05,
+      "loss": 0.9411,
+      "step": 3931
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.03915557759645e-05,
+      "loss": 24.1555,
+      "step": 3932
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.038130818159004e-05,
+      "loss": 0.1991,
+      "step": 3933
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.037105856367941e-05,
+      "loss": 0.9776,
+      "step": 3934
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.036080692291532e-05,
+      "loss": 23.958,
+      "step": 3935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.035055325998053e-05,
+      "loss": 20.6943,
+      "step": 3936
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.034029757555802e-05,
+      "loss": 19.9657,
+      "step": 3937
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.033003987033083e-05,
+      "loss": 0.9281,
+      "step": 3938
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.03197801449822e-05,
+      "loss": 25.395,
+      "step": 3939
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.030951840019547e-05,
+      "loss": 23.6928,
+      "step": 3940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.029925463665411e-05,
+      "loss": 0.9198,
+      "step": 3941
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.028898885504174e-05,
+      "loss": 0.9834,
+      "step": 3942
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.027872105604211e-05,
+      "loss": 22.4039,
+      "step": 3943
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.02684512403391e-05,
+      "loss": 0.2049,
+      "step": 3944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.025817940861672e-05,
+      "loss": 20.9956,
+      "step": 3945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.024790556155915e-05,
+      "loss": 0.9816,
+      "step": 3946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.023762969985065e-05,
+      "loss": 22.7797,
+      "step": 3947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.022735182417563e-05,
+      "loss": 22.537,
+      "step": 3948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.021707193521865e-05,
+      "loss": 25.2208,
+      "step": 3949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.020679003366445e-05,
+      "loss": 0.8959,
+      "step": 3950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.019650612019778e-05,
+      "loss": 23.3339,
+      "step": 3951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.018622019550363e-05,
+      "loss": 21.5283,
+      "step": 3952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.01759322602671e-05,
+      "loss": 0.1906,
+      "step": 3953
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.01656423151734e-05,
+      "loss": 21.7805,
+      "step": 3954
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.015535036090788e-05,
+      "loss": 21.673,
+      "step": 3955
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.014505639815604e-05,
+      "loss": 0.9339,
+      "step": 3956
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.013476042760352e-05,
+      "loss": 0.9308,
+      "step": 3957
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.012446244993606e-05,
+      "loss": 22.3751,
+      "step": 3958
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.011416246583955e-05,
+      "loss": 0.9342,
+      "step": 3959
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.010386047600005e-05,
+      "loss": 22.3754,
+      "step": 3960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.009355648110368e-05,
+      "loss": 1.0155,
+      "step": 3961
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.008325048183677e-05,
+      "loss": 22.9098,
+      "step": 3962
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.007294247888571e-05,
+      "loss": 0.2472,
+      "step": 3963
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.006263247293708e-05,
+      "loss": 20.5116,
+      "step": 3964
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.005232046467759e-05,
+      "loss": 24.4034,
+      "step": 3965
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.004200645479403e-05,
+      "loss": 20.7333,
+      "step": 3966
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.003169044397339e-05,
+      "loss": 0.1887,
+      "step": 3967
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.002137243290278e-05,
+      "loss": 0.9837,
+      "step": 3968
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.001105242226936e-05,
+      "loss": 22.7461,
+      "step": 3969
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.000073041276057e-05,
+      "loss": 23.829,
+      "step": 3970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.999040640506384e-05,
+      "loss": 0.2162,
+      "step": 3971
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.998008039986683e-05,
+      "loss": 0.9718,
+      "step": 3972
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.996975239785734e-05,
+      "loss": 0.2167,
+      "step": 3973
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.995942239972316e-05,
+      "loss": 0.9636,
+      "step": 3974
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.994909040615239e-05,
+      "loss": 0.9207,
+      "step": 3975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.993875641783319e-05,
+      "loss": 22.6171,
+      "step": 3976
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.992842043545382e-05,
+      "loss": 0.9496,
+      "step": 3977
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.991808245970272e-05,
+      "loss": 0.2153,
+      "step": 3978
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.990774249126844e-05,
+      "loss": 0.9269,
+      "step": 3979
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.98974005308397e-05,
+      "loss": 21.4296,
+      "step": 3980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.988705657910527e-05,
+      "loss": 0.1942,
+      "step": 3981
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.987671063675415e-05,
+      "loss": 21.4104,
+      "step": 3982
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.98663627044754e-05,
+      "loss": 21.6481,
+      "step": 3983
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.985601278295826e-05,
+      "loss": 0.9778,
+      "step": 3984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.984566087289207e-05,
+      "loss": 0.8975,
+      "step": 3985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.98353069749663e-05,
+      "loss": 21.2088,
+      "step": 3986
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.982495108987062e-05,
+      "loss": 22.7469,
+      "step": 3987
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.981459321829474e-05,
+      "loss": 23.1635,
+      "step": 3988
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.980423336092853e-05,
+      "loss": 0.9637,
+      "step": 3989
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.979387151846204e-05,
+      "loss": 23.0674,
+      "step": 3990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.97835076915854e-05,
+      "loss": 0.2186,
+      "step": 3991
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.977314188098889e-05,
+      "loss": 21.629,
+      "step": 3992
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.976277408736291e-05,
+      "loss": 22.5026,
+      "step": 3993
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.975240431139803e-05,
+      "loss": 23.301,
+      "step": 3994
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.97420325537849e-05,
+      "loss": 0.9155,
+      "step": 3995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.973165881521434e-05,
+      "loss": 24.0572,
+      "step": 3996
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.972128309637728e-05,
+      "loss": 21.3615,
+      "step": 3997
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.971090539796479e-05,
+      "loss": 1.0126,
+      "step": 3998
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.97005257206681e-05,
+      "loss": 21.3293,
+      "step": 3999
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.969014406517852e-05,
+      "loss": 0.9532,
+      "step": 4000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.96797604321875e-05,
+      "loss": 20.6906,
+      "step": 4001
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.966937482238668e-05,
+      "loss": 21.2952,
+      "step": 4002
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.965898723646776e-05,
+      "loss": 23.3453,
+      "step": 4003
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.964859767512262e-05,
+      "loss": 1.0044,
+      "step": 4004
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.963820613904324e-05,
+      "loss": 22.8329,
+      "step": 4005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.962781262892174e-05,
+      "loss": 0.2235,
+      "step": 4006
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.961741714545038e-05,
+      "loss": 20.8127,
+      "step": 4007
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.960701968932156e-05,
+      "loss": 0.2228,
+      "step": 4008
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.95966202612278e-05,
+      "loss": 0.9652,
+      "step": 4009
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.958621886186173e-05,
+      "loss": 0.208,
+      "step": 4010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.957581549191612e-05,
+      "loss": 20.7709,
+      "step": 4011
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.956541015208394e-05,
+      "loss": 0.2407,
+      "step": 4012
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.955500284305818e-05,
+      "loss": 23.1707,
+      "step": 4013
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.954459356553204e-05,
+      "loss": 0.1933,
+      "step": 4014
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.95341823201988e-05,
+      "loss": 0.2363,
+      "step": 4015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.952376910775195e-05,
+      "loss": 22.3784,
+      "step": 4016
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.9513353928885e-05,
+      "loss": 1.0157,
+      "step": 4017
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.950293678429168e-05,
+      "loss": 24.1805,
+      "step": 4018
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.949251767466583e-05,
+      "loss": 0.1616,
+      "step": 4019
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.948209660070138e-05,
+      "loss": 0.9506,
+      "step": 4020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.947167356309244e-05,
+      "loss": 0.2002,
+      "step": 4021
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.946124856253323e-05,
+      "loss": 0.9988,
+      "step": 4022
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.94508215997181e-05,
+      "loss": 0.9724,
+      "step": 4023
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.944039267534154e-05,
+      "loss": 21.6552,
+      "step": 4024
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.942996179009818e-05,
+      "loss": 0.9479,
+      "step": 4025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.941952894468272e-05,
+      "loss": 22.1145,
+      "step": 4026
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.940909413979009e-05,
+      "loss": 0.9656,
+      "step": 4027
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.939865737611525e-05,
+      "loss": 0.2184,
+      "step": 4028
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.938821865435338e-05,
+      "loss": 0.9576,
+      "step": 4029
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.937777797519969e-05,
+      "loss": 21.8259,
+      "step": 4030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.936733533934963e-05,
+      "loss": 0.2009,
+      "step": 4031
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.935689074749873e-05,
+      "loss": 22.6226,
+      "step": 4032
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.93464442003426e-05,
+      "loss": 20.7568,
+      "step": 4033
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.933599569857707e-05,
+      "loss": 0.2264,
+      "step": 4034
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.932554524289805e-05,
+      "loss": 1.0058,
+      "step": 4035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.931509283400158e-05,
+      "loss": 0.2173,
+      "step": 4036
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.930463847258384e-05,
+      "loss": 0.2288,
+      "step": 4037
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.929418215934115e-05,
+      "loss": 0.193,
+      "step": 4038
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.928372389496995e-05,
+      "loss": 0.9607,
+      "step": 4039
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.927326368016677e-05,
+      "loss": 0.2253,
+      "step": 4040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.926280151562836e-05,
+      "loss": 0.172,
+      "step": 4041
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.925233740205154e-05,
+      "loss": 0.2224,
+      "step": 4042
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.924187134013323e-05,
+      "loss": 24.8325,
+      "step": 4043
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.923140333057056e-05,
+      "loss": 0.2232,
+      "step": 4044
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.922093337406073e-05,
+      "loss": 22.0328,
+      "step": 4045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.92104614713011e-05,
+      "loss": 0.1931,
+      "step": 4046
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.919998762298913e-05,
+      "loss": 23.4601,
+      "step": 4047
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.918951182982243e-05,
+      "loss": 0.1846,
+      "step": 4048
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.917903409249874e-05,
+      "loss": 21.301,
+      "step": 4049
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.916855441171594e-05,
+      "loss": 21.5483,
+      "step": 4050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.9158072788172e-05,
+      "loss": 0.2013,
+      "step": 4051
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.914758922256507e-05,
+      "loss": 21.1933,
+      "step": 4052
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.91371037155934e-05,
+      "loss": 23.0427,
+      "step": 4053
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.912661626795534e-05,
+      "loss": 21.2787,
+      "step": 4054
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.911612688034944e-05,
+      "loss": 0.207,
+      "step": 4055
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.910563555347433e-05,
+      "loss": 0.2448,
+      "step": 4056
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.909514228802879e-05,
+      "loss": 0.997,
+      "step": 4057
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.90846470847117e-05,
+      "loss": 22.5651,
+      "step": 4058
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.90741499442221e-05,
+      "loss": 22.0727,
+      "step": 4059
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.906365086725915e-05,
+      "loss": 22.6009,
+      "step": 4060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.905314985452215e-05,
+      "loss": 0.9715,
+      "step": 4061
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.904264690671048e-05,
+      "loss": 21.0982,
+      "step": 4062
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.903214202452372e-05,
+      "loss": 21.4161,
+      "step": 4063
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.902163520866152e-05,
+      "loss": 0.1857,
+      "step": 4064
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.90111264598237e-05,
+      "loss": 21.8837,
+      "step": 4065
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.900061577871019e-05,
+      "loss": 0.2352,
+      "step": 4066
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.899010316602104e-05,
+      "loss": 0.9848,
+      "step": 4067
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.897958862245644e-05,
+      "loss": 21.5296,
+      "step": 4068
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.896907214871674e-05,
+      "loss": 22.6818,
+      "step": 4069
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.895855374550232e-05,
+      "loss": 1.0114,
+      "step": 4070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.894803341351382e-05,
+      "loss": 21.1138,
+      "step": 4071
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.89375111534519e-05,
+      "loss": 0.958,
+      "step": 4072
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.892698696601741e-05,
+      "loss": 0.2192,
+      "step": 4073
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.891646085191131e-05,
+      "loss": 21.6768,
+      "step": 4074
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.890593281183467e-05,
+      "loss": 0.933,
+      "step": 4075
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.889540284648873e-05,
+      "loss": 21.4794,
+      "step": 4076
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.888487095657484e-05,
+      "loss": 23.3424,
+      "step": 4077
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.887433714279444e-05,
+      "loss": 21.4879,
+      "step": 4078
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.886380140584915e-05,
+      "loss": 21.2877,
+      "step": 4079
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.885326374644072e-05,
+      "loss": 20.2731,
+      "step": 4080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.884272416527094e-05,
+      "loss": 22.4964,
+      "step": 4081
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.883218266304187e-05,
+      "loss": 20.129,
+      "step": 4082
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.88216392404556e-05,
+      "loss": 0.2118,
+      "step": 4083
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.881109389821433e-05,
+      "loss": 23.5236,
+      "step": 4084
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.88005466370205e-05,
+      "loss": 22.5755,
+      "step": 4085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.878999745757653e-05,
+      "loss": 0.2263,
+      "step": 4086
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.877944636058512e-05,
+      "loss": 20.4366,
+      "step": 4087
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.876889334674897e-05,
+      "loss": 0.9466,
+      "step": 4088
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.875833841677099e-05,
+      "loss": 23.202,
+      "step": 4089
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.874778157135415e-05,
+      "loss": 22.7037,
+      "step": 4090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.873722281120162e-05,
+      "loss": 0.2049,
+      "step": 4091
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.872666213701665e-05,
+      "loss": 0.9936,
+      "step": 4092
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.871609954950263e-05,
+      "loss": 21.7471,
+      "step": 4093
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.870553504936306e-05,
+      "loss": 22.867,
+      "step": 4094
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.869496863730161e-05,
+      "loss": 24.9188,
+      "step": 4095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.868440031402206e-05,
+      "loss": 0.1913,
+      "step": 4096
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.867383008022828e-05,
+      "loss": 21.9583,
+      "step": 4097
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.866325793662432e-05,
+      "loss": 20.9724,
+      "step": 4098
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.865268388391431e-05,
+      "loss": 0.1798,
+      "step": 4099
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.864210792280255e-05,
+      "loss": 22.7275,
+      "step": 4100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.863153005399344e-05,
+      "loss": 0.2456,
+      "step": 4101
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.862095027819155e-05,
+      "loss": 0.1872,
+      "step": 4102
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.861036859610146e-05,
+      "loss": 0.2563,
+      "step": 4103
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.859978500842804e-05,
+      "loss": 21.8858,
+      "step": 4104
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.858919951587615e-05,
+      "loss": 22.0782,
+      "step": 4105
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.857861211915089e-05,
+      "loss": 24.9101,
+      "step": 4106
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.856802281895739e-05,
+      "loss": 21.1008,
+      "step": 4107
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.855743161600093e-05,
+      "loss": 0.9814,
+      "step": 4108
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.854683851098698e-05,
+      "loss": 21.0855,
+      "step": 4109
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.853624350462107e-05,
+      "loss": 23.5163,
+      "step": 4110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.852564659760885e-05,
+      "loss": 0.9601,
+      "step": 4111
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.851504779065619e-05,
+      "loss": 0.8749,
+      "step": 4112
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.850444708446895e-05,
+      "loss": 0.9729,
+      "step": 4113
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.849384447975321e-05,
+      "loss": 21.6591,
+      "step": 4114
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.848323997721517e-05,
+      "loss": 24.6833,
+      "step": 4115
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.847263357756113e-05,
+      "loss": 0.2118,
+      "step": 4116
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.846202528149751e-05,
+      "loss": 0.2427,
+      "step": 4117
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.845141508973088e-05,
+      "loss": 0.909,
+      "step": 4118
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.844080300296793e-05,
+      "loss": 0.9709,
+      "step": 4119
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.843018902191548e-05,
+      "loss": 23.9565,
+      "step": 4120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.841957314728047e-05,
+      "loss": 21.2006,
+      "step": 4121
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.840895537976995e-05,
+      "loss": 20.3763,
+      "step": 4122
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.839833572009115e-05,
+      "loss": 0.2115,
+      "step": 4123
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.838771416895135e-05,
+      "loss": 0.9141,
+      "step": 4124
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.837709072705799e-05,
+      "loss": 22.7523,
+      "step": 4125
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.836646539511867e-05,
+      "loss": 22.2761,
+      "step": 4126
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.835583817384106e-05,
+      "loss": 22.4718,
+      "step": 4127
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.834520906393302e-05,
+      "loss": 21.1994,
+      "step": 4128
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.833457806610247e-05,
+      "loss": 0.9577,
+      "step": 4129
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.832394518105747e-05,
+      "loss": 0.1893,
+      "step": 4130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.831331040950627e-05,
+      "loss": 0.9948,
+      "step": 4131
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.830267375215716e-05,
+      "loss": 0.9561,
+      "step": 4132
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.829203520971858e-05,
+      "loss": 27.4465,
+      "step": 4133
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.828139478289913e-05,
+      "loss": 0.2302,
+      "step": 4134
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.82707524724075e-05,
+      "loss": 0.9747,
+      "step": 4135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.826010827895251e-05,
+      "loss": 25.6992,
+      "step": 4136
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.824946220324312e-05,
+      "loss": 0.9711,
+      "step": 4137
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.823881424598843e-05,
+      "loss": 0.2201,
+      "step": 4138
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.82281644078976e-05,
+      "loss": 21.0993,
+      "step": 4139
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.821751268968e-05,
+      "loss": 0.9457,
+      "step": 4140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.820685909204506e-05,
+      "loss": 23.6449,
+      "step": 4141
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.819620361570236e-05,
+      "loss": 0.961,
+      "step": 4142
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.818554626136164e-05,
+      "loss": 23.3397,
+      "step": 4143
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.817488702973265e-05,
+      "loss": 22.6483,
+      "step": 4144
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.816422592152543e-05,
+      "loss": 0.2065,
+      "step": 4145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.815356293745001e-05,
+      "loss": 0.1734,
+      "step": 4146
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.814289807821661e-05,
+      "loss": 21.8404,
+      "step": 4147
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.813223134453556e-05,
+      "loss": 0.9332,
+      "step": 4148
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.81215627371173e-05,
+      "loss": 0.9353,
+      "step": 4149
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.811089225667243e-05,
+      "loss": 0.9799,
+      "step": 4150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.810021990391164e-05,
+      "loss": 0.9294,
+      "step": 4151
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.808954567954574e-05,
+      "loss": 20.0699,
+      "step": 4152
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.807886958428572e-05,
+      "loss": 23.7496,
+      "step": 4153
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.806819161884265e-05,
+      "loss": 22.3162,
+      "step": 4154
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.805751178392771e-05,
+      "loss": 21.2971,
+      "step": 4155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.804683008025225e-05,
+      "loss": 23.3211,
+      "step": 4156
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.803614650852769e-05,
+      "loss": 0.9412,
+      "step": 4157
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.802546106946565e-05,
+      "loss": 23.6414,
+      "step": 4158
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.80147737637778e-05,
+      "loss": 20.3343,
+      "step": 4159
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.800408459217595e-05,
+      "loss": 24.0691,
+      "step": 4160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.79933935553721e-05,
+      "loss": 22.2688,
+      "step": 4161
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.798270065407827e-05,
+      "loss": 22.2032,
+      "step": 4162
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.797200588900667e-05,
+      "loss": 23.6172,
+      "step": 4163
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.796130926086965e-05,
+      "loss": 22.0459,
+      "step": 4164
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.795061077037962e-05,
+      "loss": 0.931,
+      "step": 4165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.793991041824917e-05,
+      "loss": 22.0503,
+      "step": 4166
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.792920820519098e-05,
+      "loss": 0.9396,
+      "step": 4167
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.791850413191789e-05,
+      "loss": 22.4159,
+      "step": 4168
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.790779819914281e-05,
+      "loss": 20.6819,
+      "step": 4169
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.789709040757882e-05,
+      "loss": 0.9211,
+      "step": 4170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.78863807579391e-05,
+      "loss": 20.5289,
+      "step": 4171
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.787566925093699e-05,
+      "loss": 0.9285,
+      "step": 4172
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.786495588728591e-05,
+      "loss": 0.9341,
+      "step": 4173
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.78542406676994e-05,
+      "loss": 0.9484,
+      "step": 4174
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.784352359289115e-05,
+      "loss": 22.3875,
+      "step": 4175
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.7832804663575e-05,
+      "loss": 23.8806,
+      "step": 4176
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.782208388046483e-05,
+      "loss": 0.9262,
+      "step": 4177
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.781136124427475e-05,
+      "loss": 23.4489,
+      "step": 4178
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.780063675571889e-05,
+      "loss": 0.9842,
+      "step": 4179
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.778991041551158e-05,
+      "loss": 0.1921,
+      "step": 4180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.777918222436722e-05,
+      "loss": 21.2403,
+      "step": 4181
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.776845218300037e-05,
+      "loss": 0.2151,
+      "step": 4182
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.775772029212572e-05,
+      "loss": 0.2115,
+      "step": 4183
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.774698655245802e-05,
+      "loss": 21.1993,
+      "step": 4184
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.773625096471222e-05,
+      "loss": 21.6149,
+      "step": 4185
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.772551352960337e-05,
+      "loss": 24.9679,
+      "step": 4186
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.77147742478466e-05,
+      "loss": 22.9756,
+      "step": 4187
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.770403312015721e-05,
+      "loss": 20.9029,
+      "step": 4188
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.76932901472506e-05,
+      "loss": 21.8065,
+      "step": 4189
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.768254532984232e-05,
+      "loss": 0.9544,
+      "step": 4190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.767179866864802e-05,
+      "loss": 0.2592,
+      "step": 4191
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.766105016438347e-05,
+      "loss": 22.385,
+      "step": 4192
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.765029981776457e-05,
+      "loss": 24.7385,
+      "step": 4193
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.763954762950735e-05,
+      "loss": 22.0968,
+      "step": 4194
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.762879360032796e-05,
+      "loss": 20.7728,
+      "step": 4195
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.761803773094264e-05,
+      "loss": 19.952,
+      "step": 4196
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.760728002206782e-05,
+      "loss": 0.9576,
+      "step": 4197
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.759652047441999e-05,
+      "loss": 21.8086,
+      "step": 4198
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.758575908871578e-05,
+      "loss": 22.0586,
+      "step": 4199
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.757499586567199e-05,
+      "loss": 21.9272,
+      "step": 4200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.756423080600543e-05,
+      "loss": 23.0235,
+      "step": 4201
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.755346391043316e-05,
+      "loss": 0.9446,
+      "step": 4202
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.75426951796723e-05,
+      "loss": 0.2235,
+      "step": 4203
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.753192461444007e-05,
+      "loss": 20.7423,
+      "step": 4204
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.752115221545386e-05,
+      "loss": 23.2332,
+      "step": 4205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.751037798343114e-05,
+      "loss": 20.12,
+      "step": 4206
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.749960191908956e-05,
+      "loss": 0.196,
+      "step": 4207
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.748882402314685e-05,
+      "loss": 0.2481,
+      "step": 4208
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.747804429632082e-05,
+      "loss": 0.931,
+      "step": 4209
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.74672627393295e-05,
+      "loss": 0.918,
+      "step": 4210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.745647935289097e-05,
+      "loss": 0.2024,
+      "step": 4211
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.744569413772348e-05,
+      "loss": 0.2011,
+      "step": 4212
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.743490709454534e-05,
+      "loss": 25.0265,
+      "step": 4213
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.742411822407503e-05,
+      "loss": 0.2073,
+      "step": 4214
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.741332752703114e-05,
+      "loss": 0.9364,
+      "step": 4215
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.740253500413239e-05,
+      "loss": 23.2356,
+      "step": 4216
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.73917406560976e-05,
+      "loss": 23.3242,
+      "step": 4217
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.738094448364572e-05,
+      "loss": 0.9487,
+      "step": 4218
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.737014648749584e-05,
+      "loss": 23.332,
+      "step": 4219
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.735934666836715e-05,
+      "loss": 24.5501,
+      "step": 4220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.734854502697896e-05,
+      "loss": 0.8968,
+      "step": 4221
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.733774156405072e-05,
+      "loss": 0.9883,
+      "step": 4222
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.7326936280302e-05,
+      "loss": 0.9646,
+      "step": 4223
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.731612917645246e-05,
+      "loss": 0.2171,
+      "step": 4224
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.73053202532219e-05,
+      "loss": 23.9944,
+      "step": 4225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.729450951133026e-05,
+      "loss": 0.2185,
+      "step": 4226
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.72836969514976e-05,
+      "loss": 25.1911,
+      "step": 4227
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.727288257444406e-05,
+      "loss": 0.9862,
+      "step": 4228
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.726206638088993e-05,
+      "loss": 0.9957,
+      "step": 4229
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.725124837155565e-05,
+      "loss": 23.7989,
+      "step": 4230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.724042854716169e-05,
+      "loss": 20.7217,
+      "step": 4231
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.722960690842875e-05,
+      "loss": 0.9675,
+      "step": 4232
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.72187834560776e-05,
+      "loss": 21.8991,
+      "step": 4233
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.720795819082911e-05,
+      "loss": 0.8951,
+      "step": 4234
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.71971311134043e-05,
+      "loss": 21.3948,
+      "step": 4235
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.718630222452433e-05,
+      "loss": 21.2725,
+      "step": 4236
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.717547152491039e-05,
+      "loss": 1.0021,
+      "step": 4237
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.716463901528392e-05,
+      "loss": 0.187,
+      "step": 4238
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.71538046963664e-05,
+      "loss": 22.5831,
+      "step": 4239
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.714296856887941e-05,
+      "loss": 21.2196,
+      "step": 4240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.713213063354472e-05,
+      "loss": 0.8871,
+      "step": 4241
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.712129089108419e-05,
+      "loss": 0.2084,
+      "step": 4242
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.711044934221978e-05,
+      "loss": 1.0107,
+      "step": 4243
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.70996059876736e-05,
+      "loss": 0.9499,
+      "step": 4244
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.708876082816785e-05,
+      "loss": 0.8934,
+      "step": 4245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.707791386442488e-05,
+      "loss": 21.7513,
+      "step": 4246
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.706706509716719e-05,
+      "loss": 1.0158,
+      "step": 4247
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.705621452711726e-05,
+      "loss": 22.3638,
+      "step": 4248
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.704536215499788e-05,
+      "loss": 20.7667,
+      "step": 4249
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.703450798153182e-05,
+      "loss": 22.3905,
+      "step": 4250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.702365200744204e-05,
+      "loss": 22.3424,
+      "step": 4251
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.701279423345158e-05,
+      "loss": 20.1973,
+      "step": 4252
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.700193466028363e-05,
+      "loss": 23.3435,
+      "step": 4253
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.699107328866148e-05,
+      "loss": 0.2173,
+      "step": 4254
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.698021011930859e-05,
+      "loss": 0.2508,
+      "step": 4255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.696934515294842e-05,
+      "loss": 21.1911,
+      "step": 4256
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.695847839030468e-05,
+      "loss": 19.448,
+      "step": 4257
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.694760983210114e-05,
+      "loss": 0.9651,
+      "step": 4258
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.693673947906168e-05,
+      "loss": 0.2004,
+      "step": 4259
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.692586733191034e-05,
+      "loss": 22.9395,
+      "step": 4260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.691499339137125e-05,
+      "loss": 0.9708,
+      "step": 4261
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.690411765816864e-05,
+      "loss": 0.2384,
+      "step": 4262
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.689324013302692e-05,
+      "loss": 22.351,
+      "step": 4263
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.688236081667055e-05,
+      "loss": 21.5127,
+      "step": 4264
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.687147970982415e-05,
+      "loss": 21.3918,
+      "step": 4265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.686059681321249e-05,
+      "loss": 22.121,
+      "step": 4266
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.684971212756038e-05,
+      "loss": 19.9097,
+      "step": 4267
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.683882565359279e-05,
+      "loss": 22.4368,
+      "step": 4268
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.682793739203485e-05,
+      "loss": 0.954,
+      "step": 4269
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.681704734361173e-05,
+      "loss": 20.8444,
+      "step": 4270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.680615550904876e-05,
+      "loss": 24.3549,
+      "step": 4271
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.679526188907141e-05,
+      "loss": 0.9633,
+      "step": 4272
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.678436648440522e-05,
+      "loss": 0.9771,
+      "step": 4273
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.67734692957759e-05,
+      "loss": 22.0753,
+      "step": 4274
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.676257032390923e-05,
+      "loss": 0.2036,
+      "step": 4275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.675166956953114e-05,
+      "loss": 23.5139,
+      "step": 4276
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.674076703336768e-05,
+      "loss": 0.9492,
+      "step": 4277
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.6729862716145e-05,
+      "loss": 0.9441,
+      "step": 4278
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.671895661858938e-05,
+      "loss": 20.7924,
+      "step": 4279
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.67080487414272e-05,
+      "loss": 0.2183,
+      "step": 4280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.669713908538502e-05,
+      "loss": 0.9508,
+      "step": 4281
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.668622765118945e-05,
+      "loss": 24.8615,
+      "step": 4282
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.667531443956722e-05,
+      "loss": 24.1414,
+      "step": 4283
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.666439945124522e-05,
+      "loss": 24.3789,
+      "step": 4284
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.665348268695043e-05,
+      "loss": 21.1466,
+      "step": 4285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.664256414740998e-05,
+      "loss": 21.9723,
+      "step": 4286
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.663164383335107e-05,
+      "loss": 20.7218,
+      "step": 4287
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.662072174550107e-05,
+      "loss": 0.1867,
+      "step": 4288
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.66097978845874e-05,
+      "loss": 0.9553,
+      "step": 4289
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.65988722513377e-05,
+      "loss": 0.2547,
+      "step": 4290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.658794484647961e-05,
+      "loss": 21.8514,
+      "step": 4291
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.657701567074097e-05,
+      "loss": 0.921,
+      "step": 4292
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.656608472484972e-05,
+      "loss": 0.1631,
+      "step": 4293
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.65551520095339e-05,
+      "loss": 20.333,
+      "step": 4294
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.654421752552169e-05,
+      "loss": 20.3483,
+      "step": 4295
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.653328127354136e-05,
+      "loss": 0.2255,
+      "step": 4296
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.652234325432135e-05,
+      "loss": 21.0039,
+      "step": 4297
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.651140346859013e-05,
+      "loss": 19.8884,
+      "step": 4298
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.650046191707641e-05,
+      "loss": 0.9329,
+      "step": 4299
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.648951860050887e-05,
+      "loss": 1.0064,
+      "step": 4300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.647857351961644e-05,
+      "loss": 0.9298,
+      "step": 4301
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.64676266751281e-05,
+      "loss": 19.9558,
+      "step": 4302
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.645667806777295e-05,
+      "loss": 0.1959,
+      "step": 4303
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.644572769828025e-05,
+      "loss": 20.9646,
+      "step": 4304
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.643477556737929e-05,
+      "loss": 22.3025,
+      "step": 4305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.642382167579957e-05,
+      "loss": 20.5406,
+      "step": 4306
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.641286602427067e-05,
+      "loss": 0.2282,
+      "step": 4307
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.64019086135223e-05,
+      "loss": 0.1758,
+      "step": 4308
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.639094944428424e-05,
+      "loss": 22.2143,
+      "step": 4309
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.637998851728643e-05,
+      "loss": 0.2082,
+      "step": 4310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.636902583325893e-05,
+      "loss": 0.215,
+      "step": 4311
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.63580613929319e-05,
+      "loss": 0.1912,
+      "step": 4312
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.634709519703564e-05,
+      "loss": 0.2162,
+      "step": 4313
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.633612724630053e-05,
+      "loss": 0.936,
+      "step": 4314
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.632515754145708e-05,
+      "loss": 21.7323,
+      "step": 4315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.631418608323593e-05,
+      "loss": 0.9537,
+      "step": 4316
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.630321287236785e-05,
+      "loss": 0.2111,
+      "step": 4317
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.629223790958368e-05,
+      "loss": 0.9711,
+      "step": 4318
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.628126119561442e-05,
+      "loss": 20.2258,
+      "step": 4319
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.627028273119116e-05,
+      "loss": 0.9797,
+      "step": 4320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.625930251704512e-05,
+      "loss": 0.1813,
+      "step": 4321
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.624832055390763e-05,
+      "loss": 0.2126,
+      "step": 4322
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.623733684251015e-05,
+      "loss": 21.0449,
+      "step": 4323
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.622635138358426e-05,
+      "loss": 0.2014,
+      "step": 4324
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.621536417786159e-05,
+      "loss": 21.0511,
+      "step": 4325
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.620437522607397e-05,
+      "loss": 23.8365,
+      "step": 4326
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.619338452895335e-05,
+      "loss": 22.3135,
+      "step": 4327
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.61823920872317e-05,
+      "loss": 22.1565,
+      "step": 4328
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.61713979016412e-05,
+      "loss": 0.9712,
+      "step": 4329
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.616040197291412e-05,
+      "loss": 0.9429,
+      "step": 4330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.614940430178284e-05,
+      "loss": 21.3447,
+      "step": 4331
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.613840488897981e-05,
+      "loss": 0.9439,
+      "step": 4332
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.61274037352377e-05,
+      "loss": 20.3539,
+      "step": 4333
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.611640084128921e-05,
+      "loss": 20.4903,
+      "step": 4334
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.610539620786719e-05,
+      "loss": 0.1973,
+      "step": 4335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.60943898357046e-05,
+      "loss": 0.9403,
+      "step": 4336
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.608338172553453e-05,
+      "loss": 22.834,
+      "step": 4337
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.607237187809013e-05,
+      "loss": 20.1284,
+      "step": 4338
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.606136029410475e-05,
+      "loss": 0.253,
+      "step": 4339
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.605034697431178e-05,
+      "loss": 20.6552,
+      "step": 4340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.603933191944478e-05,
+      "loss": 0.9785,
+      "step": 4341
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.60283151302374e-05,
+      "loss": 20.0124,
+      "step": 4342
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.601729660742341e-05,
+      "loss": 20.5884,
+      "step": 4343
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.600627635173668e-05,
+      "loss": 0.2072,
+      "step": 4344
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.599525436391122e-05,
+      "loss": 0.2226,
+      "step": 4345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.598423064468115e-05,
+      "loss": 20.9004,
+      "step": 4346
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.597320519478071e-05,
+      "loss": 20.7425,
+      "step": 4347
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.596217801494422e-05,
+      "loss": 18.7691,
+      "step": 4348
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.595114910590618e-05,
+      "loss": 0.9885,
+      "step": 4349
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.59401184684011e-05,
+      "loss": 21.6318,
+      "step": 4350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.592908610316374e-05,
+      "loss": 0.2477,
+      "step": 4351
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.591805201092888e-05,
+      "loss": 21.1835,
+      "step": 4352
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.590701619243143e-05,
+      "loss": 22.0707,
+      "step": 4353
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.589597864840646e-05,
+      "loss": 0.8877,
+      "step": 4354
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.588493937958907e-05,
+      "loss": 0.9313,
+      "step": 4355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.587389838671459e-05,
+      "loss": 0.226,
+      "step": 4356
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.586285567051834e-05,
+      "loss": 23.3925,
+      "step": 4357
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.585181123173584e-05,
+      "loss": 21.698,
+      "step": 4358
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.584076507110274e-05,
+      "loss": 22.4428,
+      "step": 4359
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.58297171893547e-05,
+      "loss": 0.9655,
+      "step": 4360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.581866758722762e-05,
+      "loss": 22.6449,
+      "step": 4361
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.580761626545741e-05,
+      "loss": 22.9375,
+      "step": 4362
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.579656322478013e-05,
+      "loss": 19.8081,
+      "step": 4363
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.578550846593203e-05,
+      "loss": 20.0696,
+      "step": 4364
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.577445198964936e-05,
+      "loss": 23.3249,
+      "step": 4365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.576339379666852e-05,
+      "loss": 0.2029,
+      "step": 4366
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.575233388772606e-05,
+      "loss": 22.2904,
+      "step": 4367
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.574127226355864e-05,
+      "loss": 0.2078,
+      "step": 4368
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.573020892490298e-05,
+      "loss": 0.2,
+      "step": 4369
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.571914387249597e-05,
+      "loss": 23.1228,
+      "step": 4370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.570807710707458e-05,
+      "loss": 20.2868,
+      "step": 4371
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.56970086293759e-05,
+      "loss": 19.3834,
+      "step": 4372
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.568593844013718e-05,
+      "loss": 19.8896,
+      "step": 4373
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.567486654009572e-05,
+      "loss": 0.2141,
+      "step": 4374
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.566379292998896e-05,
+      "loss": 21.8747,
+      "step": 4375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.565271761055446e-05,
+      "loss": 0.2062,
+      "step": 4376
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.564164058252987e-05,
+      "loss": 22.8399,
+      "step": 4377
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5630561846653e-05,
+      "loss": 0.9656,
+      "step": 4378
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.561948140366173e-05,
+      "loss": 23.6899,
+      "step": 4379
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.560839925429407e-05,
+      "loss": 24.1536,
+      "step": 4380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.559731539928815e-05,
+      "loss": 21.3563,
+      "step": 4381
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.558622983938219e-05,
+      "loss": 21.3375,
+      "step": 4382
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.557514257531454e-05,
+      "loss": 20.1972,
+      "step": 4383
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.55640536078237e-05,
+      "loss": 0.2025,
+      "step": 4384
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.555296293764818e-05,
+      "loss": 23.0293,
+      "step": 4385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.554187056552675e-05,
+      "loss": 0.2118,
+      "step": 4386
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.553077649219815e-05,
+      "loss": 21.1455,
+      "step": 4387
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.551968071840132e-05,
+      "loss": 0.2069,
+      "step": 4388
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.55085832448753e-05,
+      "loss": 22.7914,
+      "step": 4389
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.549748407235921e-05,
+      "loss": 0.9411,
+      "step": 4390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.548638320159231e-05,
+      "loss": 20.5963,
+      "step": 4391
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.547528063331401e-05,
+      "loss": 0.1934,
+      "step": 4392
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.546417636826374e-05,
+      "loss": 24.703,
+      "step": 4393
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.545307040718111e-05,
+      "loss": 20.538,
+      "step": 4394
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.544196275080585e-05,
+      "loss": 22.9273,
+      "step": 4395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.543085339987777e-05,
+      "loss": 0.1811,
+      "step": 4396
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.541974235513679e-05,
+      "loss": 19.8896,
+      "step": 4397
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.540862961732297e-05,
+      "loss": 0.1981,
+      "step": 4398
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.539751518717646e-05,
+      "loss": 22.745,
+      "step": 4399
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.538639906543757e-05,
+      "loss": 20.0619,
+      "step": 4400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.537528125284663e-05,
+      "loss": 21.9511,
+      "step": 4401
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.536416175014415e-05,
+      "loss": 22.0789,
+      "step": 4402
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.53530405580708e-05,
+      "loss": 0.2303,
+      "step": 4403
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.534191767736722e-05,
+      "loss": 20.8291,
+      "step": 4404
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.533079310877429e-05,
+      "loss": 0.9901,
+      "step": 4405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.531966685303296e-05,
+      "loss": 20.8124,
+      "step": 4406
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.530853891088428e-05,
+      "loss": 0.9308,
+      "step": 4407
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.529740928306943e-05,
+      "loss": 0.9631,
+      "step": 4408
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.528627797032968e-05,
+      "loss": 0.949,
+      "step": 4409
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.527514497340642e-05,
+      "loss": 21.2769,
+      "step": 4410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.52640102930412e-05,
+      "loss": 20.8377,
+      "step": 4411
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.52528739299756e-05,
+      "loss": 0.9112,
+      "step": 4412
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.524173588495138e-05,
+      "loss": 21.2616,
+      "step": 4413
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.523059615871038e-05,
+      "loss": 0.9776,
+      "step": 4414
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.521945475199455e-05,
+      "loss": 20.2969,
+      "step": 4415
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.520831166554597e-05,
+      "loss": 22.3862,
+      "step": 4416
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.519716690010679e-05,
+      "loss": 21.0485,
+      "step": 4417
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.518602045641934e-05,
+      "loss": 0.2032,
+      "step": 4418
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5174872335226e-05,
+      "loss": 0.2093,
+      "step": 4419
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.516372253726931e-05,
+      "loss": 0.221,
+      "step": 4420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.515257106329188e-05,
+      "loss": 21.462,
+      "step": 4421
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.514141791403646e-05,
+      "loss": 0.2259,
+      "step": 4422
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.513026309024591e-05,
+      "loss": 20.1648,
+      "step": 4423
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.511910659266317e-05,
+      "loss": 19.4344,
+      "step": 4424
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.510794842203132e-05,
+      "loss": 20.5069,
+      "step": 4425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.509678857909357e-05,
+      "loss": 21.5323,
+      "step": 4426
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.508562706459318e-05,
+      "loss": 20.3907,
+      "step": 4427
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.50744638792736e-05,
+      "loss": 21.7707,
+      "step": 4428
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.50632990238783e-05,
+      "loss": 21.3877,
+      "step": 4429
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.505213249915095e-05,
+      "loss": 22.1816,
+      "step": 4430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.504096430583528e-05,
+      "loss": 21.0013,
+      "step": 4431
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.502979444467514e-05,
+      "loss": 0.9248,
+      "step": 4432
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.501862291641451e-05,
+      "loss": 19.0419,
+      "step": 4433
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.500744972179745e-05,
+      "loss": 23.6552,
+      "step": 4434
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.499627486156816e-05,
+      "loss": 0.1608,
+      "step": 4435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.498509833647091e-05,
+      "loss": 21.4279,
+      "step": 4436
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.497392014725013e-05,
+      "loss": 0.9786,
+      "step": 4437
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.496274029465036e-05,
+      "loss": 20.0325,
+      "step": 4438
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.495155877941619e-05,
+      "loss": 21.0142,
+      "step": 4439
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.494037560229239e-05,
+      "loss": 21.2668,
+      "step": 4440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.492919076402379e-05,
+      "loss": 19.5856,
+      "step": 4441
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.491800426535535e-05,
+      "loss": 1.0053,
+      "step": 4442
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.490681610703216e-05,
+      "loss": 20.9693,
+      "step": 4443
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.489562628979942e-05,
+      "loss": 21.27,
+      "step": 4444
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.488443481440238e-05,
+      "loss": 20.0272,
+      "step": 4445
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.487324168158646e-05,
+      "loss": 21.6595,
+      "step": 4446
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.48620468920972e-05,
+      "loss": 24.5912,
+      "step": 4447
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.485085044668019e-05,
+      "loss": 0.2579,
+      "step": 4448
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.483965234608119e-05,
+      "loss": 0.1943,
+      "step": 4449
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.482845259104602e-05,
+      "loss": 21.6588,
+      "step": 4450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.481725118232066e-05,
+      "loss": 0.189,
+      "step": 4451
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.480604812065117e-05,
+      "loss": 0.2126,
+      "step": 4452
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.479484340678373e-05,
+      "loss": 0.9684,
+      "step": 4453
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.47836370414646e-05,
+      "loss": 23.3949,
+      "step": 4454
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.477242902544021e-05,
+      "loss": 22.0848,
+      "step": 4455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.476121935945704e-05,
+      "loss": 21.3246,
+      "step": 4456
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.475000804426172e-05,
+      "loss": 19.5245,
+      "step": 4457
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.473879508060098e-05,
+      "loss": 20.0925,
+      "step": 4458
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.472758046922163e-05,
+      "loss": 0.9651,
+      "step": 4459
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.471636421087067e-05,
+      "loss": 23.3493,
+      "step": 4460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.470514630629508e-05,
+      "loss": 0.8885,
+      "step": 4461
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.469392675624207e-05,
+      "loss": 0.2366,
+      "step": 4462
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.468270556145894e-05,
+      "loss": 0.2139,
+      "step": 4463
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4671482722693e-05,
+      "loss": 20.0591,
+      "step": 4464
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.466025824069179e-05,
+      "loss": 21.0293,
+      "step": 4465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.464903211620291e-05,
+      "loss": 0.2425,
+      "step": 4466
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.463780434997407e-05,
+      "loss": 0.916,
+      "step": 4467
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.462657494275309e-05,
+      "loss": 0.9186,
+      "step": 4468
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.46153438952879e-05,
+      "loss": 0.9663,
+      "step": 4469
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.460411120832653e-05,
+      "loss": 20.7138,
+      "step": 4470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.459287688261716e-05,
+      "loss": 0.9388,
+      "step": 4471
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.458164091890802e-05,
+      "loss": 0.2388,
+      "step": 4472
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.457040331794747e-05,
+      "loss": 0.9559,
+      "step": 4473
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4559164080484e-05,
+      "loss": 1.0069,
+      "step": 4474
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.454792320726623e-05,
+      "loss": 0.1914,
+      "step": 4475
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.453668069904278e-05,
+      "loss": 0.9367,
+      "step": 4476
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.452543655656252e-05,
+      "loss": 20.3012,
+      "step": 4477
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.451419078057432e-05,
+      "loss": 0.9798,
+      "step": 4478
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.450294337182722e-05,
+      "loss": 0.1973,
+      "step": 4479
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.449169433107034e-05,
+      "loss": 1.0291,
+      "step": 4480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.448044365905293e-05,
+      "loss": 23.4331,
+      "step": 4481
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.446919135652434e-05,
+      "loss": 22.9997,
+      "step": 4482
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.445793742423403e-05,
+      "loss": 0.2194,
+      "step": 4483
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.444668186293153e-05,
+      "loss": 0.2225,
+      "step": 4484
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.443542467336654e-05,
+      "loss": 22.1913,
+      "step": 4485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.442416585628883e-05,
+      "loss": 0.9576,
+      "step": 4486
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.44129054124483e-05,
+      "loss": 20.1377,
+      "step": 4487
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.440164334259494e-05,
+      "loss": 20.5668,
+      "step": 4488
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.439037964747886e-05,
+      "loss": 20.0271,
+      "step": 4489
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.437911432785027e-05,
+      "loss": 21.0846,
+      "step": 4490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.43678473844595e-05,
+      "loss": 0.975,
+      "step": 4491
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.435657881805698e-05,
+      "loss": 20.1794,
+      "step": 4492
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.434530862939324e-05,
+      "loss": 20.9866,
+      "step": 4493
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.433403681921892e-05,
+      "loss": 0.9189,
+      "step": 4494
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.43227633882848e-05,
+      "loss": 0.1792,
+      "step": 4495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.431148833734173e-05,
+      "loss": 21.2858,
+      "step": 4496
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.430021166714067e-05,
+      "loss": 22.673,
+      "step": 4497
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.428893337843272e-05,
+      "loss": 0.9655,
+      "step": 4498
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.427765347196905e-05,
+      "loss": 22.3033,
+      "step": 4499
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.426637194850097e-05,
+      "loss": 22.9755,
+      "step": 4500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.425508880877986e-05,
+      "loss": 0.2021,
+      "step": 4501
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.424380405355724e-05,
+      "loss": 1.0143,
+      "step": 4502
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.423251768358474e-05,
+      "loss": 0.2274,
+      "step": 4503
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.422122969961406e-05,
+      "loss": 0.9293,
+      "step": 4504
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.420994010239706e-05,
+      "loss": 0.2239,
+      "step": 4505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.419864889268563e-05,
+      "loss": 22.6853,
+      "step": 4506
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.418735607123189e-05,
+      "loss": 0.2294,
+      "step": 4507
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.417606163878793e-05,
+      "loss": 0.1964,
+      "step": 4508
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.416476559610606e-05,
+      "loss": 19.8236,
+      "step": 4509
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.415346794393861e-05,
+      "loss": 22.0261,
+      "step": 4510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.414216868303807e-05,
+      "loss": 21.8357,
+      "step": 4511
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.413086781415704e-05,
+      "loss": 20.2621,
+      "step": 4512
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.411956533804818e-05,
+      "loss": 0.2158,
+      "step": 4513
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.41082612554643e-05,
+      "loss": 0.1974,
+      "step": 4514
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.409695556715834e-05,
+      "loss": 20.8477,
+      "step": 4515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.408564827388326e-05,
+      "loss": 20.6411,
+      "step": 4516
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.407433937639219e-05,
+      "loss": 18.4383,
+      "step": 4517
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.406302887543836e-05,
+      "loss": 23.5632,
+      "step": 4518
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.405171677177511e-05,
+      "loss": 21.8876,
+      "step": 4519
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.404040306615587e-05,
+      "loss": 0.9504,
+      "step": 4520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.402908775933419e-05,
+      "loss": 21.3136,
+      "step": 4521
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.401777085206373e-05,
+      "loss": 19.3835,
+      "step": 4522
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.400645234509822e-05,
+      "loss": 0.9132,
+      "step": 4523
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.399513223919156e-05,
+      "loss": 20.0455,
+      "step": 4524
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.398381053509768e-05,
+      "loss": 21.8457,
+      "step": 4525
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.397248723357071e-05,
+      "loss": 0.9774,
+      "step": 4526
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.396116233536481e-05,
+      "loss": 21.5894,
+      "step": 4527
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.394983584123424e-05,
+      "loss": 22.2906,
+      "step": 4528
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.393850775193344e-05,
+      "loss": 0.9554,
+      "step": 4529
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.392717806821692e-05,
+      "loss": 22.5722,
+      "step": 4530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.391584679083924e-05,
+      "loss": 22.7944,
+      "step": 4531
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.390451392055516e-05,
+      "loss": 22.7552,
+      "step": 4532
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.389317945811947e-05,
+      "loss": 22.0976,
+      "step": 4533
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.388184340428712e-05,
+      "loss": 0.1801,
+      "step": 4534
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.387050575981315e-05,
+      "loss": 0.9805,
+      "step": 4535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.385916652545267e-05,
+      "loss": 22.1768,
+      "step": 4536
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.384782570196096e-05,
+      "loss": 21.1298,
+      "step": 4537
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.383648329009335e-05,
+      "loss": 0.9645,
+      "step": 4538
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.382513929060529e-05,
+      "loss": 21.9371,
+      "step": 4539
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.381379370425237e-05,
+      "loss": 21.7937,
+      "step": 4540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.380244653179023e-05,
+      "loss": 0.2048,
+      "step": 4541
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.379109777397467e-05,
+      "loss": 0.9286,
+      "step": 4542
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.377974743156153e-05,
+      "loss": 22.8149,
+      "step": 4543
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.376839550530684e-05,
+      "loss": 20.2503,
+      "step": 4544
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.375704199596669e-05,
+      "loss": 0.9587,
+      "step": 4545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.374568690429724e-05,
+      "loss": 22.7927,
+      "step": 4546
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.373433023105482e-05,
+      "loss": 0.9558,
+      "step": 4547
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.372297197699582e-05,
+      "loss": 20.1249,
+      "step": 4548
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.371161214287678e-05,
+      "loss": 0.9147,
+      "step": 4549
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.370025072945429e-05,
+      "loss": 22.3817,
+      "step": 4550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.368888773748507e-05,
+      "loss": 19.4386,
+      "step": 4551
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3677523167726e-05,
+      "loss": 22.2304,
+      "step": 4552
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.366615702093393e-05,
+      "loss": 0.2369,
+      "step": 4553
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.365478929786597e-05,
+      "loss": 22.3287,
+      "step": 4554
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.364341999927921e-05,
+      "loss": 21.1445,
+      "step": 4555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.363204912593095e-05,
+      "loss": 0.1932,
+      "step": 4556
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.36206766785785e-05,
+      "loss": 21.4381,
+      "step": 4557
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.360930265797935e-05,
+      "loss": 23.3425,
+      "step": 4558
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.359792706489102e-05,
+      "loss": 0.2455,
+      "step": 4559
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.358654990007122e-05,
+      "loss": 21.7949,
+      "step": 4560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.35751711642777e-05,
+      "loss": 0.1785,
+      "step": 4561
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.356379085826836e-05,
+      "loss": 20.3534,
+      "step": 4562
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.355240898280113e-05,
+      "loss": 21.2246,
+      "step": 4563
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.354102553863416e-05,
+      "loss": 0.2123,
+      "step": 4564
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.35296405265256e-05,
+      "loss": 0.9782,
+      "step": 4565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.351825394723373e-05,
+      "loss": 21.5936,
+      "step": 4566
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.350686580151701e-05,
+      "loss": 0.9349,
+      "step": 4567
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.349547609013388e-05,
+      "loss": 21.7244,
+      "step": 4568
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.348408481384299e-05,
+      "loss": 0.9549,
+      "step": 4569
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.347269197340303e-05,
+      "loss": 0.2272,
+      "step": 4570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.346129756957283e-05,
+      "loss": 22.4725,
+      "step": 4571
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.34499016031113e-05,
+      "loss": 0.2254,
+      "step": 4572
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.343850407477745e-05,
+      "loss": 0.9616,
+      "step": 4573
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.342710498533045e-05,
+      "loss": 24.172,
+      "step": 4574
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.341570433552948e-05,
+      "loss": 22.4472,
+      "step": 4575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.340430212613392e-05,
+      "loss": 0.2216,
+      "step": 4576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.339289835790318e-05,
+      "loss": 0.226,
+      "step": 4577
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.338149303159684e-05,
+      "loss": 23.2573,
+      "step": 4578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.337008614797452e-05,
+      "loss": 0.1899,
+      "step": 4579
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.335867770779595e-05,
+      "loss": 18.922,
+      "step": 4580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.334726771182103e-05,
+      "loss": 0.223,
+      "step": 4581
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.333585616080972e-05,
+      "loss": 22.9397,
+      "step": 4582
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.332444305552203e-05,
+      "loss": 0.9375,
+      "step": 4583
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.331302839671817e-05,
+      "loss": 0.9203,
+      "step": 4584
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.330161218515841e-05,
+      "loss": 0.9545,
+      "step": 4585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.329019442160307e-05,
+      "loss": 0.9289,
+      "step": 4586
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.327877510681272e-05,
+      "loss": 0.9593,
+      "step": 4587
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.326735424154785e-05,
+      "loss": 0.2332,
+      "step": 4588
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.32559318265692e-05,
+      "loss": 21.6986,
+      "step": 4589
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.324450786263752e-05,
+      "loss": 0.888,
+      "step": 4590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.323308235051372e-05,
+      "loss": 20.1998,
+      "step": 4591
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.322165529095877e-05,
+      "loss": 0.9673,
+      "step": 4592
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.321022668473378e-05,
+      "loss": 0.8807,
+      "step": 4593
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.319879653259997e-05,
+      "loss": 0.9606,
+      "step": 4594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.31873648353186e-05,
+      "loss": 22.1899,
+      "step": 4595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.317593159365111e-05,
+      "loss": 0.1869,
+      "step": 4596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.316449680835898e-05,
+      "loss": 21.5247,
+      "step": 4597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.315306048020384e-05,
+      "loss": 20.2353,
+      "step": 4598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.314162260994739e-05,
+      "loss": 20.0799,
+      "step": 4599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.313018319835144e-05,
+      "loss": 0.1958,
+      "step": 4600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.311874224617793e-05,
+      "loss": 0.2278,
+      "step": 4601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.310729975418887e-05,
+      "loss": 0.9605,
+      "step": 4602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.309585572314636e-05,
+      "loss": 0.8757,
+      "step": 4603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.308441015381267e-05,
+      "loss": 0.9338,
+      "step": 4604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.307296304695007e-05,
+      "loss": 0.9321,
+      "step": 4605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.306151440332106e-05,
+      "loss": 0.9629,
+      "step": 4606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.305006422368811e-05,
+      "loss": 24.7741,
+      "step": 4607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.303861250881389e-05,
+      "loss": 23.1187,
+      "step": 4608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.302715925946115e-05,
+      "loss": 0.9741,
+      "step": 4609
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.301570447639267e-05,
+      "loss": 0.9426,
+      "step": 4610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.300424816037146e-05,
+      "loss": 19.9916,
+      "step": 4611
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.299279031216053e-05,
+      "loss": 21.8515,
+      "step": 4612
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.298133093252301e-05,
+      "loss": 19.2089,
+      "step": 4613
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.296987002222219e-05,
+      "loss": 20.3889,
+      "step": 4614
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.29584075820214e-05,
+      "loss": 0.1875,
+      "step": 4615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.294694361268409e-05,
+      "loss": 22.3828,
+      "step": 4616
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.29354781149738e-05,
+      "loss": 22.4532,
+      "step": 4617
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.29240110896542e-05,
+      "loss": 0.9463,
+      "step": 4618
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.291254253748906e-05,
+      "loss": 21.8746,
+      "step": 4619
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.290107245924222e-05,
+      "loss": 0.9336,
+      "step": 4620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.288960085567764e-05,
+      "loss": 20.9206,
+      "step": 4621
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.287812772755938e-05,
+      "loss": 21.0803,
+      "step": 4622
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.286665307565163e-05,
+      "loss": 0.2055,
+      "step": 4623
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.285517690071861e-05,
+      "loss": 20.3788,
+      "step": 4624
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.284369920352474e-05,
+      "loss": 22.8577,
+      "step": 4625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.283221998483445e-05,
+      "loss": 21.0449,
+      "step": 4626
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.282073924541232e-05,
+      "loss": 0.1993,
+      "step": 4627
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.280925698602301e-05,
+      "loss": 19.0173,
+      "step": 4628
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.279777320743131e-05,
+      "loss": 1.0034,
+      "step": 4629
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.278628791040208e-05,
+      "loss": 0.9263,
+      "step": 4630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.277480109570032e-05,
+      "loss": 20.0576,
+      "step": 4631
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.276331276409106e-05,
+      "loss": 0.2239,
+      "step": 4632
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.27518229163395e-05,
+      "loss": 0.1952,
+      "step": 4633
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.274033155321094e-05,
+      "loss": 20.8874,
+      "step": 4634
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.272883867547072e-05,
+      "loss": 22.7619,
+      "step": 4635
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.271734428388434e-05,
+      "loss": 20.9402,
+      "step": 4636
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.270584837921737e-05,
+      "loss": 22.8055,
+      "step": 4637
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.269435096223549e-05,
+      "loss": 21.649,
+      "step": 4638
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.26828520337045e-05,
+      "loss": 21.0433,
+      "step": 4639
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.267135159439025e-05,
+      "loss": 0.199,
+      "step": 4640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.265984964505876e-05,
+      "loss": 21.2072,
+      "step": 4641
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.264834618647609e-05,
+      "loss": 21.8425,
+      "step": 4642
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.263684121940842e-05,
+      "loss": 21.0368,
+      "step": 4643
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.262533474462205e-05,
+      "loss": 0.1833,
+      "step": 4644
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.261382676288335e-05,
+      "loss": 0.9472,
+      "step": 4645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.260231727495882e-05,
+      "loss": 0.205,
+      "step": 4646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.259080628161504e-05,
+      "loss": 21.3769,
+      "step": 4647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.257929378361869e-05,
+      "loss": 0.9387,
+      "step": 4648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.256777978173655e-05,
+      "loss": 0.9656,
+      "step": 4649
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.255626427673552e-05,
+      "loss": 19.8395,
+      "step": 4650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.254474726938258e-05,
+      "loss": 22.0356,
+      "step": 4651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.253322876044481e-05,
+      "loss": 0.9258,
+      "step": 4652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.252170875068942e-05,
+      "loss": 0.9752,
+      "step": 4653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.251018724088367e-05,
+      "loss": 22.4562,
+      "step": 4654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.249866423179496e-05,
+      "loss": 0.1695,
+      "step": 4655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.248713972419077e-05,
+      "loss": 23.2231,
+      "step": 4656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.247561371883867e-05,
+      "loss": 20.0078,
+      "step": 4657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.246408621650638e-05,
+      "loss": 22.6278,
+      "step": 4658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.245255721796168e-05,
+      "loss": 22.2358,
+      "step": 4659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.244102672397241e-05,
+      "loss": 0.9547,
+      "step": 4660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.242949473530662e-05,
+      "loss": 20.7423,
+      "step": 4661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.241796125273233e-05,
+      "loss": 0.9404,
+      "step": 4662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.240642627701776e-05,
+      "loss": 0.2091,
+      "step": 4663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.23948898089312e-05,
+      "loss": 24.0314,
+      "step": 4664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.238335184924101e-05,
+      "loss": 23.1002,
+      "step": 4665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.237181239871568e-05,
+      "loss": 21.2551,
+      "step": 4666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.236027145812381e-05,
+      "loss": 20.0731,
+      "step": 4667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.234872902823405e-05,
+      "loss": 0.9336,
+      "step": 4668
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.23371851098152e-05,
+      "loss": 0.9563,
+      "step": 4669
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.232563970363611e-05,
+      "loss": 0.9515,
+      "step": 4670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.23140928104658e-05,
+      "loss": 0.2026,
+      "step": 4671
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.23025444310733e-05,
+      "loss": 0.1748,
+      "step": 4672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.229099456622784e-05,
+      "loss": 19.9094,
+      "step": 4673
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.227944321669864e-05,
+      "loss": 0.9581,
+      "step": 4674
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.226789038325511e-05,
+      "loss": 0.2117,
+      "step": 4675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.22563360666667e-05,
+      "loss": 23.409,
+      "step": 4676
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.224478026770301e-05,
+      "loss": 21.1891,
+      "step": 4677
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.223322298713369e-05,
+      "loss": 22.3794,
+      "step": 4678
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.22216642257285e-05,
+      "loss": 0.9399,
+      "step": 4679
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.221010398425731e-05,
+      "loss": 0.9356,
+      "step": 4680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.219854226349012e-05,
+      "loss": 23.6199,
+      "step": 4681
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.218697906419693e-05,
+      "loss": 0.2256,
+      "step": 4682
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.217541438714797e-05,
+      "loss": 0.2197,
+      "step": 4683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.216384823311346e-05,
+      "loss": 22.9767,
+      "step": 4684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.215228060286378e-05,
+      "loss": 21.3297,
+      "step": 4685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.214071149716938e-05,
+      "loss": 22.4674,
+      "step": 4686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.21291409168008e-05,
+      "loss": 21.3561,
+      "step": 4687
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.211756886252871e-05,
+      "loss": 0.2389,
+      "step": 4688
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.210599533512388e-05,
+      "loss": 0.9812,
+      "step": 4689
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.209442033535712e-05,
+      "loss": 0.1962,
+      "step": 4690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.208284386399942e-05,
+      "loss": 0.1788,
+      "step": 4691
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.207126592182179e-05,
+      "loss": 21.7747,
+      "step": 4692
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.205968650959541e-05,
+      "loss": 23.1955,
+      "step": 4693
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.20481056280915e-05,
+      "loss": 0.9402,
+      "step": 4694
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.20365232780814e-05,
+      "loss": 22.1522,
+      "step": 4695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.202493946033656e-05,
+      "loss": 22.0007,
+      "step": 4696
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.201335417562851e-05,
+      "loss": 0.2225,
+      "step": 4697
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.200176742472888e-05,
+      "loss": 20.134,
+      "step": 4698
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.19901792084094e-05,
+      "loss": 0.9643,
+      "step": 4699
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.197858952744193e-05,
+      "loss": 21.6606,
+      "step": 4700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.196699838259834e-05,
+      "loss": 24.9522,
+      "step": 4701
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.19554057746507e-05,
+      "loss": 19.839,
+      "step": 4702
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.194381170437112e-05,
+      "loss": 21.2958,
+      "step": 4703
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.19322161725318e-05,
+      "loss": 0.1875,
+      "step": 4704
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.192061917990506e-05,
+      "loss": 0.9197,
+      "step": 4705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.190902072726335e-05,
+      "loss": 21.2991,
+      "step": 4706
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.189742081537916e-05,
+      "loss": 21.9889,
+      "step": 4707
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.18858194450251e-05,
+      "loss": 21.9881,
+      "step": 4708
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.187421661697384e-05,
+      "loss": 22.3115,
+      "step": 4709
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.186261233199821e-05,
+      "loss": 0.2246,
+      "step": 4710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.185100659087113e-05,
+      "loss": 23.0774,
+      "step": 4711
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.183939939436556e-05,
+      "loss": 21.5291,
+      "step": 4712
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.182779074325464e-05,
+      "loss": 0.2304,
+      "step": 4713
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.18161806383115e-05,
+      "loss": 20.3705,
+      "step": 4714
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.180456908030948e-05,
+      "loss": 21.5942,
+      "step": 4715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.179295607002192e-05,
+      "loss": 19.6528,
+      "step": 4716
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.178134160822232e-05,
+      "loss": 20.2481,
+      "step": 4717
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.176972569568427e-05,
+      "loss": 1.0299,
+      "step": 4718
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.175810833318144e-05,
+      "loss": 21.9982,
+      "step": 4719
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.174648952148757e-05,
+      "loss": 20.654,
+      "step": 4720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.173486926137656e-05,
+      "loss": 22.9142,
+      "step": 4721
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.172324755362237e-05,
+      "loss": 0.1654,
+      "step": 4722
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.171162439899904e-05,
+      "loss": 21.2188,
+      "step": 4723
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.169999979828072e-05,
+      "loss": 19.7626,
+      "step": 4724
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.168837375224172e-05,
+      "loss": 0.9139,
+      "step": 4725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.167674626165631e-05,
+      "loss": 0.208,
+      "step": 4726
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1665117327299e-05,
+      "loss": 20.5662,
+      "step": 4727
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.165348694994429e-05,
+      "loss": 0.9614,
+      "step": 4728
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.164185513036683e-05,
+      "loss": 0.1941,
+      "step": 4729
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.163022186934136e-05,
+      "loss": 20.9633,
+      "step": 4730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.16185871676427e-05,
+      "loss": 0.9933,
+      "step": 4731
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.160695102604579e-05,
+      "loss": 0.9628,
+      "step": 4732
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.159531344532564e-05,
+      "loss": 19.7733,
+      "step": 4733
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.158367442625735e-05,
+      "loss": 18.708,
+      "step": 4734
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.157203396961617e-05,
+      "loss": 22.2963,
+      "step": 4735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.156039207617738e-05,
+      "loss": 19.8846,
+      "step": 4736
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.154874874671639e-05,
+      "loss": 22.8424,
+      "step": 4737
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.15371039820087e-05,
+      "loss": 21.3064,
+      "step": 4738
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.15254577828299e-05,
+      "loss": 21.7267,
+      "step": 4739
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.151381014995571e-05,
+      "loss": 0.226,
+      "step": 4740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.150216108416189e-05,
+      "loss": 19.9082,
+      "step": 4741
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.149051058622431e-05,
+      "loss": 0.9646,
+      "step": 4742
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.147885865691899e-05,
+      "loss": 0.9282,
+      "step": 4743
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.146720529702196e-05,
+      "loss": 20.5446,
+      "step": 4744
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.14555505073094e-05,
+      "loss": 0.9089,
+      "step": 4745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.144389428855758e-05,
+      "loss": 22.8247,
+      "step": 4746
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.143223664154286e-05,
+      "loss": 21.92,
+      "step": 4747
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.142057756704168e-05,
+      "loss": 0.9884,
+      "step": 4748
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.140891706583061e-05,
+      "loss": 19.3786,
+      "step": 4749
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.139725513868628e-05,
+      "loss": 20.7023,
+      "step": 4750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.138559178638542e-05,
+      "loss": 23.0956,
+      "step": 4751
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.137392700970489e-05,
+      "loss": 0.1719,
+      "step": 4752
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.136226080942158e-05,
+      "loss": 0.2295,
+      "step": 4753
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.135059318631254e-05,
+      "loss": 1.0,
+      "step": 4754
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.13389241411549e-05,
+      "loss": 0.227,
+      "step": 4755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.132725367472584e-05,
+      "loss": 19.7738,
+      "step": 4756
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.131558178780267e-05,
+      "loss": 21.6392,
+      "step": 4757
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.130390848116283e-05,
+      "loss": 0.2149,
+      "step": 4758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.129223375558377e-05,
+      "loss": 1.0184,
+      "step": 4759
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.128055761184313e-05,
+      "loss": 0.2039,
+      "step": 4760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.126888005071852e-05,
+      "loss": 0.9633,
+      "step": 4761
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.12572010729878e-05,
+      "loss": 0.9371,
+      "step": 4762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.124552067942881e-05,
+      "loss": 0.19,
+      "step": 4763
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.12338388708195e-05,
+      "loss": 23.4718,
+      "step": 4764
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.122215564793798e-05,
+      "loss": 22.8639,
+      "step": 4765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.121047101156236e-05,
+      "loss": 21.9738,
+      "step": 4766
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.119878496247091e-05,
+      "loss": 21.4372,
+      "step": 4767
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.118709750144197e-05,
+      "loss": 21.7508,
+      "step": 4768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.117540862925399e-05,
+      "loss": 0.175,
+      "step": 4769
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.11637183466855e-05,
+      "loss": 0.2087,
+      "step": 4770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.115202665451513e-05,
+      "loss": 0.9597,
+      "step": 4771
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.114033355352158e-05,
+      "loss": 23.3012,
+      "step": 4772
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.112863904448369e-05,
+      "loss": 0.8807,
+      "step": 4773
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.111694312818036e-05,
+      "loss": 19.5549,
+      "step": 4774
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.110524580539057e-05,
+      "loss": 23.0638,
+      "step": 4775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.109354707689346e-05,
+      "loss": 0.9732,
+      "step": 4776
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.108184694346819e-05,
+      "loss": 19.8056,
+      "step": 4777
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.107014540589405e-05,
+      "loss": 0.928,
+      "step": 4778
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.105844246495042e-05,
+      "loss": 0.9308,
+      "step": 4779
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.104673812141675e-05,
+      "loss": 0.9726,
+      "step": 4780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.103503237607263e-05,
+      "loss": 22.1824,
+      "step": 4781
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.102332522969772e-05,
+      "loss": 0.2304,
+      "step": 4782
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.101161668307173e-05,
+      "loss": 0.2001,
+      "step": 4783
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.099990673697456e-05,
+      "loss": 0.922,
+      "step": 4784
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.09881953921861e-05,
+      "loss": 0.9555,
+      "step": 4785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.09764826494864e-05,
+      "loss": 0.9453,
+      "step": 4786
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.096476850965559e-05,
+      "loss": 0.9281,
+      "step": 4787
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.095305297347386e-05,
+      "loss": 22.6879,
+      "step": 4788
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.094133604172155e-05,
+      "loss": 20.9352,
+      "step": 4789
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.092961771517905e-05,
+      "loss": 0.994,
+      "step": 4790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.091789799462683e-05,
+      "loss": 21.5697,
+      "step": 4791
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.090617688084553e-05,
+      "loss": 0.1988,
+      "step": 4792
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.089445437461579e-05,
+      "loss": 0.2302,
+      "step": 4793
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.088273047671838e-05,
+      "loss": 0.918,
+      "step": 4794
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.087100518793421e-05,
+      "loss": 0.2021,
+      "step": 4795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.085927850904418e-05,
+      "loss": 23.0104,
+      "step": 4796
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.084755044082939e-05,
+      "loss": 0.9491,
+      "step": 4797
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.083582098407095e-05,
+      "loss": 21.638,
+      "step": 4798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.082409013955012e-05,
+      "loss": 19.9513,
+      "step": 4799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.081235790804821e-05,
+      "loss": 0.9004,
+      "step": 4800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.080062429034666e-05,
+      "loss": 22.2668,
+      "step": 4801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.078888928722697e-05,
+      "loss": 0.9284,
+      "step": 4802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.077715289947076e-05,
+      "loss": 0.2122,
+      "step": 4803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.07654151278597e-05,
+      "loss": 21.0521,
+      "step": 4804
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.07536759731756e-05,
+      "loss": 20.38,
+      "step": 4805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.074193543620036e-05,
+      "loss": 20.651,
+      "step": 4806
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.073019351771593e-05,
+      "loss": 19.8315,
+      "step": 4807
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.071845021850435e-05,
+      "loss": 0.9366,
+      "step": 4808
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.070670553934782e-05,
+      "loss": 21.3767,
+      "step": 4809
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.069495948102859e-05,
+      "loss": 20.9772,
+      "step": 4810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.068321204432898e-05,
+      "loss": 0.9637,
+      "step": 4811
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.067146323003146e-05,
+      "loss": 0.9089,
+      "step": 4812
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.06597130389185e-05,
+      "loss": 0.2198,
+      "step": 4813
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.064796147177275e-05,
+      "loss": 20.5078,
+      "step": 4814
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.063620852937693e-05,
+      "loss": 0.1904,
+      "step": 4815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.062445421251381e-05,
+      "loss": 19.4426,
+      "step": 4816
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.061269852196632e-05,
+      "loss": 23.0747,
+      "step": 4817
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.060094145851741e-05,
+      "loss": 21.6027,
+      "step": 4818
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.058918302295017e-05,
+      "loss": 23.522,
+      "step": 4819
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.057742321604776e-05,
+      "loss": 21.9488,
+      "step": 4820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.056566203859343e-05,
+      "loss": 0.9209,
+      "step": 4821
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.055389949137057e-05,
+      "loss": 0.9414,
+      "step": 4822
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.054213557516258e-05,
+      "loss": 0.9566,
+      "step": 4823
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0530370290753e-05,
+      "loss": 22.1282,
+      "step": 4824
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.051860363892546e-05,
+      "loss": 0.9267,
+      "step": 4825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.050683562046366e-05,
+      "loss": 20.0925,
+      "step": 4826
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.04950662361514e-05,
+      "loss": 0.9295,
+      "step": 4827
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.04832954867726e-05,
+      "loss": 0.9727,
+      "step": 4828
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.047152337311123e-05,
+      "loss": 20.4955,
+      "step": 4829
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.045974989595139e-05,
+      "loss": 0.9271,
+      "step": 4830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.044797505607722e-05,
+      "loss": 21.5379,
+      "step": 4831
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.043619885427299e-05,
+      "loss": 18.9911,
+      "step": 4832
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.042442129132305e-05,
+      "loss": 0.9151,
+      "step": 4833
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.041264236801185e-05,
+      "loss": 0.9455,
+      "step": 4834
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.040086208512389e-05,
+      "loss": 21.0896,
+      "step": 4835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.038908044344381e-05,
+      "loss": 0.9463,
+      "step": 4836
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.037729744375635e-05,
+      "loss": 0.9523,
+      "step": 4837
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.036551308684627e-05,
+      "loss": 0.2189,
+      "step": 4838
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.035372737349849e-05,
+      "loss": 21.3782,
+      "step": 4839
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.034194030449797e-05,
+      "loss": 20.7389,
+      "step": 4840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.03301518806298e-05,
+      "loss": 22.7116,
+      "step": 4841
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.031836210267915e-05,
+      "loss": 0.9087,
+      "step": 4842
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.030657097143126e-05,
+      "loss": 20.9895,
+      "step": 4843
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.029477848767148e-05,
+      "loss": 20.598,
+      "step": 4844
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.028298465218523e-05,
+      "loss": 23.1891,
+      "step": 4845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.027118946575804e-05,
+      "loss": 18.6317,
+      "step": 4846
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.025939292917556e-05,
+      "loss": 0.9533,
+      "step": 4847
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.024759504322344e-05,
+      "loss": 19.4015,
+      "step": 4848
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.023579580868752e-05,
+      "loss": 20.8561,
+      "step": 4849
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.022399522635365e-05,
+      "loss": 0.1722,
+      "step": 4850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.02121932970078e-05,
+      "loss": 20.2457,
+      "step": 4851
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.020039002143609e-05,
+      "loss": 0.2094,
+      "step": 4852
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.01885854004246e-05,
+      "loss": 20.3251,
+      "step": 4853
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.017677943475961e-05,
+      "loss": 23.4287,
+      "step": 4854
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.016497212522746e-05,
+      "loss": 22.5335,
+      "step": 4855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.015316347261455e-05,
+      "loss": 22.0876,
+      "step": 4856
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.01413534777074e-05,
+      "loss": 22.0572,
+      "step": 4857
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.01295421412926e-05,
+      "loss": 0.1992,
+      "step": 4858
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.011772946415688e-05,
+      "loss": 20.05,
+      "step": 4859
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.010591544708696e-05,
+      "loss": 0.2113,
+      "step": 4860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.009410009086974e-05,
+      "loss": 0.2091,
+      "step": 4861
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.008228339629219e-05,
+      "loss": 0.2299,
+      "step": 4862
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.007046536414133e-05,
+      "loss": 21.0033,
+      "step": 4863
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.005864599520432e-05,
+      "loss": 22.678,
+      "step": 4864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.004682529026836e-05,
+      "loss": 20.6857,
+      "step": 4865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.003500325012078e-05,
+      "loss": 20.8739,
+      "step": 4866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.002317987554899e-05,
+      "loss": 20.1361,
+      "step": 4867
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.001135516734045e-05,
+      "loss": 20.3282,
+      "step": 4868
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.999952912628278e-05,
+      "loss": 19.6515,
+      "step": 4869
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.998770175316361e-05,
+      "loss": 22.4815,
+      "step": 4870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.997587304877074e-05,
+      "loss": 0.9247,
+      "step": 4871
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.996404301389198e-05,
+      "loss": 21.9177,
+      "step": 4872
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.995221164931529e-05,
+      "loss": 0.9527,
+      "step": 4873
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.994037895582868e-05,
+      "loss": 0.9995,
+      "step": 4874
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.992854493422026e-05,
+      "loss": 0.9224,
+      "step": 4875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.991670958527824e-05,
+      "loss": 21.3111,
+      "step": 4876
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.990487290979092e-05,
+      "loss": 0.2269,
+      "step": 4877
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.989303490854665e-05,
+      "loss": 0.1937,
+      "step": 4878
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.988119558233392e-05,
+      "loss": 21.3152,
+      "step": 4879
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.986935493194125e-05,
+      "loss": 19.4539,
+      "step": 4880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.985751295815733e-05,
+      "loss": 0.9612,
+      "step": 4881
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.984566966177087e-05,
+      "loss": 20.5609,
+      "step": 4882
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.983382504357066e-05,
+      "loss": 0.2302,
+      "step": 4883
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.982197910434565e-05,
+      "loss": 0.9426,
+      "step": 4884
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.98101318448848e-05,
+      "loss": 23.2239,
+      "step": 4885
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.979828326597724e-05,
+      "loss": 0.9417,
+      "step": 4886
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.978643336841209e-05,
+      "loss": 20.3243,
+      "step": 4887
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.977458215297861e-05,
+      "loss": 20.5175,
+      "step": 4888
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.976272962046619e-05,
+      "loss": 20.2471,
+      "step": 4889
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.975087577166423e-05,
+      "loss": 20.956,
+      "step": 4890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.973902060736226e-05,
+      "loss": 0.1901,
+      "step": 4891
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.972716412834987e-05,
+      "loss": 0.9879,
+      "step": 4892
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.971530633541678e-05,
+      "loss": 23.729,
+      "step": 4893
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.970344722935279e-05,
+      "loss": 21.1619,
+      "step": 4894
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.969158681094771e-05,
+      "loss": 0.2485,
+      "step": 4895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.967972508099158e-05,
+      "loss": 20.5703,
+      "step": 4896
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.966786204027439e-05,
+      "loss": 0.2324,
+      "step": 4897
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.965599768958627e-05,
+      "loss": 0.967,
+      "step": 4898
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.964413202971748e-05,
+      "loss": 0.976,
+      "step": 4899
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.96322650614583e-05,
+      "loss": 0.9572,
+      "step": 4900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.962039678559913e-05,
+      "loss": 22.9166,
+      "step": 4901
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.960852720293046e-05,
+      "loss": 20.1079,
+      "step": 4902
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.959665631424285e-05,
+      "loss": 21.4582,
+      "step": 4903
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.958478412032695e-05,
+      "loss": 20.4295,
+      "step": 4904
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.957291062197351e-05,
+      "loss": 19.1751,
+      "step": 4905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.956103581997339e-05,
+      "loss": 0.9526,
+      "step": 4906
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.954915971511747e-05,
+      "loss": 0.1811,
+      "step": 4907
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.953728230819675e-05,
+      "loss": 0.1911,
+      "step": 4908
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.952540360000235e-05,
+      "loss": 0.9047,
+      "step": 4909
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.951352359132542e-05,
+      "loss": 0.221,
+      "step": 4910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.950164228295724e-05,
+      "loss": 20.7446,
+      "step": 4911
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.948975967568916e-05,
+      "loss": 21.4611,
+      "step": 4912
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.947787577031259e-05,
+      "loss": 21.959,
+      "step": 4913
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.946599056761911e-05,
+      "loss": 22.2408,
+      "step": 4914
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.945410406840028e-05,
+      "loss": 0.1955,
+      "step": 4915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.944221627344782e-05,
+      "loss": 0.9111,
+      "step": 4916
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.943032718355346e-05,
+      "loss": 22.8122,
+      "step": 4917
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.941843679950917e-05,
+      "loss": 19.0215,
+      "step": 4918
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.940654512210682e-05,
+      "loss": 21.8878,
+      "step": 4919
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.939465215213848e-05,
+      "loss": 20.9515,
+      "step": 4920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.938275789039628e-05,
+      "loss": 22.1682,
+      "step": 4921
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.937086233767241e-05,
+      "loss": 20.9263,
+      "step": 4922
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.93589654947592e-05,
+      "loss": 19.5853,
+      "step": 4923
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.934706736244902e-05,
+      "loss": 20.0088,
+      "step": 4924
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.933516794153434e-05,
+      "loss": 21.5755,
+      "step": 4925
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.932326723280773e-05,
+      "loss": 0.9425,
+      "step": 4926
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.931136523706181e-05,
+      "loss": 0.2089,
+      "step": 4927
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.929946195508932e-05,
+      "loss": 0.975,
+      "step": 4928
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.928755738768309e-05,
+      "loss": 22.0605,
+      "step": 4929
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.927565153563599e-05,
+      "loss": 0.9611,
+      "step": 4930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.926374439974102e-05,
+      "loss": 21.9161,
+      "step": 4931
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.925183598079124e-05,
+      "loss": 18.183,
+      "step": 4932
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.923992627957983e-05,
+      "loss": 21.5766,
+      "step": 4933
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.922801529690001e-05,
+      "loss": 1.0093,
+      "step": 4934
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.921610303354511e-05,
+      "loss": 21.3059,
+      "step": 4935
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.920418949030856e-05,
+      "loss": 0.9833,
+      "step": 4936
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.919227466798382e-05,
+      "loss": 0.9397,
+      "step": 4937
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.91803585673645e-05,
+      "loss": 23.2967,
+      "step": 4938
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.916844118924428e-05,
+      "loss": 20.5359,
+      "step": 4939
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.915652253441687e-05,
+      "loss": 22.6364,
+      "step": 4940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.914460260367616e-05,
+      "loss": 21.8657,
+      "step": 4941
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.913268139781601e-05,
+      "loss": 0.9862,
+      "step": 4942
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.91207589176305e-05,
+      "loss": 20.9132,
+      "step": 4943
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.910883516391367e-05,
+      "loss": 20.1978,
+      "step": 4944
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.90969101374597e-05,
+      "loss": 21.8293,
+      "step": 4945
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.908498383906289e-05,
+      "loss": 0.9371,
+      "step": 4946
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.907305626951755e-05,
+      "loss": 20.9634,
+      "step": 4947
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.906112742961812e-05,
+      "loss": 19.7112,
+      "step": 4948
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.904919732015912e-05,
+      "loss": 21.7599,
+      "step": 4949
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.903726594193515e-05,
+      "loss": 21.099,
+      "step": 4950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.90253332957409e-05,
+      "loss": 1.0074,
+      "step": 4951
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.901339938237114e-05,
+      "loss": 0.208,
+      "step": 4952
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.90014642026207e-05,
+      "loss": 0.9349,
+      "step": 4953
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.898952775728456e-05,
+      "loss": 0.9856,
+      "step": 4954
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.897759004715771e-05,
+      "loss": 20.0553,
+      "step": 4955
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.896565107303526e-05,
+      "loss": 0.9686,
+      "step": 4956
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.895371083571241e-05,
+      "loss": 0.2289,
+      "step": 4957
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.894176933598443e-05,
+      "loss": 0.9014,
+      "step": 4958
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.89298265746467e-05,
+      "loss": 19.288,
+      "step": 4959
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.891788255249465e-05,
+      "loss": 0.934,
+      "step": 4960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.890593727032379e-05,
+      "loss": 0.2312,
+      "step": 4961
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.889399072892974e-05,
+      "loss": 20.2357,
+      "step": 4962
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.888204292910822e-05,
+      "loss": 20.0569,
+      "step": 4963
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.887009387165495e-05,
+      "loss": 0.1839,
+      "step": 4964
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.885814355736586e-05,
+      "loss": 0.9127,
+      "step": 4965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.884619198703688e-05,
+      "loss": 0.9385,
+      "step": 4966
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.883423916146399e-05,
+      "loss": 0.2256,
+      "step": 4967
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.882228508144337e-05,
+      "loss": 0.9219,
+      "step": 4968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.881032974777118e-05,
+      "loss": 20.6363,
+      "step": 4969
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.879837316124371e-05,
+      "loss": 0.9659,
+      "step": 4970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.878641532265733e-05,
+      "loss": 22.2342,
+      "step": 4971
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.877445623280847e-05,
+      "loss": 0.19,
+      "step": 4972
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.876249589249368e-05,
+      "loss": 0.9611,
+      "step": 4973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.875053430250956e-05,
+      "loss": 0.2541,
+      "step": 4974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.873857146365281e-05,
+      "loss": 0.2063,
+      "step": 4975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.872660737672023e-05,
+      "loss": 0.9335,
+      "step": 4976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.871464204250866e-05,
+      "loss": 0.9447,
+      "step": 4977
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.870267546181504e-05,
+      "loss": 0.9523,
+      "step": 4978
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.869070763543643e-05,
+      "loss": 22.5674,
+      "step": 4979
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.867873856416993e-05,
+      "loss": 21.1854,
+      "step": 4980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.866676824881273e-05,
+      "loss": 19.142,
+      "step": 4981
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.865479669016212e-05,
+      "loss": 20.5167,
+      "step": 4982
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.864282388901544e-05,
+      "loss": 0.204,
+      "step": 4983
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.863084984617015e-05,
+      "loss": 21.7162,
+      "step": 4984
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.861887456242377e-05,
+      "loss": 0.8927,
+      "step": 4985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.860689803857392e-05,
+      "loss": 0.2108,
+      "step": 4986
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.859492027541829e-05,
+      "loss": 0.9342,
+      "step": 4987
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.858294127375465e-05,
+      "loss": 21.4851,
+      "step": 4988
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.857096103438083e-05,
+      "loss": 0.9695,
+      "step": 4989
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.855897955809484e-05,
+      "loss": 21.2459,
+      "step": 4990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.854699684569462e-05,
+      "loss": 21.9701,
+      "step": 4991
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.853501289797832e-05,
+      "loss": 20.2478,
+      "step": 4992
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.852302771574415e-05,
+      "loss": 0.8956,
+      "step": 4993
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.851104129979031e-05,
+      "loss": 0.1663,
+      "step": 4994
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.849905365091522e-05,
+      "loss": 20.8943,
+      "step": 4995
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.848706476991728e-05,
+      "loss": 20.6944,
+      "step": 4996
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.847507465759499e-05,
+      "loss": 0.9872,
+      "step": 4997
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.846308331474697e-05,
+      "loss": 0.2113,
+      "step": 4998
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.84510907421719e-05,
+      "loss": 22.462,
+      "step": 4999
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.843909694066853e-05,
+      "loss": 21.3202,
+      "step": 5000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.842710191103571e-05,
+      "loss": 0.9268,
+      "step": 5001
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.841510565407235e-05,
+      "loss": 0.9265,
+      "step": 5002
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.840310817057749e-05,
+      "loss": 21.1755,
+      "step": 5003
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.839110946135018e-05,
+      "loss": 0.9677,
+      "step": 5004
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.837910952718962e-05,
+      "loss": 0.1841,
+      "step": 5005
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.836710836889504e-05,
+      "loss": 23.3232,
+      "step": 5006
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.835510598726578e-05,
+      "loss": 21.044,
+      "step": 5007
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.834310238310127e-05,
+      "loss": 0.2035,
+      "step": 5008
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.833109755720096e-05,
+      "loss": 22.0736,
+      "step": 5009
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.831909151036448e-05,
+      "loss": 0.9373,
+      "step": 5010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.830708424339146e-05,
+      "loss": 21.2773,
+      "step": 5011
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.829507575708165e-05,
+      "loss": 19.9992,
+      "step": 5012
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.828306605223487e-05,
+      "loss": 22.0168,
+      "step": 5013
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8271055129651e-05,
+      "loss": 0.9436,
+      "step": 5014
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.825904299013008e-05,
+      "loss": 23.4504,
+      "step": 5015
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.82470296344721e-05,
+      "loss": 1.0025,
+      "step": 5016
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.823501506347725e-05,
+      "loss": 0.978,
+      "step": 5017
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.822299927794576e-05,
+      "loss": 0.8951,
+      "step": 5018
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.821098227867791e-05,
+      "loss": 0.9812,
+      "step": 5019
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.819896406647413e-05,
+      "loss": 19.8973,
+      "step": 5020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.818694464213485e-05,
+      "loss": 0.9486,
+      "step": 5021
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.817492400646062e-05,
+      "loss": 0.9251,
+      "step": 5022
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.816290216025208e-05,
+      "loss": 0.9472,
+      "step": 5023
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.815087910430994e-05,
+      "loss": 19.7697,
+      "step": 5024
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8138854839435e-05,
+      "loss": 0.9451,
+      "step": 5025
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.812682936642813e-05,
+      "loss": 20.1241,
+      "step": 5026
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.811480268609028e-05,
+      "loss": 0.2341,
+      "step": 5027
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.810277479922247e-05,
+      "loss": 0.2135,
+      "step": 5028
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.809074570662584e-05,
+      "loss": 21.5461,
+      "step": 5029
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.807871540910154e-05,
+      "loss": 0.8911,
+      "step": 5030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.80666839074509e-05,
+      "loss": 0.186,
+      "step": 5031
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.805465120247523e-05,
+      "loss": 0.2153,
+      "step": 5032
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.804261729497597e-05,
+      "loss": 21.3498,
+      "step": 5033
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.803058218575465e-05,
+      "loss": 0.9549,
+      "step": 5034
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.801854587561286e-05,
+      "loss": 0.2076,
+      "step": 5035
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.800650836535227e-05,
+      "loss": 0.9475,
+      "step": 5036
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.799446965577464e-05,
+      "loss": 0.9539,
+      "step": 5037
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.798242974768179e-05,
+      "loss": 0.914,
+      "step": 5038
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.797038864187564e-05,
+      "loss": 0.9343,
+      "step": 5039
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.79583463391582e-05,
+      "loss": 0.9482,
+      "step": 5040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.794630284033152e-05,
+      "loss": 22.0874,
+      "step": 5041
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.793425814619777e-05,
+      "loss": 20.9464,
+      "step": 5042
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.792221225755916e-05,
+      "loss": 21.5893,
+      "step": 5043
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.791016517521802e-05,
+      "loss": 19.1999,
+      "step": 5044
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.789811689997677e-05,
+      "loss": 0.9769,
+      "step": 5045
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.78860674326378e-05,
+      "loss": 0.9376,
+      "step": 5046
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.787401677400374e-05,
+      "loss": 0.9856,
+      "step": 5047
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.786196492487719e-05,
+      "loss": 23.8049,
+      "step": 5048
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.784991188606085e-05,
+      "loss": 20.317,
+      "step": 5049
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.783785765835752e-05,
+      "loss": 0.9442,
+      "step": 5050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.782580224257007e-05,
+      "loss": 21.3511,
+      "step": 5051
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.781374563950142e-05,
+      "loss": 0.9484,
+      "step": 5052
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.780168784995464e-05,
+      "loss": 0.251,
+      "step": 5053
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.778962887473282e-05,
+      "loss": 21.2512,
+      "step": 5054
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.777756871463911e-05,
+      "loss": 20.6945,
+      "step": 5055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.776550737047681e-05,
+      "loss": 0.2004,
+      "step": 5056
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.775344484304925e-05,
+      "loss": 0.1965,
+      "step": 5057
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.774138113315985e-05,
+      "loss": 20.7926,
+      "step": 5058
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.772931624161212e-05,
+      "loss": 20.8885,
+      "step": 5059
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.77172501692096e-05,
+      "loss": 20.2101,
+      "step": 5060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.770518291675599e-05,
+      "loss": 22.7248,
+      "step": 5061
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.769311448505503e-05,
+      "loss": 20.0051,
+      "step": 5062
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.768104487491047e-05,
+      "loss": 0.9364,
+      "step": 5063
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.766897408712627e-05,
+      "loss": 0.9755,
+      "step": 5064
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.765690212250637e-05,
+      "loss": 0.9636,
+      "step": 5065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.764482898185482e-05,
+      "loss": 0.9598,
+      "step": 5066
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.763275466597577e-05,
+      "loss": 0.2316,
+      "step": 5067
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.762067917567338e-05,
+      "loss": 0.169,
+      "step": 5068
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.760860251175196e-05,
+      "loss": 0.9835,
+      "step": 5069
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.75965246750159e-05,
+      "loss": 0.205,
+      "step": 5070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.758444566626958e-05,
+      "loss": 0.2276,
+      "step": 5071
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.757236548631758e-05,
+      "loss": 0.9995,
+      "step": 5072
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.756028413596444e-05,
+      "loss": 19.0117,
+      "step": 5073
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.754820161601487e-05,
+      "loss": 0.9112,
+      "step": 5074
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.753611792727362e-05,
+      "loss": 20.3497,
+      "step": 5075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.752403307054549e-05,
+      "loss": 21.8362,
+      "step": 5076
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.751194704663543e-05,
+      "loss": 20.4911,
+      "step": 5077
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.749985985634841e-05,
+      "loss": 0.9608,
+      "step": 5078
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.748777150048946e-05,
+      "loss": 22.1615,
+      "step": 5079
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.747568197986378e-05,
+      "loss": 18.3047,
+      "step": 5080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.746359129527652e-05,
+      "loss": 21.4528,
+      "step": 5081
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.745149944753304e-05,
+      "loss": 0.9427,
+      "step": 5082
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.743940643743868e-05,
+      "loss": 20.2184,
+      "step": 5083
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.742731226579889e-05,
+      "loss": 21.3502,
+      "step": 5084
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.741521693341922e-05,
+      "loss": 20.4562,
+      "step": 5085
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.740312044110523e-05,
+      "loss": 0.9313,
+      "step": 5086
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.739102278966265e-05,
+      "loss": 0.2188,
+      "step": 5087
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.737892397989721e-05,
+      "loss": 0.9059,
+      "step": 5088
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.736682401261477e-05,
+      "loss": 22.8637,
+      "step": 5089
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.735472288862123e-05,
+      "loss": 21.4855,
+      "step": 5090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.734262060872258e-05,
+      "loss": 0.9512,
+      "step": 5091
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.73305171737249e-05,
+      "loss": 19.8766,
+      "step": 5092
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.731841258443432e-05,
+      "loss": 0.9495,
+      "step": 5093
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.730630684165707e-05,
+      "loss": 0.1875,
+      "step": 5094
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.729419994619946e-05,
+      "loss": 1.0128,
+      "step": 5095
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.728209189886781e-05,
+      "loss": 19.3415,
+      "step": 5096
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.726998270046866e-05,
+      "loss": 19.5605,
+      "step": 5097
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.725787235180846e-05,
+      "loss": 0.9626,
+      "step": 5098
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.724576085369388e-05,
+      "loss": 0.1975,
+      "step": 5099
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.723364820693154e-05,
+      "loss": 0.8816,
+      "step": 5100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.722153441232823e-05,
+      "loss": 19.9805,
+      "step": 5101
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.720941947069082e-05,
+      "loss": 22.4677,
+      "step": 5102
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.719730338282616e-05,
+      "loss": 0.1789,
+      "step": 5103
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.718518614954127e-05,
+      "loss": 22.714,
+      "step": 5104
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.717306777164321e-05,
+      "loss": 0.221,
+      "step": 5105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.71609482499391e-05,
+      "loss": 0.9806,
+      "step": 5106
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.714882758523618e-05,
+      "loss": 0.9865,
+      "step": 5107
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.713670577834175e-05,
+      "loss": 19.8247,
+      "step": 5108
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.712458283006316e-05,
+      "loss": 20.0912,
+      "step": 5109
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.711245874120784e-05,
+      "loss": 0.2203,
+      "step": 5110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.710033351258335e-05,
+      "loss": 19.7004,
+      "step": 5111
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.708820714499726e-05,
+      "loss": 0.2166,
+      "step": 5112
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.707607963925724e-05,
+      "loss": 21.1456,
+      "step": 5113
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.706395099617107e-05,
+      "loss": 0.9733,
+      "step": 5114
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.705182121654654e-05,
+      "loss": 0.9597,
+      "step": 5115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.703969030119155e-05,
+      "loss": 0.9742,
+      "step": 5116
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.702755825091412e-05,
+      "loss": 0.1954,
+      "step": 5117
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.701542506652223e-05,
+      "loss": 0.9003,
+      "step": 5118
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.700329074882406e-05,
+      "loss": 22.9414,
+      "step": 5119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.699115529862778e-05,
+      "loss": 0.9216,
+      "step": 5120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.69790187167417e-05,
+      "loss": 20.0259,
+      "step": 5121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.696688100397416e-05,
+      "loss": 0.9854,
+      "step": 5122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.695474216113356e-05,
+      "loss": 21.7902,
+      "step": 5123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.694260218902844e-05,
+      "loss": 0.9068,
+      "step": 5124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.693046108846737e-05,
+      "loss": 20.0379,
+      "step": 5125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.691831886025898e-05,
+      "loss": 19.6441,
+      "step": 5126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.690617550521205e-05,
+      "loss": 19.1155,
+      "step": 5127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.689403102413531e-05,
+      "loss": 20.8454,
+      "step": 5128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.688188541783771e-05,
+      "loss": 21.5609,
+      "step": 5129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.686973868712818e-05,
+      "loss": 0.22,
+      "step": 5130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.685759083281571e-05,
+      "loss": 21.2162,
+      "step": 5131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.684544185570945e-05,
+      "loss": 19.4635,
+      "step": 5132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.683329175661858e-05,
+      "loss": 0.2071,
+      "step": 5133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.682114053635232e-05,
+      "loss": 0.9473,
+      "step": 5134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.680898819572001e-05,
+      "loss": 19.7969,
+      "step": 5135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.679683473553108e-05,
+      "loss": 0.9789,
+      "step": 5136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.678468015659498e-05,
+      "loss": 0.2359,
+      "step": 5137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.677252445972125e-05,
+      "loss": 0.9482,
+      "step": 5138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.676036764571954e-05,
+      "loss": 0.1896,
+      "step": 5139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.674820971539952e-05,
+      "loss": 0.9483,
+      "step": 5140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.673605066957102e-05,
+      "loss": 20.855,
+      "step": 5141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.672389050904384e-05,
+      "loss": 19.1072,
+      "step": 5142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.671172923462792e-05,
+      "loss": 0.2165,
+      "step": 5143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.669956684713326e-05,
+      "loss": 21.6802,
+      "step": 5144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.668740334736992e-05,
+      "loss": 0.1744,
+      "step": 5145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.667523873614805e-05,
+      "loss": 20.4301,
+      "step": 5146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.666307301427787e-05,
+      "loss": 20.8717,
+      "step": 5147
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.665090618256968e-05,
+      "loss": 23.4293,
+      "step": 5148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.663873824183386e-05,
+      "loss": 0.2424,
+      "step": 5149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.66265691928808e-05,
+      "loss": 21.5375,
+      "step": 5150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.661439903652108e-05,
+      "loss": 21.4111,
+      "step": 5151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.660222777356524e-05,
+      "loss": 21.9946,
+      "step": 5152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.659005540482397e-05,
+      "loss": 20.5102,
+      "step": 5153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6577881931108e-05,
+      "loss": 0.2157,
+      "step": 5154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.656570735322814e-05,
+      "loss": 18.1882,
+      "step": 5155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.655353167199525e-05,
+      "loss": 21.7468,
+      "step": 5156
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.654135488822033e-05,
+      "loss": 20.239,
+      "step": 5157
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.652917700271436e-05,
+      "loss": 0.1981,
+      "step": 5158
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.65169980162885e-05,
+      "loss": 0.9712,
+      "step": 5159
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.650481792975389e-05,
+      "loss": 0.21,
+      "step": 5160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.649263674392179e-05,
+      "loss": 20.6676,
+      "step": 5161
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.648045445960352e-05,
+      "loss": 0.2015,
+      "step": 5162
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.646827107761049e-05,
+      "loss": 0.2537,
+      "step": 5163
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.645608659875414e-05,
+      "loss": 0.9966,
+      "step": 5164
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.644390102384604e-05,
+      "loss": 21.0795,
+      "step": 5165
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.64317143536978e-05,
+      "loss": 21.631,
+      "step": 5166
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.64195265891211e-05,
+      "loss": 21.7756,
+      "step": 5167
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.640733773092771e-05,
+      "loss": 1.0046,
+      "step": 5168
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.639514777992948e-05,
+      "loss": 0.9153,
+      "step": 5169
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.638295673693828e-05,
+      "loss": 0.9702,
+      "step": 5170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.637076460276613e-05,
+      "loss": 20.2078,
+      "step": 5171
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.635857137822504e-05,
+      "loss": 22.004,
+      "step": 5172
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.634637706412715e-05,
+      "loss": 18.8666,
+      "step": 5173
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.63341816612847e-05,
+      "loss": 0.1763,
+      "step": 5174
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.63219851705099e-05,
+      "loss": 20.4779,
+      "step": 5175
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.630978759261513e-05,
+      "loss": 0.2151,
+      "step": 5176
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.62975889284128e-05,
+      "loss": 20.876,
+      "step": 5177
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.628538917871538e-05,
+      "loss": 0.1862,
+      "step": 5178
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.627318834433544e-05,
+      "loss": 21.2508,
+      "step": 5179
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.626098642608562e-05,
+      "loss": 20.3754,
+      "step": 5180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.624878342477863e-05,
+      "loss": 23.0111,
+      "step": 5181
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.623657934122722e-05,
+      "loss": 21.0559,
+      "step": 5182
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.622437417624426e-05,
+      "loss": 0.9352,
+      "step": 5183
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.621216793064267e-05,
+      "loss": 21.2249,
+      "step": 5184
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.619996060523543e-05,
+      "loss": 21.8991,
+      "step": 5185
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.618775220083562e-05,
+      "loss": 18.5558,
+      "step": 5186
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.617554271825636e-05,
+      "loss": 0.9206,
+      "step": 5187
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.616333215831086e-05,
+      "loss": 0.1995,
+      "step": 5188
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.615112052181243e-05,
+      "loss": 21.2087,
+      "step": 5189
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.613890780957438e-05,
+      "loss": 0.9566,
+      "step": 5190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.612669402241016e-05,
+      "loss": 20.8261,
+      "step": 5191
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.611447916113324e-05,
+      "loss": 0.9213,
+      "step": 5192
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.610226322655722e-05,
+      "loss": 0.1857,
+      "step": 5193
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.609004621949572e-05,
+      "loss": 22.8311,
+      "step": 5194
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.607782814076244e-05,
+      "loss": 20.6675,
+      "step": 5195
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.60656089911712e-05,
+      "loss": 20.4907,
+      "step": 5196
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.605338877153577e-05,
+      "loss": 0.2272,
+      "step": 5197
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.604116748267018e-05,
+      "loss": 18.8075,
+      "step": 5198
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.602894512538834e-05,
+      "loss": 0.1866,
+      "step": 5199
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.601672170050434e-05,
+      "loss": 20.4983,
+      "step": 5200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.600449720883235e-05,
+      "loss": 0.2236,
+      "step": 5201
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.599227165118655e-05,
+      "loss": 18.4422,
+      "step": 5202
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.59800450283812e-05,
+      "loss": 20.1446,
+      "step": 5203
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.596781734123068e-05,
+      "loss": 0.9278,
+      "step": 5204
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.595558859054939e-05,
+      "loss": 0.9312,
+      "step": 5205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.594335877715186e-05,
+      "loss": 0.8937,
+      "step": 5206
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.593112790185259e-05,
+      "loss": 0.206,
+      "step": 5207
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.591889596546628e-05,
+      "loss": 23.0926,
+      "step": 5208
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.590666296880758e-05,
+      "loss": 0.9434,
+      "step": 5209
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.58944289126913e-05,
+      "loss": 0.2038,
+      "step": 5210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.588219379793225e-05,
+      "loss": 20.5234,
+      "step": 5211
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.58699576253454e-05,
+      "loss": 20.0924,
+      "step": 5212
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.585772039574566e-05,
+      "loss": 21.413,
+      "step": 5213
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.584548210994816e-05,
+      "loss": 0.9038,
+      "step": 5214
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.583324276876798e-05,
+      "loss": 21.5758,
+      "step": 5215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.582100237302035e-05,
+      "loss": 0.9379,
+      "step": 5216
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.580876092352051e-05,
+      "loss": 20.3503,
+      "step": 5217
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.57965184210838e-05,
+      "loss": 0.2572,
+      "step": 5218
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.578427486652564e-05,
+      "loss": 0.2075,
+      "step": 5219
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.57720302606615e-05,
+      "loss": 0.2001,
+      "step": 5220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.575978460430695e-05,
+      "loss": 21.3935,
+      "step": 5221
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.574753789827755e-05,
+      "loss": 21.9571,
+      "step": 5222
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.573529014338904e-05,
+      "loss": 0.9493,
+      "step": 5223
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.572304134045717e-05,
+      "loss": 19.4339,
+      "step": 5224
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.571079149029775e-05,
+      "loss": 20.1036,
+      "step": 5225
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.569854059372669e-05,
+      "loss": 20.387,
+      "step": 5226
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.568628865155995e-05,
+      "loss": 0.9044,
+      "step": 5227
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.567403566461356e-05,
+      "loss": 20.7656,
+      "step": 5228
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.566178163370365e-05,
+      "loss": 0.9581,
+      "step": 5229
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.564952655964638e-05,
+      "loss": 20.2284,
+      "step": 5230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.563727044325798e-05,
+      "loss": 20.7987,
+      "step": 5231
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.562501328535479e-05,
+      "loss": 0.2459,
+      "step": 5232
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.561275508675317e-05,
+      "loss": 0.2508,
+      "step": 5233
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.56004958482696e-05,
+      "loss": 0.1967,
+      "step": 5234
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.558823557072057e-05,
+      "loss": 0.9659,
+      "step": 5235
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.557597425492272e-05,
+      "loss": 0.944,
+      "step": 5236
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.556371190169265e-05,
+      "loss": 0.9575,
+      "step": 5237
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.555144851184712e-05,
+      "loss": 22.7465,
+      "step": 5238
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.553918408620294e-05,
+      "loss": 0.9377,
+      "step": 5239
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.552691862557696e-05,
+      "loss": 0.971,
+      "step": 5240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.551465213078612e-05,
+      "loss": 19.0789,
+      "step": 5241
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.550238460264744e-05,
+      "loss": 0.2172,
+      "step": 5242
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.549011604197799e-05,
+      "loss": 0.9534,
+      "step": 5243
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.547784644959487e-05,
+      "loss": 21.9427,
+      "step": 5244
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.546557582631534e-05,
+      "loss": 20.0684,
+      "step": 5245
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.545330417295668e-05,
+      "loss": 20.9674,
+      "step": 5246
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.544103149033622e-05,
+      "loss": 22.8339,
+      "step": 5247
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.542875777927139e-05,
+      "loss": 21.8033,
+      "step": 5248
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.541648304057966e-05,
+      "loss": 20.8067,
+      "step": 5249
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.54042072750786e-05,
+      "loss": 20.3322,
+      "step": 5250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.539193048358582e-05,
+      "loss": 20.9543,
+      "step": 5251
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.537965266691902e-05,
+      "loss": 0.1952,
+      "step": 5252
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.536737382589598e-05,
+      "loss": 20.66,
+      "step": 5253
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.535509396133448e-05,
+      "loss": 0.9021,
+      "step": 5254
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.534281307405245e-05,
+      "loss": 20.6663,
+      "step": 5255
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.533053116486785e-05,
+      "loss": 21.7225,
+      "step": 5256
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.531824823459869e-05,
+      "loss": 20.2203,
+      "step": 5257
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.530596428406312e-05,
+      "loss": 0.1813,
+      "step": 5258
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.529367931407926e-05,
+      "loss": 19.446,
+      "step": 5259
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.528139332546537e-05,
+      "loss": 0.9592,
+      "step": 5260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.526910631903973e-05,
+      "loss": 0.1745,
+      "step": 5261
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.525681829562074e-05,
+      "loss": 20.7284,
+      "step": 5262
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.524452925602684e-05,
+      "loss": 0.1895,
+      "step": 5263
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.523223920107651e-05,
+      "loss": 0.2339,
+      "step": 5264
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.521994813158834e-05,
+      "loss": 0.1941,
+      "step": 5265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.520765604838097e-05,
+      "loss": 0.9626,
+      "step": 5266
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.519536295227312e-05,
+      "loss": 23.5531,
+      "step": 5267
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.518306884408355e-05,
+      "loss": 0.9949,
+      "step": 5268
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.517077372463112e-05,
+      "loss": 0.9304,
+      "step": 5269
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.515847759473474e-05,
+      "loss": 0.1685,
+      "step": 5270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.514618045521338e-05,
+      "loss": 0.9256,
+      "step": 5271
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.513388230688608e-05,
+      "loss": 19.6925,
+      "step": 5272
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5121583150572e-05,
+      "loss": 22.2333,
+      "step": 5273
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.510928298709025e-05,
+      "loss": 0.2114,
+      "step": 5274
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.509698181726014e-05,
+      "loss": 0.2007,
+      "step": 5275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.508467964190095e-05,
+      "loss": 20.8452,
+      "step": 5276
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.507237646183205e-05,
+      "loss": 19.2258,
+      "step": 5277
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.506007227787294e-05,
+      "loss": 0.9762,
+      "step": 5278
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.504776709084308e-05,
+      "loss": 0.9,
+      "step": 5279
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.50354609015621e-05,
+      "loss": 0.2387,
+      "step": 5280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.502315371084962e-05,
+      "loss": 0.1965,
+      "step": 5281
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.501084551952537e-05,
+      "loss": 20.6824,
+      "step": 5282
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.499853632840912e-05,
+      "loss": 0.187,
+      "step": 5283
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.498622613832073e-05,
+      "loss": 19.5845,
+      "step": 5284
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.497391495008011e-05,
+      "loss": 0.2173,
+      "step": 5285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.496160276450724e-05,
+      "loss": 0.2065,
+      "step": 5286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.494928958242217e-05,
+      "loss": 21.8385,
+      "step": 5287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.493697540464502e-05,
+      "loss": 0.9399,
+      "step": 5288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.492466023199598e-05,
+      "loss": 0.1973,
+      "step": 5289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.491234406529528e-05,
+      "loss": 19.264,
+      "step": 5290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.490002690536324e-05,
+      "loss": 0.2429,
+      "step": 5291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.488770875302024e-05,
+      "loss": 0.2265,
+      "step": 5292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.487538960908672e-05,
+      "loss": 21.1498,
+      "step": 5293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.486306947438322e-05,
+      "loss": 0.2265,
+      "step": 5294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.485074834973028e-05,
+      "loss": 19.0212,
+      "step": 5295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.483842623594856e-05,
+      "loss": 22.0738,
+      "step": 5296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.482610313385877e-05,
+      "loss": 20.5428,
+      "step": 5297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.481377904428171e-05,
+      "loss": 0.1901,
+      "step": 5298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.480145396803817e-05,
+      "loss": 0.2518,
+      "step": 5299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.47891279059491e-05,
+      "loss": 21.4773,
+      "step": 5300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.477680085883546e-05,
+      "loss": 22.3895,
+      "step": 5301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.47644728275183e-05,
+      "loss": 0.941,
+      "step": 5302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.475214381281869e-05,
+      "loss": 21.4284,
+      "step": 5303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.473981381555782e-05,
+      "loss": 21.1098,
+      "step": 5304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.472748283655694e-05,
+      "loss": 20.151,
+      "step": 5305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.471515087663733e-05,
+      "loss": 18.6758,
+      "step": 5306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.470281793662036e-05,
+      "loss": 20.3401,
+      "step": 5307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.469048401732747e-05,
+      "loss": 22.6764,
+      "step": 5308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.467814911958012e-05,
+      "loss": 0.9264,
+      "step": 5309
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.466581324419995e-05,
+      "loss": 22.9517,
+      "step": 5310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.465347639200852e-05,
+      "loss": 0.1879,
+      "step": 5311
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.464113856382752e-05,
+      "loss": 0.1729,
+      "step": 5312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.462879976047874e-05,
+      "loss": 20.5189,
+      "step": 5313
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.461645998278399e-05,
+      "loss": 19.4813,
+      "step": 5314
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.460411923156515e-05,
+      "loss": 0.9676,
+      "step": 5315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.459177750764418e-05,
+      "loss": 19.158,
+      "step": 5316
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.457943481184309e-05,
+      "loss": 0.234,
+      "step": 5317
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.456709114498396e-05,
+      "loss": 0.2222,
+      "step": 5318
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.455474650788894e-05,
+      "loss": 0.907,
+      "step": 5319
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.454240090138024e-05,
+      "loss": 19.0271,
+      "step": 5320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.453005432628014e-05,
+      "loss": 20.9135,
+      "step": 5321
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.451770678341096e-05,
+      "loss": 0.9634,
+      "step": 5322
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.450535827359511e-05,
+      "loss": 21.3644,
+      "step": 5323
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.449300879765508e-05,
+      "loss": 20.0773,
+      "step": 5324
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.448065835641336e-05,
+      "loss": 19.9683,
+      "step": 5325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.446830695069259e-05,
+      "loss": 21.6306,
+      "step": 5326
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.445595458131543e-05,
+      "loss": 19.7686,
+      "step": 5327
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.444360124910457e-05,
+      "loss": 21.5097,
+      "step": 5328
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.44312469548828e-05,
+      "loss": 21.4084,
+      "step": 5329
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.441889169947304e-05,
+      "loss": 0.2408,
+      "step": 5330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.440653548369812e-05,
+      "loss": 0.9763,
+      "step": 5331
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.439417830838107e-05,
+      "loss": 0.1925,
+      "step": 5332
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.438182017434494e-05,
+      "loss": 21.8956,
+      "step": 5333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.43694610824128e-05,
+      "loss": 0.9324,
+      "step": 5334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.435710103340786e-05,
+      "loss": 20.3859,
+      "step": 5335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.434474002815335e-05,
+      "loss": 0.928,
+      "step": 5336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.433237806747257e-05,
+      "loss": 0.2137,
+      "step": 5337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.432001515218886e-05,
+      "loss": 0.9717,
+      "step": 5338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.430765128312568e-05,
+      "loss": 0.1886,
+      "step": 5339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.429528646110648e-05,
+      "loss": 20.6324,
+      "step": 5340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.428292068695488e-05,
+      "loss": 20.2929,
+      "step": 5341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.427055396149443e-05,
+      "loss": 0.9225,
+      "step": 5342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.425818628554884e-05,
+      "loss": 19.638,
+      "step": 5343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.424581765994188e-05,
+      "loss": 19.8779,
+      "step": 5344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.423344808549732e-05,
+      "loss": 20.3421,
+      "step": 5345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.422107756303903e-05,
+      "loss": 19.6045,
+      "step": 5346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4208706093391e-05,
+      "loss": 0.9559,
+      "step": 5347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.419633367737714e-05,
+      "loss": 21.806,
+      "step": 5348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.418396031582158e-05,
+      "loss": 20.6107,
+      "step": 5349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.417158600954843e-05,
+      "loss": 22.2372,
+      "step": 5350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.415921075938184e-05,
+      "loss": 21.6798,
+      "step": 5351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.41468345661461e-05,
+      "loss": 0.9699,
+      "step": 5352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.41344574306655e-05,
+      "loss": 20.9738,
+      "step": 5353
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.412207935376442e-05,
+      "loss": 21.0854,
+      "step": 5354
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.410970033626731e-05,
+      "loss": 18.1439,
+      "step": 5355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.409732037899865e-05,
+      "loss": 21.6192,
+      "step": 5356
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.408493948278302e-05,
+      "loss": 0.9686,
+      "step": 5357
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.407255764844503e-05,
+      "loss": 0.2272,
+      "step": 5358
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.406017487680937e-05,
+      "loss": 0.1868,
+      "step": 5359
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.404779116870082e-05,
+      "loss": 0.9675,
+      "step": 5360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.403540652494413e-05,
+      "loss": 20.7675,
+      "step": 5361
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.402302094636424e-05,
+      "loss": 0.9035,
+      "step": 5362
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.401063443378603e-05,
+      "loss": 0.9723,
+      "step": 5363
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.399824698803456e-05,
+      "loss": 20.2462,
+      "step": 5364
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.398585860993484e-05,
+      "loss": 20.2584,
+      "step": 5365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.397346930031202e-05,
+      "loss": 0.1774,
+      "step": 5366
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.396107905999127e-05,
+      "loss": 0.1766,
+      "step": 5367
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.394868788979785e-05,
+      "loss": 18.2462,
+      "step": 5368
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.393629579055709e-05,
+      "loss": 0.1957,
+      "step": 5369
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.39239027630943e-05,
+      "loss": 23.2528,
+      "step": 5370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.391150880823496e-05,
+      "loss": 21.3059,
+      "step": 5371
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.389911392680456e-05,
+      "loss": 19.9807,
+      "step": 5372
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.388671811962865e-05,
+      "loss": 22.7012,
+      "step": 5373
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.387432138753286e-05,
+      "loss": 0.1993,
+      "step": 5374
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.386192373134283e-05,
+      "loss": 0.2122,
+      "step": 5375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.384952515188436e-05,
+      "loss": 21.2152,
+      "step": 5376
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.383712564998322e-05,
+      "loss": 0.9623,
+      "step": 5377
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.382472522646527e-05,
+      "loss": 0.8988,
+      "step": 5378
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.381232388215645e-05,
+      "loss": 0.2471,
+      "step": 5379
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.379992161788273e-05,
+      "loss": 21.8557,
+      "step": 5380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.378751843447018e-05,
+      "loss": 18.7494,
+      "step": 5381
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.37751143327449e-05,
+      "loss": 20.153,
+      "step": 5382
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.376270931353307e-05,
+      "loss": 19.0721,
+      "step": 5383
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.375030337766089e-05,
+      "loss": 0.2042,
+      "step": 5384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.373789652595469e-05,
+      "loss": 0.9313,
+      "step": 5385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.372548875924081e-05,
+      "loss": 22.6881,
+      "step": 5386
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.371308007834564e-05,
+      "loss": 0.2032,
+      "step": 5387
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.37006704840957e-05,
+      "loss": 0.9885,
+      "step": 5388
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.368825997731751e-05,
+      "loss": 20.9081,
+      "step": 5389
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.367584855883765e-05,
+      "loss": 19.8308,
+      "step": 5390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.36634362294828e-05,
+      "loss": 0.2051,
+      "step": 5391
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.365102299007965e-05,
+      "loss": 19.5097,
+      "step": 5392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.363860884145502e-05,
+      "loss": 20.4744,
+      "step": 5393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.362619378443575e-05,
+      "loss": 20.2439,
+      "step": 5394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.361377781984868e-05,
+      "loss": 19.682,
+      "step": 5395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.360136094852083e-05,
+      "loss": 22.2106,
+      "step": 5396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.35889431712792e-05,
+      "loss": 0.2029,
+      "step": 5397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.357652448895086e-05,
+      "loss": 21.3319,
+      "step": 5398
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3564104902363e-05,
+      "loss": 0.211,
+      "step": 5399
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.355168441234275e-05,
+      "loss": 0.9231,
+      "step": 5400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.353926301971742e-05,
+      "loss": 0.9517,
+      "step": 5401
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.352684072531435e-05,
+      "loss": 0.9597,
+      "step": 5402
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.351441752996084e-05,
+      "loss": 22.2399,
+      "step": 5403
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.350199343448444e-05,
+      "loss": 22.4502,
+      "step": 5404
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.348956843971257e-05,
+      "loss": 19.1944,
+      "step": 5405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.347714254647284e-05,
+      "loss": 0.9555,
+      "step": 5406
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.346471575559284e-05,
+      "loss": 21.9839,
+      "step": 5407
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.345228806790026e-05,
+      "loss": 20.8221,
+      "step": 5408
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.343985948422287e-05,
+      "loss": 0.9235,
+      "step": 5409
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.342743000538845e-05,
+      "loss": 0.9426,
+      "step": 5410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.341499963222484e-05,
+      "loss": 0.9765,
+      "step": 5411
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.340256836556e-05,
+      "loss": 0.9383,
+      "step": 5412
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.339013620622187e-05,
+      "loss": 0.2191,
+      "step": 5413
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.337770315503851e-05,
+      "loss": 0.2215,
+      "step": 5414
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.336526921283804e-05,
+      "loss": 22.0973,
+      "step": 5415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.335283438044859e-05,
+      "loss": 21.4245,
+      "step": 5416
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.334039865869838e-05,
+      "loss": 0.2221,
+      "step": 5417
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.332796204841569e-05,
+      "loss": 0.1974,
+      "step": 5418
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.331552455042886e-05,
+      "loss": 0.9269,
+      "step": 5419
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.330308616556629e-05,
+      "loss": 0.192,
+      "step": 5420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.329064689465643e-05,
+      "loss": 0.2333,
+      "step": 5421
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.327820673852777e-05,
+      "loss": 0.9558,
+      "step": 5422
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.326576569800892e-05,
+      "loss": 22.585,
+      "step": 5423
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.32533237739285e-05,
+      "loss": 0.9183,
+      "step": 5424
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.324088096711517e-05,
+      "loss": 0.9518,
+      "step": 5425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.322843727839772e-05,
+      "loss": 0.9543,
+      "step": 5426
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.321599270860492e-05,
+      "loss": 0.1856,
+      "step": 5427
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.320354725856568e-05,
+      "loss": 20.788,
+      "step": 5428
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.319110092910889e-05,
+      "loss": 0.2018,
+      "step": 5429
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.317865372106354e-05,
+      "loss": 22.0615,
+      "step": 5430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.31662056352587e-05,
+      "loss": 21.13,
+      "step": 5431
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.315375667252341e-05,
+      "loss": 0.9577,
+      "step": 5432
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.314130683368686e-05,
+      "loss": 24.6303,
+      "step": 5433
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.31288561195783e-05,
+      "loss": 0.965,
+      "step": 5434
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.311640453102699e-05,
+      "loss": 19.0592,
+      "step": 5435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.310395206886222e-05,
+      "loss": 0.1926,
+      "step": 5436
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.309149873391342e-05,
+      "loss": 0.9581,
+      "step": 5437
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.307904452701003e-05,
+      "loss": 0.9612,
+      "step": 5438
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.306658944898156e-05,
+      "loss": 0.9343,
+      "step": 5439
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.305413350065759e-05,
+      "loss": 0.9643,
+      "step": 5440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.304167668286771e-05,
+      "loss": 20.7244,
+      "step": 5441
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.302921899644164e-05,
+      "loss": 0.2027,
+      "step": 5442
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.30167604422091e-05,
+      "loss": 0.9607,
+      "step": 5443
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.300430102099988e-05,
+      "loss": 18.3537,
+      "step": 5444
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.299184073364384e-05,
+      "loss": 0.2049,
+      "step": 5445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.297937958097094e-05,
+      "loss": 20.2696,
+      "step": 5446
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.296691756381104e-05,
+      "loss": 22.3791,
+      "step": 5447
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.295445468299428e-05,
+      "loss": 21.3618,
+      "step": 5448
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.29419909393507e-05,
+      "loss": 0.2013,
+      "step": 5449
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.292952633371043e-05,
+      "loss": 18.8897,
+      "step": 5450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.291706086690372e-05,
+      "loss": 19.8444,
+      "step": 5451
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.290459453976075e-05,
+      "loss": 0.986,
+      "step": 5452
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.28921273531119e-05,
+      "loss": 19.632,
+      "step": 5453
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.287965930778753e-05,
+      "loss": 0.1848,
+      "step": 5454
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.286719040461804e-05,
+      "loss": 19.6349,
+      "step": 5455
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.285472064443396e-05,
+      "loss": 22.1245,
+      "step": 5456
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.28422500280658e-05,
+      "loss": 20.7375,
+      "step": 5457
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.282977855634418e-05,
+      "loss": 0.9584,
+      "step": 5458
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.281730623009974e-05,
+      "loss": 0.9973,
+      "step": 5459
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.28048330501632e-05,
+      "loss": 19.6805,
+      "step": 5460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.279235901736536e-05,
+      "loss": 0.1924,
+      "step": 5461
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.277988413253703e-05,
+      "loss": 21.8447,
+      "step": 5462
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.276740839650908e-05,
+      "loss": 0.9241,
+      "step": 5463
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.275493181011244e-05,
+      "loss": 0.1979,
+      "step": 5464
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.274245437417817e-05,
+      "loss": 0.9383,
+      "step": 5465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.272997608953726e-05,
+      "loss": 21.9941,
+      "step": 5466
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.271749695702087e-05,
+      "loss": 20.2475,
+      "step": 5467
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.270501697746013e-05,
+      "loss": 22.3415,
+      "step": 5468
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.269253615168628e-05,
+      "loss": 20.506,
+      "step": 5469
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.26800544805306e-05,
+      "loss": 21.5234,
+      "step": 5470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.266757196482445e-05,
+      "loss": 22.2549,
+      "step": 5471
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.26550886053992e-05,
+      "loss": 19.9993,
+      "step": 5472
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.26426044030863e-05,
+      "loss": 20.5065,
+      "step": 5473
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.263011935871723e-05,
+      "loss": 0.901,
+      "step": 5474
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.261763347312361e-05,
+      "loss": 19.5384,
+      "step": 5475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.260514674713704e-05,
+      "loss": 0.1699,
+      "step": 5476
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.259265918158914e-05,
+      "loss": 22.1001,
+      "step": 5477
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.258017077731173e-05,
+      "loss": 21.5953,
+      "step": 5478
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.256768153513651e-05,
+      "loss": 0.2384,
+      "step": 5479
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.255519145589537e-05,
+      "loss": 18.9472,
+      "step": 5480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.254270054042021e-05,
+      "loss": 0.1883,
+      "step": 5481
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.253020878954295e-05,
+      "loss": 21.9553,
+      "step": 5482
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.251771620409563e-05,
+      "loss": 21.1688,
+      "step": 5483
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.25052227849103e-05,
+      "loss": 20.7573,
+      "step": 5484
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.249272853281908e-05,
+      "loss": 21.5099,
+      "step": 5485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.248023344865416e-05,
+      "loss": 19.8836,
+      "step": 5486
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.246773753324774e-05,
+      "loss": 0.1994,
+      "step": 5487
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.245524078743213e-05,
+      "loss": 0.9588,
+      "step": 5488
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.244274321203966e-05,
+      "loss": 22.0132,
+      "step": 5489
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.243024480790275e-05,
+      "loss": 0.1849,
+      "step": 5490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.241774557585382e-05,
+      "loss": 0.2188,
+      "step": 5491
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.240524551672538e-05,
+      "loss": 0.2435,
+      "step": 5492
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.239274463135002e-05,
+      "loss": 0.9882,
+      "step": 5493
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.238024292056033e-05,
+      "loss": 23.627,
+      "step": 5494
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2367740385189e-05,
+      "loss": 22.498,
+      "step": 5495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.235523702606873e-05,
+      "loss": 23.5901,
+      "step": 5496
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.234273284403233e-05,
+      "loss": 0.9761,
+      "step": 5497
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.233022783991265e-05,
+      "loss": 0.986,
+      "step": 5498
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.231772201454252e-05,
+      "loss": 0.9539,
+      "step": 5499
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.230521536875494e-05,
+      "loss": 0.1713,
+      "step": 5500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.229270790338289e-05,
+      "loss": 21.5269,
+      "step": 5501
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.228019961925944e-05,
+      "loss": 0.9092,
+      "step": 5502
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.22676905172177e-05,
+      "loss": 20.6817,
+      "step": 5503
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.22551805980908e-05,
+      "loss": 22.3899,
+      "step": 5504
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.224266986271198e-05,
+      "loss": 0.9428,
+      "step": 5505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.223015831191454e-05,
+      "loss": 21.0688,
+      "step": 5506
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.221764594653175e-05,
+      "loss": 20.7666,
+      "step": 5507
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.220513276739706e-05,
+      "loss": 0.1813,
+      "step": 5508
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.219261877534385e-05,
+      "loss": 0.9334,
+      "step": 5509
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.218010397120563e-05,
+      "loss": 1.0101,
+      "step": 5510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.216758835581595e-05,
+      "loss": 20.5105,
+      "step": 5511
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.21550719300084e-05,
+      "loss": 0.2239,
+      "step": 5512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.214255469461663e-05,
+      "loss": 0.2199,
+      "step": 5513
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.213003665047435e-05,
+      "loss": 20.0648,
+      "step": 5514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.211751779841534e-05,
+      "loss": 0.1935,
+      "step": 5515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.210499813927337e-05,
+      "loss": 0.1977,
+      "step": 5516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.209247767388234e-05,
+      "loss": 0.9119,
+      "step": 5517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.207995640307616e-05,
+      "loss": 0.9723,
+      "step": 5518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.20674343276888e-05,
+      "loss": 22.207,
+      "step": 5519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.205491144855432e-05,
+      "loss": 0.9345,
+      "step": 5520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.204238776650676e-05,
+      "loss": 21.0331,
+      "step": 5521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.202986328238027e-05,
+      "loss": 22.4585,
+      "step": 5522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.201733799700906e-05,
+      "loss": 1.0059,
+      "step": 5523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.200481191122733e-05,
+      "loss": 0.9473,
+      "step": 5524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.199228502586943e-05,
+      "loss": 0.9507,
+      "step": 5525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.197975734176964e-05,
+      "loss": 21.4816,
+      "step": 5526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.196722885976241e-05,
+      "loss": 0.9531,
+      "step": 5527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.195469958068221e-05,
+      "loss": 0.9693,
+      "step": 5528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.19421695053635e-05,
+      "loss": 0.9383,
+      "step": 5529
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.192963863464088e-05,
+      "loss": 19.7259,
+      "step": 5530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.191710696934894e-05,
+      "loss": 0.1696,
+      "step": 5531
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.190457451032234e-05,
+      "loss": 20.5925,
+      "step": 5532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.189204125839584e-05,
+      "loss": 22.806,
+      "step": 5533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.187950721440417e-05,
+      "loss": 0.2945,
+      "step": 5534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.186697237918219e-05,
+      "loss": 0.2074,
+      "step": 5535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.185443675356475e-05,
+      "loss": 19.5245,
+      "step": 5536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.184190033838678e-05,
+      "loss": 0.9642,
+      "step": 5537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.18293631344833e-05,
+      "loss": 19.8854,
+      "step": 5538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.181682514268933e-05,
+      "loss": 0.2337,
+      "step": 5539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.180428636383993e-05,
+      "loss": 0.9726,
+      "step": 5540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.179174679877026e-05,
+      "loss": 20.0646,
+      "step": 5541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.177920644831553e-05,
+      "loss": 0.955,
+      "step": 5542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.176666531331096e-05,
+      "loss": 19.991,
+      "step": 5543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.175412339459186e-05,
+      "loss": 20.7633,
+      "step": 5544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.174158069299357e-05,
+      "loss": 0.1966,
+      "step": 5545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.172903720935152e-05,
+      "loss": 20.0214,
+      "step": 5546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.171649294450113e-05,
+      "loss": 0.9265,
+      "step": 5547
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.170394789927791e-05,
+      "loss": 0.9642,
+      "step": 5548
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.169140207451744e-05,
+      "loss": 0.9248,
+      "step": 5549
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.167885547105534e-05,
+      "loss": 0.9406,
+      "step": 5550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.166630808972722e-05,
+      "loss": 20.891,
+      "step": 5551
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.165375993136884e-05,
+      "loss": 19.9665,
+      "step": 5552
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.164121099681597e-05,
+      "loss": 0.2231,
+      "step": 5553
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.162866128690436e-05,
+      "loss": 0.9192,
+      "step": 5554
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.161611080246997e-05,
+      "loss": 19.1708,
+      "step": 5555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.160355954434865e-05,
+      "loss": 20.0527,
+      "step": 5556
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.159100751337642e-05,
+      "loss": 19.4245,
+      "step": 5557
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.157845471038928e-05,
+      "loss": 0.2152,
+      "step": 5558
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.15659011362233e-05,
+      "loss": 20.3952,
+      "step": 5559
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.155334679171462e-05,
+      "loss": 19.285,
+      "step": 5560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.15407916776994e-05,
+      "loss": 0.1913,
+      "step": 5561
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.152823579501388e-05,
+      "loss": 21.0089,
+      "step": 5562
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.151567914449437e-05,
+      "loss": 0.1578,
+      "step": 5563
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.150312172697714e-05,
+      "loss": 18.4081,
+      "step": 5564
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.149056354329862e-05,
+      "loss": 0.9148,
+      "step": 5565
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.147800459429522e-05,
+      "loss": 0.9224,
+      "step": 5566
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.146544488080343e-05,
+      "loss": 19.8958,
+      "step": 5567
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.14528844036598e-05,
+      "loss": 0.2007,
+      "step": 5568
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.14403231637009e-05,
+      "loss": 0.9752,
+      "step": 5569
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.142776116176336e-05,
+      "loss": 20.6575,
+      "step": 5570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.141519839868388e-05,
+      "loss": 19.5607,
+      "step": 5571
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.14026348752992e-05,
+      "loss": 0.1958,
+      "step": 5572
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.139007059244611e-05,
+      "loss": 0.9391,
+      "step": 5573
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.137750555096145e-05,
+      "loss": 0.1991,
+      "step": 5574
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.136493975168209e-05,
+      "loss": 0.9377,
+      "step": 5575
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.135237319544498e-05,
+      "loss": 0.954,
+      "step": 5576
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.133980588308714e-05,
+      "loss": 0.2375,
+      "step": 5577
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.132723781544558e-05,
+      "loss": 21.9099,
+      "step": 5578
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.131466899335738e-05,
+      "loss": 19.6222,
+      "step": 5579
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.130209941765974e-05,
+      "loss": 0.9626,
+      "step": 5580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.128952908918977e-05,
+      "loss": 0.2186,
+      "step": 5581
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.127695800878478e-05,
+      "loss": 0.9559,
+      "step": 5582
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.126438617728203e-05,
+      "loss": 20.7169,
+      "step": 5583
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.125181359551886e-05,
+      "loss": 20.9141,
+      "step": 5584
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.12392402643327e-05,
+      "loss": 18.963,
+      "step": 5585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.122666618456094e-05,
+      "loss": 0.1663,
+      "step": 5586
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.121409135704111e-05,
+      "loss": 21.2635,
+      "step": 5587
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.120151578261074e-05,
+      "loss": 0.9266,
+      "step": 5588
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.11889394621074e-05,
+      "loss": 0.217,
+      "step": 5589
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.117636239636879e-05,
+      "loss": 0.9843,
+      "step": 5590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.116378458623253e-05,
+      "loss": 0.2186,
+      "step": 5591
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.115120603253641e-05,
+      "loss": 0.186,
+      "step": 5592
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.113862673611819e-05,
+      "loss": 0.2352,
+      "step": 5593
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.112604669781572e-05,
+      "loss": 22.698,
+      "step": 5594
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.11134659184669e-05,
+      "loss": 0.204,
+      "step": 5595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.110088439890966e-05,
+      "loss": 0.92,
+      "step": 5596
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1088302139982e-05,
+      "loss": 0.9526,
+      "step": 5597
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.107571914252192e-05,
+      "loss": 0.9679,
+      "step": 5598
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.106313540736755e-05,
+      "loss": 20.6147,
+      "step": 5599
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1050550935357e-05,
+      "loss": 22.9076,
+      "step": 5600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.103796572732846e-05,
+      "loss": 21.0201,
+      "step": 5601
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1025379784120174e-05,
+      "loss": 18.338,
+      "step": 5602
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1012793106570386e-05,
+      "loss": 19.6531,
+      "step": 5603
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1000205695517475e-05,
+      "loss": 22.0414,
+      "step": 5604
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.09876175517998e-05,
+      "loss": 21.9082,
+      "step": 5605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.097502867625577e-05,
+      "loss": 0.9188,
+      "step": 5606
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.096243906972392e-05,
+      "loss": 20.7043,
+      "step": 5607
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0949848733042704e-05,
+      "loss": 19.4178,
+      "step": 5608
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0937257667050764e-05,
+      "loss": 20.1772,
+      "step": 5609
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.092466587258667e-05,
+      "loss": 21.1247,
+      "step": 5610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.091207335048911e-05,
+      "loss": 0.9274,
+      "step": 5611
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0899480101596825e-05,
+      "loss": 18.9593,
+      "step": 5612
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.088688612674856e-05,
+      "loss": 0.9709,
+      "step": 5613
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.087429142678316e-05,
+      "loss": 0.8941,
+      "step": 5614
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.086169600253945e-05,
+      "loss": 0.9827,
+      "step": 5615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.084909985485637e-05,
+      "loss": 0.2248,
+      "step": 5616
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.083650298457289e-05,
+      "loss": 0.9328,
+      "step": 5617
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.082390539252798e-05,
+      "loss": 0.9845,
+      "step": 5618
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0811307079560745e-05,
+      "loss": 21.6562,
+      "step": 5619
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.079870804651027e-05,
+      "loss": 20.0171,
+      "step": 5620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.078610829421571e-05,
+      "loss": 20.8916,
+      "step": 5621
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.077350782351626e-05,
+      "loss": 20.3321,
+      "step": 5622
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.076090663525118e-05,
+      "loss": 1.0036,
+      "step": 5623
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.074830473025977e-05,
+      "loss": 19.4914,
+      "step": 5624
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.073570210938136e-05,
+      "loss": 21.466,
+      "step": 5625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0723098773455365e-05,
+      "loss": 20.7788,
+      "step": 5626
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.071049472332122e-05,
+      "loss": 21.0942,
+      "step": 5627
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.069788995981839e-05,
+      "loss": 23.1441,
+      "step": 5628
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.068528448378644e-05,
+      "loss": 0.9228,
+      "step": 5629
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0672678296064936e-05,
+      "loss": 21.3393,
+      "step": 5630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0660071397493514e-05,
+      "loss": 0.9586,
+      "step": 5631
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.064746378891186e-05,
+      "loss": 21.4072,
+      "step": 5632
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0634855471159666e-05,
+      "loss": 0.9395,
+      "step": 5633
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.062224644507677e-05,
+      "loss": 18.3032,
+      "step": 5634
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.060963671150293e-05,
+      "loss": 19.741,
+      "step": 5635
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.059702627127801e-05,
+      "loss": 21.0364,
+      "step": 5636
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0584415125241976e-05,
+      "loss": 0.2223,
+      "step": 5637
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.057180327423476e-05,
+      "loss": 0.9185,
+      "step": 5638
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.055919071909637e-05,
+      "loss": 0.9049,
+      "step": 5639
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0546577460666855e-05,
+      "loss": 0.2003,
+      "step": 5640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0533963499786314e-05,
+      "loss": 19.1376,
+      "step": 5641
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0521348837294925e-05,
+      "loss": 0.9473,
+      "step": 5642
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.050873347403285e-05,
+      "loss": 0.9154,
+      "step": 5643
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0496117410840346e-05,
+      "loss": 0.1635,
+      "step": 5644
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.04835006485577e-05,
+      "loss": 23.4706,
+      "step": 5645
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.047088318802524e-05,
+      "loss": 0.9619,
+      "step": 5646
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.045826503008336e-05,
+      "loss": 19.9801,
+      "step": 5647
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.044564617557247e-05,
+      "loss": 21.9935,
+      "step": 5648
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.043302662533308e-05,
+      "loss": 0.992,
+      "step": 5649
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.042040638020565e-05,
+      "loss": 19.7061,
+      "step": 5650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0407785441030804e-05,
+      "loss": 0.1944,
+      "step": 5651
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.039516380864913e-05,
+      "loss": 21.6264,
+      "step": 5652
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.038254148390129e-05,
+      "loss": 0.9521,
+      "step": 5653
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.036991846762799e-05,
+      "loss": 19.0793,
+      "step": 5654
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.035729476066998e-05,
+      "loss": 0.9329,
+      "step": 5655
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0344670363868064e-05,
+      "loss": 18.8065,
+      "step": 5656
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.033204527806309e-05,
+      "loss": 0.1721,
+      "step": 5657
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.031941950409592e-05,
+      "loss": 0.9103,
+      "step": 5658
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.030679304280752e-05,
+      "loss": 23.2942,
+      "step": 5659
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.029416589503886e-05,
+      "loss": 0.965,
+      "step": 5660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.028153806163095e-05,
+      "loss": 0.9586,
+      "step": 5661
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0268909543424876e-05,
+      "loss": 19.7615,
+      "step": 5662
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.025628034126176e-05,
+      "loss": 21.1774,
+      "step": 5663
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.024365045598278e-05,
+      "loss": 0.1888,
+      "step": 5664
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.023101988842912e-05,
+      "loss": 19.8997,
+      "step": 5665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.021838863944204e-05,
+      "loss": 0.9718,
+      "step": 5666
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.020575670986284e-05,
+      "loss": 19.8255,
+      "step": 5667
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.019312410053286e-05,
+      "loss": 21.1175,
+      "step": 5668
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0180490812293514e-05,
+      "loss": 19.4733,
+      "step": 5669
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.01678568459862e-05,
+      "loss": 20.7177,
+      "step": 5670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.015522220245243e-05,
+      "loss": 19.8448,
+      "step": 5671
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.014258688253372e-05,
+      "loss": 0.953,
+      "step": 5672
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.012995088707165e-05,
+      "loss": 20.3487,
+      "step": 5673
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.011731421690782e-05,
+      "loss": 0.9414,
+      "step": 5674
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.010467687288389e-05,
+      "loss": 19.8285,
+      "step": 5675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.009203885584158e-05,
+      "loss": 0.9777,
+      "step": 5676
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0079400166622636e-05,
+      "loss": 21.1968,
+      "step": 5677
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0066760806068844e-05,
+      "loss": 19.2836,
+      "step": 5678
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.005412077502207e-05,
+      "loss": 19.9365,
+      "step": 5679
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.004148007432415e-05,
+      "loss": 0.2291,
+      "step": 5680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.002883870481708e-05,
+      "loss": 0.207,
+      "step": 5681
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.001619666734276e-05,
+      "loss": 0.9546,
+      "step": 5682
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0003553962743266e-05,
+      "loss": 0.1679,
+      "step": 5683
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.999091059186065e-05,
+      "loss": 22.2769,
+      "step": 5684
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.997826655553699e-05,
+      "loss": 20.8695,
+      "step": 5685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.996562185461447e-05,
+      "loss": 22.6875,
+      "step": 5686
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.995297648993526e-05,
+      "loss": 0.1937,
+      "step": 5687
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.994033046234162e-05,
+      "loss": 0.9742,
+      "step": 5688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9927683772675834e-05,
+      "loss": 20.4036,
+      "step": 5689
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.991503642178022e-05,
+      "loss": 0.9819,
+      "step": 5690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9902388410497155e-05,
+      "loss": 19.9167,
+      "step": 5691
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.988973973966905e-05,
+      "loss": 21.9186,
+      "step": 5692
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.987709041013837e-05,
+      "loss": 0.2143,
+      "step": 5693
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.986444042274762e-05,
+      "loss": 21.5887,
+      "step": 5694
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.985178977833934e-05,
+      "loss": 0.2238,
+      "step": 5695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.983913847775613e-05,
+      "loss": 0.1929,
+      "step": 5696
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.982648652184063e-05,
+      "loss": 19.5143,
+      "step": 5697
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.98138339114355e-05,
+      "loss": 20.6063,
+      "step": 5698
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.980118064738348e-05,
+      "loss": 0.1795,
+      "step": 5699
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9788526730527336e-05,
+      "loss": 20.7327,
+      "step": 5700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.977587216170988e-05,
+      "loss": 0.9168,
+      "step": 5701
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9763216941773936e-05,
+      "loss": 0.2115,
+      "step": 5702
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9750561071562436e-05,
+      "loss": 0.9052,
+      "step": 5703
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.97379045519183e-05,
+      "loss": 21.509,
+      "step": 5704
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.972524738368452e-05,
+      "loss": 0.216,
+      "step": 5705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9712589567704126e-05,
+      "loss": 20.2389,
+      "step": 5706
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9699931104820163e-05,
+      "loss": 0.9841,
+      "step": 5707
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.968727199587577e-05,
+      "loss": 0.9447,
+      "step": 5708
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.96746122417141e-05,
+      "loss": 21.2595,
+      "step": 5709
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.966195184317832e-05,
+      "loss": 20.3059,
+      "step": 5710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9649290801111724e-05,
+      "loss": 0.9444,
+      "step": 5711
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9636629116357557e-05,
+      "loss": 21.0148,
+      "step": 5712
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.962396678975916e-05,
+      "loss": 20.3879,
+      "step": 5713
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.961130382215989e-05,
+      "loss": 0.9573,
+      "step": 5714
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.959864021440315e-05,
+      "loss": 20.2823,
+      "step": 5715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.958597596733246e-05,
+      "loss": 0.2354,
+      "step": 5716
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.957331108179125e-05,
+      "loss": 19.6907,
+      "step": 5717
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.956064555862308e-05,
+      "loss": 0.1881,
+      "step": 5718
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9547979398671536e-05,
+      "loss": 22.4486,
+      "step": 5719
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9535312602780235e-05,
+      "loss": 0.9262,
+      "step": 5720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.952264517179286e-05,
+      "loss": 18.7884,
+      "step": 5721
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.950997710655311e-05,
+      "loss": 18.5741,
+      "step": 5722
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.949730840790473e-05,
+      "loss": 0.9264,
+      "step": 5723
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.948463907669153e-05,
+      "loss": 0.9268,
+      "step": 5724
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.947196911375733e-05,
+      "loss": 20.7708,
+      "step": 5725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.945929851994603e-05,
+      "loss": 19.5116,
+      "step": 5726
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.944662729610153e-05,
+      "loss": 0.9098,
+      "step": 5727
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.94339554430678e-05,
+      "loss": 18.1002,
+      "step": 5728
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.942128296168885e-05,
+      "loss": 20.4535,
+      "step": 5729
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.940860985280872e-05,
+      "loss": 0.9183,
+      "step": 5730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.93959361172715e-05,
+      "loss": 19.5893,
+      "step": 5731
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9383261755921304e-05,
+      "loss": 0.2391,
+      "step": 5732
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.937058676960233e-05,
+      "loss": 0.9783,
+      "step": 5733
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.935791115915877e-05,
+      "loss": 21.0622,
+      "step": 5734
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.934523492543489e-05,
+      "loss": 21.9437,
+      "step": 5735
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.933255806927499e-05,
+      "loss": 20.1709,
+      "step": 5736
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.931988059152338e-05,
+      "loss": 20.1626,
+      "step": 5737
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9307202493024484e-05,
+      "loss": 0.2218,
+      "step": 5738
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.929452377462268e-05,
+      "loss": 19.8278,
+      "step": 5739
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.928184443716244e-05,
+      "loss": 20.9645,
+      "step": 5740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9269164481488296e-05,
+      "loss": 0.9552,
+      "step": 5741
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.925648390844476e-05,
+      "loss": 22.8409,
+      "step": 5742
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.924380271887642e-05,
+      "loss": 20.574,
+      "step": 5743
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9231120913627915e-05,
+      "loss": 20.0827,
+      "step": 5744
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.921843849354389e-05,
+      "loss": 0.9387,
+      "step": 5745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.920575545946909e-05,
+      "loss": 0.9132,
+      "step": 5746
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9193071812248226e-05,
+      "loss": 0.9582,
+      "step": 5747
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9180387552726116e-05,
+      "loss": 20.6085,
+      "step": 5748
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.916770268174757e-05,
+      "loss": 0.9617,
+      "step": 5749
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.915501720015747e-05,
+      "loss": 20.772,
+      "step": 5750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.914233110880073e-05,
+      "loss": 0.9491,
+      "step": 5751
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.91296444085223e-05,
+      "loss": 0.9913,
+      "step": 5752
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.911695710016718e-05,
+      "loss": 19.7496,
+      "step": 5753
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.910426918458036e-05,
+      "loss": 18.2461,
+      "step": 5754
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.909158066260698e-05,
+      "loss": 0.2222,
+      "step": 5755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9078891535092115e-05,
+      "loss": 18.292,
+      "step": 5756
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9066201802880924e-05,
+      "loss": 0.2298,
+      "step": 5757
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.905351146681863e-05,
+      "loss": 0.8967,
+      "step": 5758
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.904082052775042e-05,
+      "loss": 0.8943,
+      "step": 5759
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.90281289865216e-05,
+      "loss": 0.9054,
+      "step": 5760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.901543684397749e-05,
+      "loss": 0.9421,
+      "step": 5761
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9002744100963424e-05,
+      "loss": 20.1739,
+      "step": 5762
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.899005075832482e-05,
+      "loss": 1.0148,
+      "step": 5763
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.897735681690709e-05,
+      "loss": 21.2541,
+      "step": 5764
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.896466227755574e-05,
+      "loss": 0.9579,
+      "step": 5765
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.895196714111625e-05,
+      "loss": 20.5754,
+      "step": 5766
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8939271408434185e-05,
+      "loss": 20.1896,
+      "step": 5767
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8926575080355174e-05,
+      "loss": 0.9741,
+      "step": 5768
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.891387815772481e-05,
+      "loss": 0.2381,
+      "step": 5769
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.890118064138879e-05,
+      "loss": 0.9459,
+      "step": 5770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.88884825321928e-05,
+      "loss": 0.1894,
+      "step": 5771
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.887578383098263e-05,
+      "loss": 22.039,
+      "step": 5772
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.886308453860405e-05,
+      "loss": 21.0584,
+      "step": 5773
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.885038465590289e-05,
+      "loss": 18.6884,
+      "step": 5774
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.883768418372504e-05,
+      "loss": 0.9268,
+      "step": 5775
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.882498312291639e-05,
+      "loss": 0.9457,
+      "step": 5776
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.881228147432289e-05,
+      "loss": 19.4002,
+      "step": 5777
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.879957923879055e-05,
+      "loss": 21.1142,
+      "step": 5778
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.878687641716538e-05,
+      "loss": 0.9509,
+      "step": 5779
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8774173010293446e-05,
+      "loss": 0.2521,
+      "step": 5780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.876146901902086e-05,
+      "loss": 0.9663,
+      "step": 5781
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.874876444419377e-05,
+      "loss": 21.1389,
+      "step": 5782
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8736059286658375e-05,
+      "loss": 0.962,
+      "step": 5783
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.872335354726084e-05,
+      "loss": 21.4325,
+      "step": 5784
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8710647226847494e-05,
+      "loss": 20.4166,
+      "step": 5785
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.869794032626459e-05,
+      "loss": 20.2409,
+      "step": 5786
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.868523284635847e-05,
+      "loss": 0.1985,
+      "step": 5787
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.867252478797556e-05,
+      "loss": 0.9305,
+      "step": 5788
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.865981615196221e-05,
+      "loss": 22.3439,
+      "step": 5789
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.864710693916493e-05,
+      "loss": 0.99,
+      "step": 5790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.863439715043018e-05,
+      "loss": 0.9323,
+      "step": 5791
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.862168678660448e-05,
+      "loss": 0.9381,
+      "step": 5792
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.860897584853443e-05,
+      "loss": 20.1723,
+      "step": 5793
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.859626433706663e-05,
+      "loss": 0.9028,
+      "step": 5794
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.858355225304771e-05,
+      "loss": 0.9683,
+      "step": 5795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.857083959732438e-05,
+      "loss": 0.918,
+      "step": 5796
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.855812637074333e-05,
+      "loss": 0.9144,
+      "step": 5797
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.854541257415134e-05,
+      "loss": 0.9569,
+      "step": 5798
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.853269820839521e-05,
+      "loss": 20.6973,
+      "step": 5799
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8519983274321766e-05,
+      "loss": 17.2786,
+      "step": 5800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.850726777277788e-05,
+      "loss": 0.1903,
+      "step": 5801
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.849455170461047e-05,
+      "loss": 0.9553,
+      "step": 5802
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.848183507066649e-05,
+      "loss": 0.9407,
+      "step": 5803
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8469117871792925e-05,
+      "loss": 0.9226,
+      "step": 5804
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.845640010883679e-05,
+      "loss": 21.4529,
+      "step": 5805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.844368178264514e-05,
+      "loss": 1.0042,
+      "step": 5806
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.84309628940651e-05,
+      "loss": 21.2977,
+      "step": 5807
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.84182434439438e-05,
+      "loss": 19.6427,
+      "step": 5808
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.84055234331284e-05,
+      "loss": 20.5226,
+      "step": 5809
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8392802862466125e-05,
+      "loss": 0.9547,
+      "step": 5810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.838008173280421e-05,
+      "loss": 0.2001,
+      "step": 5811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8367360044989935e-05,
+      "loss": 21.2573,
+      "step": 5812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.835463779987067e-05,
+      "loss": 22.5292,
+      "step": 5813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8341914998293736e-05,
+      "loss": 20.3772,
+      "step": 5814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.832919164110654e-05,
+      "loss": 0.9519,
+      "step": 5815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.831646772915651e-05,
+      "loss": 21.3593,
+      "step": 5816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8303743263291125e-05,
+      "loss": 21.7953,
+      "step": 5817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8291018244357896e-05,
+      "loss": 20.7992,
+      "step": 5818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.827829267320436e-05,
+      "loss": 18.9378,
+      "step": 5819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8265566550678095e-05,
+      "loss": 0.9344,
+      "step": 5820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.825283987762674e-05,
+      "loss": 20.9278,
+      "step": 5821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.824011265489794e-05,
+      "loss": 21.8404,
+      "step": 5822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8227384883339375e-05,
+      "loss": 21.3104,
+      "step": 5823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.82146565637988e-05,
+      "loss": 0.9505,
+      "step": 5824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.820192769712395e-05,
+      "loss": 18.3923,
+      "step": 5825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.818919828416265e-05,
+      "loss": 19.5386,
+      "step": 5826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8176468325762726e-05,
+      "loss": 0.9258,
+      "step": 5827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.816373782277205e-05,
+      "loss": 0.2163,
+      "step": 5828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8151006776038544e-05,
+      "loss": 0.2047,
+      "step": 5829
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.813827518641015e-05,
+      "loss": 0.8979,
+      "step": 5830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8125543054734844e-05,
+      "loss": 23.2496,
+      "step": 5831
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.811281038186065e-05,
+      "loss": 0.8922,
+      "step": 5832
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8100077168635634e-05,
+      "loss": 20.6612,
+      "step": 5833
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.808734341590787e-05,
+      "loss": 0.9613,
+      "step": 5834
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.807460912452551e-05,
+      "loss": 0.9853,
+      "step": 5835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8061874295336675e-05,
+      "loss": 1.0082,
+      "step": 5836
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8049138929189604e-05,
+      "loss": 0.9317,
+      "step": 5837
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8036403026932504e-05,
+      "loss": 20.1075,
+      "step": 5838
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.802366658941365e-05,
+      "loss": 0.9175,
+      "step": 5839
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.801092961748138e-05,
+      "loss": 0.9901,
+      "step": 5840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7998192111984e-05,
+      "loss": 22.4765,
+      "step": 5841
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7985454073769885e-05,
+      "loss": 0.9232,
+      "step": 5842
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.797271550368747e-05,
+      "loss": 0.2034,
+      "step": 5843
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.795997640258518e-05,
+      "loss": 20.3309,
+      "step": 5844
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.794723677131153e-05,
+      "loss": 19.5796,
+      "step": 5845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7934496610715005e-05,
+      "loss": 0.9325,
+      "step": 5846
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7921755921644185e-05,
+      "loss": 0.9973,
+      "step": 5847
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.790901470494763e-05,
+      "loss": 0.9288,
+      "step": 5848
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7896272961474e-05,
+      "loss": 19.6105,
+      "step": 5849
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.788353069207192e-05,
+      "loss": 0.9433,
+      "step": 5850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.787078789759011e-05,
+      "loss": 20.0339,
+      "step": 5851
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.785804457887728e-05,
+      "loss": 20.8058,
+      "step": 5852
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7845300736782204e-05,
+      "loss": 19.2047,
+      "step": 5853
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.783255637215369e-05,
+      "loss": 0.1696,
+      "step": 5854
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.781981148584055e-05,
+      "loss": 0.9567,
+      "step": 5855
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7807066078691676e-05,
+      "loss": 19.2228,
+      "step": 5856
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.779432015155597e-05,
+      "loss": 0.9737,
+      "step": 5857
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.778157370528233e-05,
+      "loss": 20.3105,
+      "step": 5858
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7768826740719775e-05,
+      "loss": 19.6846,
+      "step": 5859
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.77560792587173e-05,
+      "loss": 21.855,
+      "step": 5860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7743331260123925e-05,
+      "loss": 0.9093,
+      "step": 5861
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.773058274578876e-05,
+      "loss": 0.9585,
+      "step": 5862
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7717833716560886e-05,
+      "loss": 20.6547,
+      "step": 5863
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.770508417328946e-05,
+      "loss": 21.6963,
+      "step": 5864
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7692334116823675e-05,
+      "loss": 20.9273,
+      "step": 5865
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7679583548012715e-05,
+      "loss": 22.3793,
+      "step": 5866
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.766683246770585e-05,
+      "loss": 0.1835,
+      "step": 5867
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7654080876752344e-05,
+      "loss": 20.4711,
+      "step": 5868
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.764132877600153e-05,
+      "loss": 0.9218,
+      "step": 5869
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.762857616630275e-05,
+      "loss": 0.1987,
+      "step": 5870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.761582304850537e-05,
+      "loss": 22.3256,
+      "step": 5871
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7603069423458834e-05,
+      "loss": 21.8948,
+      "step": 5872
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.759031529201257e-05,
+      "loss": 0.1935,
+      "step": 5873
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.757756065501608e-05,
+      "loss": 20.9665,
+      "step": 5874
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7564805513318875e-05,
+      "loss": 0.9419,
+      "step": 5875
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.75520498677705e-05,
+      "loss": 20.6444,
+      "step": 5876
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7539293719220555e-05,
+      "loss": 19.985,
+      "step": 5877
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7526537068518636e-05,
+      "loss": 0.9736,
+      "step": 5878
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7513779916514424e-05,
+      "loss": 20.9566,
+      "step": 5879
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.750102226405758e-05,
+      "loss": 20.2745,
+      "step": 5880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.748826411199784e-05,
+      "loss": 20.7711,
+      "step": 5881
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7475505461184966e-05,
+      "loss": 20.5928,
+      "step": 5882
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7462746312468686e-05,
+      "loss": 21.3906,
+      "step": 5883
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.744998666669888e-05,
+      "loss": 0.2099,
+      "step": 5884
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.743722652472538e-05,
+      "loss": 18.7737,
+      "step": 5885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.742446588739807e-05,
+      "loss": 0.9569,
+      "step": 5886
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7411704755566876e-05,
+      "loss": 21.5049,
+      "step": 5887
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7398943130081704e-05,
+      "loss": 21.1083,
+      "step": 5888
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7386181011792614e-05,
+      "loss": 19.6481,
+      "step": 5889
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.737341840154956e-05,
+      "loss": 0.9371,
+      "step": 5890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.73606553002026e-05,
+      "loss": 0.1963,
+      "step": 5891
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.734789170860187e-05,
+      "loss": 0.956,
+      "step": 5892
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.733512762759741e-05,
+      "loss": 19.7132,
+      "step": 5893
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.73223630580394e-05,
+      "loss": 21.0392,
+      "step": 5894
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.730959800077803e-05,
+      "loss": 0.9222,
+      "step": 5895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7296832456663486e-05,
+      "loss": 0.9483,
+      "step": 5896
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.728406642654605e-05,
+      "loss": 20.4431,
+      "step": 5897
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.727129991127596e-05,
+      "loss": 19.5548,
+      "step": 5898
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.725853291170354e-05,
+      "loss": 0.1974,
+      "step": 5899
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7245765428679144e-05,
+      "loss": 20.1741,
+      "step": 5900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.723299746305313e-05,
+      "loss": 0.1965,
+      "step": 5901
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.722022901567591e-05,
+      "loss": 20.2082,
+      "step": 5902
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7207460087397925e-05,
+      "loss": 0.9669,
+      "step": 5903
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.719469067906963e-05,
+      "loss": 0.9268,
+      "step": 5904
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7181920791541546e-05,
+      "loss": 0.9431,
+      "step": 5905
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7169150425664195e-05,
+      "loss": 20.4456,
+      "step": 5906
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.715637958228815e-05,
+      "loss": 19.5993,
+      "step": 5907
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7143608262263994e-05,
+      "loss": 0.9679,
+      "step": 5908
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7130836466442386e-05,
+      "loss": 17.8192,
+      "step": 5909
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7118064195673934e-05,
+      "loss": 0.1937,
+      "step": 5910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7105291450809385e-05,
+      "loss": 0.9919,
+      "step": 5911
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.709251823269945e-05,
+      "loss": 21.8481,
+      "step": 5912
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.707974454219486e-05,
+      "loss": 0.9799,
+      "step": 5913
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7066970380146434e-05,
+      "loss": 22.5055,
+      "step": 5914
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.705419574740497e-05,
+      "loss": 20.7361,
+      "step": 5915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.704142064482131e-05,
+      "loss": 0.9643,
+      "step": 5916
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.702864507324637e-05,
+      "loss": 0.9098,
+      "step": 5917
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.701586903353101e-05,
+      "loss": 0.9721,
+      "step": 5918
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.700309252652624e-05,
+      "loss": 0.9764,
+      "step": 5919
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.699031555308297e-05,
+      "loss": 0.9456,
+      "step": 5920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.697753811405225e-05,
+      "loss": 20.1766,
+      "step": 5921
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6964760210285095e-05,
+      "loss": 0.935,
+      "step": 5922
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6951981842632585e-05,
+      "loss": 20.2168,
+      "step": 5923
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.693920301194581e-05,
+      "loss": 20.1491,
+      "step": 5924
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6926423719075895e-05,
+      "loss": 16.7519,
+      "step": 5925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.691364396487401e-05,
+      "loss": 0.9982,
+      "step": 5926
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6900863750191347e-05,
+      "loss": 0.2183,
+      "step": 5927
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.688808307587912e-05,
+      "loss": 0.9731,
+      "step": 5928
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.687530194278858e-05,
+      "loss": 0.9735,
+      "step": 5929
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.686252035177103e-05,
+      "loss": 0.9289,
+      "step": 5930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.684973830367776e-05,
+      "loss": 0.2257,
+      "step": 5931
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6836955799360125e-05,
+      "loss": 0.9591,
+      "step": 5932
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.682417283966949e-05,
+      "loss": 0.2017,
+      "step": 5933
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6811389425457276e-05,
+      "loss": 0.9381,
+      "step": 5934
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.679860555757489e-05,
+      "loss": 0.2011,
+      "step": 5935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.678582123687381e-05,
+      "loss": 20.4597,
+      "step": 5936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6773036464205556e-05,
+      "loss": 24.4714,
+      "step": 5937
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.676025124042163e-05,
+      "loss": 21.4155,
+      "step": 5938
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.674746556637359e-05,
+      "loss": 21.0269,
+      "step": 5939
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6734679442913006e-05,
+      "loss": 20.243,
+      "step": 5940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.672189287089152e-05,
+      "loss": 0.2272,
+      "step": 5941
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.670910585116077e-05,
+      "loss": 22.7534,
+      "step": 5942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.669631838457241e-05,
+      "loss": 0.8951,
+      "step": 5943
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.668353047197817e-05,
+      "loss": 21.1466,
+      "step": 5944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.667074211422978e-05,
+      "loss": 0.9454,
+      "step": 5945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.665795331217899e-05,
+      "loss": 0.9779,
+      "step": 5946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.664516406667761e-05,
+      "loss": 0.266,
+      "step": 5947
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.663237437857744e-05,
+      "loss": 0.2038,
+      "step": 5948
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.661958424873037e-05,
+      "loss": 0.1803,
+      "step": 5949
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6606793677988247e-05,
+      "loss": 21.1698,
+      "step": 5950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.659400266720301e-05,
+      "loss": 19.5194,
+      "step": 5951
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.658121121722656e-05,
+      "loss": 20.6361,
+      "step": 5952
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.656841932891091e-05,
+      "loss": 21.9315,
+      "step": 5953
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.655562700310803e-05,
+      "loss": 0.9224,
+      "step": 5954
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6542834240669975e-05,
+      "loss": 20.673,
+      "step": 5955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.653004104244878e-05,
+      "loss": 0.208,
+      "step": 5956
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6517247409296535e-05,
+      "loss": 22.9482,
+      "step": 5957
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6504453342065364e-05,
+      "loss": 20.6835,
+      "step": 5958
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.649165884160741e-05,
+      "loss": 0.2381,
+      "step": 5959
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.647886390877482e-05,
+      "loss": 21.901,
+      "step": 5960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6466068544419846e-05,
+      "loss": 19.46,
+      "step": 5961
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.645327274939467e-05,
+      "loss": 0.943,
+      "step": 5962
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.644047652455159e-05,
+      "loss": 20.611,
+      "step": 5963
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.642767987074288e-05,
+      "loss": 20.8232,
+      "step": 5964
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.641488278882082e-05,
+      "loss": 0.9079,
+      "step": 5965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6402085279637816e-05,
+      "loss": 18.8387,
+      "step": 5966
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.638928734404621e-05,
+      "loss": 19.5921,
+      "step": 5967
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.637648898289839e-05,
+      "loss": 20.7572,
+      "step": 5968
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.636369019704682e-05,
+      "loss": 20.6084,
+      "step": 5969
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6350890987343944e-05,
+      "loss": 19.6729,
+      "step": 5970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.633809135464224e-05,
+      "loss": 19.5489,
+      "step": 5971
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6325291299794237e-05,
+      "loss": 0.2426,
+      "step": 5972
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6312490823652465e-05,
+      "loss": 20.4361,
+      "step": 5973
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.62996899270695e-05,
+      "loss": 19.9444,
+      "step": 5974
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6286888610897934e-05,
+      "loss": 19.0127,
+      "step": 5975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6274086875990414e-05,
+      "loss": 20.3654,
+      "step": 5976
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.626128472319958e-05,
+      "loss": 0.1757,
+      "step": 5977
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.624848215337811e-05,
+      "loss": 0.9249,
+      "step": 5978
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.623567916737873e-05,
+      "loss": 19.9384,
+      "step": 5979
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.622287576605417e-05,
+      "loss": 0.221,
+      "step": 5980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.621007195025721e-05,
+      "loss": 0.9764,
+      "step": 5981
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6197267720840604e-05,
+      "loss": 0.8794,
+      "step": 5982
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6184463078657224e-05,
+      "loss": 21.8196,
+      "step": 5983
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6171658024559884e-05,
+      "loss": 0.955,
+      "step": 5984
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.615885255940147e-05,
+      "loss": 0.1642,
+      "step": 5985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6146046684034894e-05,
+      "loss": 17.7264,
+      "step": 5986
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.613324039931306e-05,
+      "loss": 20.6465,
+      "step": 5987
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.612043370608895e-05,
+      "loss": 20.5994,
+      "step": 5988
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.610762660521556e-05,
+      "loss": 0.9558,
+      "step": 5989
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.609481909754586e-05,
+      "loss": 21.5714,
+      "step": 5990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.608201118393294e-05,
+      "loss": 0.201,
+      "step": 5991
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.606920286522982e-05,
+      "loss": 0.2201,
+      "step": 5992
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6056394142289634e-05,
+      "loss": 20.4122,
+      "step": 5993
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.604358501596547e-05,
+      "loss": 20.6517,
+      "step": 5994
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6030775487110485e-05,
+      "loss": 0.9043,
+      "step": 5995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.601796555657788e-05,
+      "loss": 0.9505,
+      "step": 5996
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6005155225220816e-05,
+      "loss": 21.1151,
+      "step": 5997
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.599234449389255e-05,
+      "loss": 0.9262,
+      "step": 5998
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.597953336344631e-05,
+      "loss": 21.4065,
+      "step": 5999
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.596672183473539e-05,
+      "loss": 19.9262,
+      "step": 6000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5953909908613114e-05,
+      "loss": 0.9481,
+      "step": 6001
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5941097585932775e-05,
+      "loss": 20.9808,
+      "step": 6002
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5928284867547774e-05,
+      "loss": 0.1909,
+      "step": 6003
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.591547175431148e-05,
+      "loss": 19.0538,
+      "step": 6004
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.59026582470773e-05,
+      "loss": 20.6123,
+      "step": 6005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5889844346698685e-05,
+      "loss": 0.2321,
+      "step": 6006
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5877030054029104e-05,
+      "loss": 0.234,
+      "step": 6007
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5864215369922037e-05,
+      "loss": 19.8339,
+      "step": 6008
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.585140029523098e-05,
+      "loss": 21.0426,
+      "step": 6009
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.583858483080953e-05,
+      "loss": 21.7842,
+      "step": 6010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.582576897751122e-05,
+      "loss": 20.1198,
+      "step": 6011
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.581295273618964e-05,
+      "loss": 20.384,
+      "step": 6012
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.580013610769844e-05,
+      "loss": 18.8299,
+      "step": 6013
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5787319092891236e-05,
+      "loss": 0.1921,
+      "step": 6014
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5774501692621716e-05,
+      "loss": 19.9817,
+      "step": 6015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.576168390774359e-05,
+      "loss": 0.1692,
+      "step": 6016
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.574886573911056e-05,
+      "loss": 19.3351,
+      "step": 6017
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.573604718757639e-05,
+      "loss": 21.733,
+      "step": 6018
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5723228253994856e-05,
+      "loss": 0.9441,
+      "step": 6019
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.571040893921973e-05,
+      "loss": 0.9409,
+      "step": 6020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.569758924410488e-05,
+      "loss": 19.1534,
+      "step": 6021
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5684769169504115e-05,
+      "loss": 0.957,
+      "step": 6022
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.567194871627137e-05,
+      "loss": 0.9573,
+      "step": 6023
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5659127885260484e-05,
+      "loss": 22.5423,
+      "step": 6024
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.564630667732542e-05,
+      "loss": 0.1897,
+      "step": 6025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5633485093320114e-05,
+      "loss": 22.4228,
+      "step": 6026
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.562066313409856e-05,
+      "loss": 0.2195,
+      "step": 6027
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.560784080051474e-05,
+      "loss": 22.1063,
+      "step": 6028
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.559501809342269e-05,
+      "loss": 0.9219,
+      "step": 6029
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5582195013676476e-05,
+      "loss": 0.9779,
+      "step": 6030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.556937156213015e-05,
+      "loss": 18.969,
+      "step": 6031
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.555654773963782e-05,
+      "loss": 21.4987,
+      "step": 6032
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5543723547053625e-05,
+      "loss": 19.8728,
+      "step": 6033
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.553089898523171e-05,
+      "loss": 20.5954,
+      "step": 6034
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.551807405502624e-05,
+      "loss": 0.9673,
+      "step": 6035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5505248757291416e-05,
+      "loss": 20.0544,
+      "step": 6036
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5492423092881486e-05,
+      "loss": 0.9109,
+      "step": 6037
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.547959706265068e-05,
+      "loss": 19.0099,
+      "step": 6038
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.546677066745326e-05,
+      "loss": 18.8312,
+      "step": 6039
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5453943908143534e-05,
+      "loss": 0.204,
+      "step": 6040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.544111678557585e-05,
+      "loss": 0.914,
+      "step": 6041
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.542828930060451e-05,
+      "loss": 0.9647,
+      "step": 6042
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.541546145408392e-05,
+      "loss": 19.942,
+      "step": 6043
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.540263324686843e-05,
+      "loss": 0.189,
+      "step": 6044
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.538980467981251e-05,
+      "loss": 19.713,
+      "step": 6045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.537697575377057e-05,
+      "loss": 0.2293,
+      "step": 6046
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.536414646959708e-05,
+      "loss": 0.2115,
+      "step": 6047
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.535131682814655e-05,
+      "loss": 19.6916,
+      "step": 6048
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.533848683027346e-05,
+      "loss": 18.8824,
+      "step": 6049
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.532565647683238e-05,
+      "loss": 0.1872,
+      "step": 6050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5312825768677844e-05,
+      "loss": 18.3686,
+      "step": 6051
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.529999470666445e-05,
+      "loss": 19.0862,
+      "step": 6052
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5287163291646814e-05,
+      "loss": 0.19,
+      "step": 6053
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.527433152447955e-05,
+      "loss": 0.9739,
+      "step": 6054
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5261499406017325e-05,
+      "loss": 0.2123,
+      "step": 6055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5248666937114825e-05,
+      "loss": 20.1157,
+      "step": 6056
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.523583411862673e-05,
+      "loss": 22.0335,
+      "step": 6057
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5223000951407786e-05,
+      "loss": 20.3683,
+      "step": 6058
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5210167436312734e-05,
+      "loss": 0.922,
+      "step": 6059
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.519733357419634e-05,
+      "loss": 21.0575,
+      "step": 6060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.51844993659134e-05,
+      "loss": 22.8263,
+      "step": 6061
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.517166481231873e-05,
+      "loss": 0.9694,
+      "step": 6062
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5158829914267194e-05,
+      "loss": 0.1838,
+      "step": 6063
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5145994672613624e-05,
+      "loss": 0.1853,
+      "step": 6064
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5133159088212936e-05,
+      "loss": 19.33,
+      "step": 6065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5120323161919995e-05,
+      "loss": 0.1587,
+      "step": 6066
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.510748689458979e-05,
+      "loss": 21.5093,
+      "step": 6067
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.509465028707724e-05,
+      "loss": 0.1968,
+      "step": 6068
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.508181334023732e-05,
+      "loss": 20.1512,
+      "step": 6069
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.506897605492507e-05,
+      "loss": 20.9713,
+      "step": 6070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5056138431995463e-05,
+      "loss": 18.631,
+      "step": 6071
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.504330047230356e-05,
+      "loss": 0.9353,
+      "step": 6072
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5030462176704446e-05,
+      "loss": 20.7822,
+      "step": 6073
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5017623546053196e-05,
+      "loss": 0.1586,
+      "step": 6074
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5004784581204927e-05,
+      "loss": 20.621,
+      "step": 6075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.499194528301477e-05,
+      "loss": 0.9554,
+      "step": 6076
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.497910565233788e-05,
+      "loss": 20.0168,
+      "step": 6077
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.496626569002944e-05,
+      "loss": 0.2306,
+      "step": 6078
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.495342539694466e-05,
+      "loss": 18.8894,
+      "step": 6079
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.494058477393875e-05,
+      "loss": 1.0021,
+      "step": 6080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.492774382186696e-05,
+      "loss": 0.9546,
+      "step": 6081
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4914902541584556e-05,
+      "loss": 19.0411,
+      "step": 6082
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4902060933946816e-05,
+      "loss": 18.6508,
+      "step": 6083
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.488921899980908e-05,
+      "loss": 19.1831,
+      "step": 6084
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.487637674002666e-05,
+      "loss": 0.9676,
+      "step": 6085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.48635341554549e-05,
+      "loss": 0.2123,
+      "step": 6086
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.485069124694919e-05,
+      "loss": 19.2548,
+      "step": 6087
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.483784801536494e-05,
+      "loss": 20.6286,
+      "step": 6088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.482500446155754e-05,
+      "loss": 0.9158,
+      "step": 6089
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.481216058638246e-05,
+      "loss": 19.3054,
+      "step": 6090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.479931639069512e-05,
+      "loss": 0.2029,
+      "step": 6091
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4786471875351056e-05,
+      "loss": 1.0184,
+      "step": 6092
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.477362704120574e-05,
+      "loss": 0.9776,
+      "step": 6093
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.476078188911469e-05,
+      "loss": 0.9344,
+      "step": 6094
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.474793641993349e-05,
+      "loss": 0.184,
+      "step": 6095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.473509063451766e-05,
+      "loss": 0.9589,
+      "step": 6096
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4722244533722836e-05,
+      "loss": 20.1287,
+      "step": 6097
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4709398118404584e-05,
+      "loss": 0.9681,
+      "step": 6098
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4696551389418546e-05,
+      "loss": 0.8779,
+      "step": 6099
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4683704347620415e-05,
+      "loss": 0.9347,
+      "step": 6100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4670856993865816e-05,
+      "loss": 0.201,
+      "step": 6101
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.465800932901046e-05,
+      "loss": 19.1847,
+      "step": 6102
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.464516135391007e-05,
+      "loss": 0.9207,
+      "step": 6103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.463231306942035e-05,
+      "loss": 0.956,
+      "step": 6104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.461946447639709e-05,
+      "loss": 18.9756,
+      "step": 6105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4606615575696054e-05,
+      "loss": 19.3811,
+      "step": 6106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.459376636817303e-05,
+      "loss": 19.8798,
+      "step": 6107
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.458091685468384e-05,
+      "loss": 19.1679,
+      "step": 6108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.456806703608432e-05,
+      "loss": 0.9385,
+      "step": 6109
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.455521691323033e-05,
+      "loss": 19.3477,
+      "step": 6110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4542366486977756e-05,
+      "loss": 19.2774,
+      "step": 6111
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4529515758182506e-05,
+      "loss": 0.2223,
+      "step": 6112
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.451666472770044e-05,
+      "loss": 22.074,
+      "step": 6113
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.450381339638755e-05,
+      "loss": 20.3795,
+      "step": 6114
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.44909617650998e-05,
+      "loss": 21.5196,
+      "step": 6115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.447810983469313e-05,
+      "loss": 0.914,
+      "step": 6116
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.446525760602357e-05,
+      "loss": 20.2674,
+      "step": 6117
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.44524050799471e-05,
+      "loss": 0.2086,
+      "step": 6118
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.44395522573198e-05,
+      "loss": 0.9379,
+      "step": 6119
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.442669913899773e-05,
+      "loss": 18.958,
+      "step": 6120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.441384572583691e-05,
+      "loss": 21.9084,
+      "step": 6121
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.440099201869351e-05,
+      "loss": 21.8189,
+      "step": 6122
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4388138018423596e-05,
+      "loss": 21.5251,
+      "step": 6123
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.437528372588331e-05,
+      "loss": 0.1959,
+      "step": 6124
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4362429141928826e-05,
+      "loss": 20.2714,
+      "step": 6125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.434957426741631e-05,
+      "loss": 0.9545,
+      "step": 6126
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.433671910320196e-05,
+      "loss": 0.9229,
+      "step": 6127
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.432386365014199e-05,
+      "loss": 0.2284,
+      "step": 6128
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.431100790909263e-05,
+      "loss": 0.9529,
+      "step": 6129
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.429815188091012e-05,
+      "loss": 0.1728,
+      "step": 6130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.428529556645074e-05,
+      "loss": 19.2365,
+      "step": 6131
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4272438966570793e-05,
+      "loss": 0.1667,
+      "step": 6132
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.425958208212657e-05,
+      "loss": 0.1893,
+      "step": 6133
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.424672491397441e-05,
+      "loss": 19.0701,
+      "step": 6134
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.423386746297066e-05,
+      "loss": 20.791,
+      "step": 6135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4221009729971685e-05,
+      "loss": 20.3584,
+      "step": 6136
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4208151715833866e-05,
+      "loss": 20.3861,
+      "step": 6137
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4195293421413596e-05,
+      "loss": 0.2092,
+      "step": 6138
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.418243484756732e-05,
+      "loss": 19.9552,
+      "step": 6139
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4169575995151466e-05,
+      "loss": 20.0078,
+      "step": 6140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.41567168650225e-05,
+      "loss": 21.6752,
+      "step": 6141
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.414385745803689e-05,
+      "loss": 20.2284,
+      "step": 6142
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.413099777505114e-05,
+      "loss": 0.9385,
+      "step": 6143
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.411813781692175e-05,
+      "loss": 22.8117,
+      "step": 6144
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4105277584505286e-05,
+      "loss": 0.9489,
+      "step": 6145
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.409241707865826e-05,
+      "loss": 0.9557,
+      "step": 6146
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4079556300237275e-05,
+      "loss": 0.2276,
+      "step": 6147
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.40666952500989e-05,
+      "loss": 0.217,
+      "step": 6148
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.405383392909973e-05,
+      "loss": 20.6538,
+      "step": 6149
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.404097233809641e-05,
+      "loss": 0.9376,
+      "step": 6150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4028110477945557e-05,
+      "loss": 21.9751,
+      "step": 6151
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4015248349503875e-05,
+      "loss": 0.1873,
+      "step": 6152
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4002385953628e-05,
+      "loss": 19.1957,
+      "step": 6153
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3989523291174635e-05,
+      "loss": 19.6864,
+      "step": 6154
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.397666036300051e-05,
+      "loss": 19.6985,
+      "step": 6155
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3963797169962336e-05,
+      "loss": 0.9527,
+      "step": 6156
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3950933712916883e-05,
+      "loss": 18.5881,
+      "step": 6157
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.39380699927209e-05,
+      "loss": 20.5718,
+      "step": 6158
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.392520601023117e-05,
+      "loss": 0.9529,
+      "step": 6159
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3912341766304495e-05,
+      "loss": 19.8426,
+      "step": 6160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.38994772617977e-05,
+      "loss": 18.8854,
+      "step": 6161
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.388661249756761e-05,
+      "loss": 19.4542,
+      "step": 6162
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3873747474471103e-05,
+      "loss": 20.0335,
+      "step": 6163
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3860882193365034e-05,
+      "loss": 20.8148,
+      "step": 6164
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.384801665510627e-05,
+      "loss": 0.9715,
+      "step": 6165
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.383515086055174e-05,
+      "loss": 0.1738,
+      "step": 6166
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3822284810558364e-05,
+      "loss": 18.6234,
+      "step": 6167
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.380941850598307e-05,
+      "loss": 0.9322,
+      "step": 6168
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.379655194768283e-05,
+      "loss": 0.9276,
+      "step": 6169
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3783685136514596e-05,
+      "loss": 22.0263,
+      "step": 6170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.377081807333536e-05,
+      "loss": 0.2043,
+      "step": 6171
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.375795075900216e-05,
+      "loss": 0.2128,
+      "step": 6172
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.374508319437197e-05,
+      "loss": 0.2028,
+      "step": 6173
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.373221538030188e-05,
+      "loss": 0.183,
+      "step": 6174
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.37193473176489e-05,
+      "loss": 19.6042,
+      "step": 6175
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.370647900727012e-05,
+      "loss": 21.4457,
+      "step": 6176
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.369361045002266e-05,
+      "loss": 20.9348,
+      "step": 6177
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3680741646763575e-05,
+      "loss": 1.003,
+      "step": 6178
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.366787259835001e-05,
+      "loss": 19.5746,
+      "step": 6179
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.365500330563911e-05,
+      "loss": 21.7062,
+      "step": 6180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3642133769488026e-05,
+      "loss": 20.0414,
+      "step": 6181
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.362926399075391e-05,
+      "loss": 0.2261,
+      "step": 6182
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.361639397029398e-05,
+      "loss": 19.8479,
+      "step": 6183
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.360352370896542e-05,
+      "loss": 20.5731,
+      "step": 6184
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3590653207625454e-05,
+      "loss": 0.9026,
+      "step": 6185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.357778246713131e-05,
+      "loss": 20.3089,
+      "step": 6186
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3564911488340254e-05,
+      "loss": 0.9526,
+      "step": 6187
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.355204027210954e-05,
+      "loss": 0.9419,
+      "step": 6188
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.353916881929646e-05,
+      "loss": 19.9294,
+      "step": 6189
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.352629713075829e-05,
+      "loss": 0.1986,
+      "step": 6190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3513425207352375e-05,
+      "loss": 0.174,
+      "step": 6191
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.350055304993604e-05,
+      "loss": 0.9771,
+      "step": 6192
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.34876806593666e-05,
+      "loss": 19.0932,
+      "step": 6193
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.347480803650147e-05,
+      "loss": 20.4036,
+      "step": 6194
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.346193518219795e-05,
+      "loss": 20.4777,
+      "step": 6195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.344906209731352e-05,
+      "loss": 0.9134,
+      "step": 6196
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.343618878270552e-05,
+      "loss": 19.2189,
+      "step": 6197
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.34233152392314e-05,
+      "loss": 22.2171,
+      "step": 6198
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.341044146774859e-05,
+      "loss": 0.962,
+      "step": 6199
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.339756746911456e-05,
+      "loss": 17.9636,
+      "step": 6200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3384693244186756e-05,
+      "loss": 0.9678,
+      "step": 6201
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.337181879382267e-05,
+      "loss": 0.2023,
+      "step": 6202
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.335894411887978e-05,
+      "loss": 21.5142,
+      "step": 6203
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.334606922021565e-05,
+      "loss": 0.9147,
+      "step": 6204
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3333194098687764e-05,
+      "loss": 19.4806,
+      "step": 6205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3320318755153685e-05,
+      "loss": 0.9635,
+      "step": 6206
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3307443190470965e-05,
+      "loss": 0.9383,
+      "step": 6207
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.329456740549716e-05,
+      "loss": 19.0926,
+      "step": 6208
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3281691401089884e-05,
+      "loss": 20.7035,
+      "step": 6209
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3268815178106724e-05,
+      "loss": 20.0066,
+      "step": 6210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.32559387374053e-05,
+      "loss": 0.9938,
+      "step": 6211
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.324306207984323e-05,
+      "loss": 0.2084,
+      "step": 6212
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.323018520627818e-05,
+      "loss": 0.9724,
+      "step": 6213
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.321730811756781e-05,
+      "loss": 20.4287,
+      "step": 6214
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.320443081456977e-05,
+      "loss": 0.977,
+      "step": 6215
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3191553298141786e-05,
+      "loss": 20.0271,
+      "step": 6216
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.31786755691415e-05,
+      "loss": 0.1989,
+      "step": 6217
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.316579762842668e-05,
+      "loss": 0.9561,
+      "step": 6218
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.315291947685507e-05,
+      "loss": 20.8923,
+      "step": 6219
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3140041115284343e-05,
+      "loss": 19.2705,
+      "step": 6220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.312716254457234e-05,
+      "loss": 0.9904,
+      "step": 6221
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3114283765576764e-05,
+      "loss": 0.239,
+      "step": 6222
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.310140477915544e-05,
+      "loss": 22.9112,
+      "step": 6223
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.308852558616617e-05,
+      "loss": 19.9491,
+      "step": 6224
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.307564618746675e-05,
+      "loss": 21.2237,
+      "step": 6225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3062766583915034e-05,
+      "loss": 0.2237,
+      "step": 6226
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.304988677636883e-05,
+      "loss": 20.4146,
+      "step": 6227
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3037006765686e-05,
+      "loss": 0.195,
+      "step": 6228
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.302412655272444e-05,
+      "loss": 19.841,
+      "step": 6229
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3011246138342006e-05,
+      "loss": 20.2304,
+      "step": 6230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.29983655233966e-05,
+      "loss": 19.9661,
+      "step": 6231
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.298548470874614e-05,
+      "loss": 0.2057,
+      "step": 6232
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.297260369524852e-05,
+      "loss": 19.3352,
+      "step": 6233
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2959722483761706e-05,
+      "loss": 0.9288,
+      "step": 6234
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.294684107514364e-05,
+      "loss": 0.1881,
+      "step": 6235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.293395947025227e-05,
+      "loss": 21.3046,
+      "step": 6236
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2921077669945586e-05,
+      "loss": 21.3475,
+      "step": 6237
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2908195675081575e-05,
+      "loss": 0.8988,
+      "step": 6238
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.289531348651822e-05,
+      "loss": 0.9698,
+      "step": 6239
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.288243110511355e-05,
+      "loss": 0.1746,
+      "step": 6240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.28695485317256e-05,
+      "loss": 0.9162,
+      "step": 6241
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.285666576721238e-05,
+      "loss": 0.9386,
+      "step": 6242
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.284378281243195e-05,
+      "loss": 20.7502,
+      "step": 6243
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.283089966824242e-05,
+      "loss": 0.1985,
+      "step": 6244
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2818016335501794e-05,
+      "loss": 19.1265,
+      "step": 6245
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.280513281506823e-05,
+      "loss": 18.8623,
+      "step": 6246
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.279224910779976e-05,
+      "loss": 21.9291,
+      "step": 6247
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.277936521455458e-05,
+      "loss": 0.9122,
+      "step": 6248
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2766481136190746e-05,
+      "loss": 0.9913,
+      "step": 6249
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.275359687356642e-05,
+      "loss": 19.1598,
+      "step": 6250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.274071242753979e-05,
+      "loss": 21.6036,
+      "step": 6251
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.272782779896898e-05,
+      "loss": 0.2088,
+      "step": 6252
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.271494298871217e-05,
+      "loss": 21.1522,
+      "step": 6253
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.270205799762755e-05,
+      "loss": 20.9373,
+      "step": 6254
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.268917282657332e-05,
+      "loss": 0.9181,
+      "step": 6255
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2676287476407705e-05,
+      "loss": 20.1693,
+      "step": 6256
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2663401947988925e-05,
+      "loss": 0.9694,
+      "step": 6257
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.265051624217521e-05,
+      "loss": 0.9412,
+      "step": 6258
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2637630359824794e-05,
+      "loss": 22.6775,
+      "step": 6259
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.262474430179597e-05,
+      "loss": 19.0044,
+      "step": 6260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2611858068946986e-05,
+      "loss": 0.1656,
+      "step": 6261
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.259897166213612e-05,
+      "loss": 19.5145,
+      "step": 6262
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.258608508222169e-05,
+      "loss": 19.6269,
+      "step": 6263
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.257319833006198e-05,
+      "loss": 18.5437,
+      "step": 6264
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.256031140651532e-05,
+      "loss": 0.2168,
+      "step": 6265
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.254742431244005e-05,
+      "loss": 20.137,
+      "step": 6266
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2534537048694465e-05,
+      "loss": 0.1913,
+      "step": 6267
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.252164961613697e-05,
+      "loss": 0.9525,
+      "step": 6268
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.250876201562589e-05,
+      "loss": 20.6597,
+      "step": 6269
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.249587424801962e-05,
+      "loss": 0.9573,
+      "step": 6270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2482986314176555e-05,
+      "loss": 20.4679,
+      "step": 6271
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2470098214955046e-05,
+      "loss": 22.4839,
+      "step": 6272
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.245720995121356e-05,
+      "loss": 19.9353,
+      "step": 6273
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2444321523810466e-05,
+      "loss": 18.5138,
+      "step": 6274
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2431432933604195e-05,
+      "loss": 20.6834,
+      "step": 6275
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.241854418145324e-05,
+      "loss": 19.4163,
+      "step": 6276
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2405655268215994e-05,
+      "loss": 20.3415,
+      "step": 6277
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2392766194750945e-05,
+      "loss": 19.7853,
+      "step": 6278
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2379876961916565e-05,
+      "loss": 19.0791,
+      "step": 6279
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.236698757057131e-05,
+      "loss": 20.0244,
+      "step": 6280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2354098021573714e-05,
+      "loss": 0.972,
+      "step": 6281
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2341208315782266e-05,
+      "loss": 18.9031,
+      "step": 6282
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.232831845405547e-05,
+      "loss": 0.1919,
+      "step": 6283
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.231542843725186e-05,
+      "loss": 21.123,
+      "step": 6284
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2302538266229963e-05,
+      "loss": 20.0293,
+      "step": 6285
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.228964794184832e-05,
+      "loss": 0.1843,
+      "step": 6286
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.227675746496551e-05,
+      "loss": 20.6653,
+      "step": 6287
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.226386683644008e-05,
+      "loss": 0.2027,
+      "step": 6288
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2250976057130596e-05,
+      "loss": 0.2213,
+      "step": 6289
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2238085127895676e-05,
+      "loss": 20.5251,
+      "step": 6290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.222519404959389e-05,
+      "loss": 0.2157,
+      "step": 6291
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.221230282308385e-05,
+      "loss": 18.9799,
+      "step": 6292
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.219941144922418e-05,
+      "loss": 19.3977,
+      "step": 6293
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.218651992887348e-05,
+      "loss": 0.1907,
+      "step": 6294
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.217362826289042e-05,
+      "loss": 0.189,
+      "step": 6295
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.216073645213363e-05,
+      "loss": 20.2581,
+      "step": 6296
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.214784449746174e-05,
+      "loss": 20.1367,
+      "step": 6297
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.213495239973346e-05,
+      "loss": 19.7787,
+      "step": 6298
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.212206015980742e-05,
+      "loss": 0.9648,
+      "step": 6299
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.210916777854233e-05,
+      "loss": 20.7682,
+      "step": 6300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.209627525679688e-05,
+      "loss": 0.9681,
+      "step": 6301
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.208338259542975e-05,
+      "loss": 0.1843,
+      "step": 6302
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.207048979529969e-05,
+      "loss": 21.3817,
+      "step": 6303
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.205759685726539e-05,
+      "loss": 20.5519,
+      "step": 6304
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.204470378218559e-05,
+      "loss": 0.1874,
+      "step": 6305
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2031810570919024e-05,
+      "loss": 18.6275,
+      "step": 6306
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2018917224324436e-05,
+      "loss": 0.2188,
+      "step": 6307
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.20060237432606e-05,
+      "loss": 0.9451,
+      "step": 6308
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1993130128586267e-05,
+      "loss": 0.9063,
+      "step": 6309
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.198023638116021e-05,
+      "loss": 20.1308,
+      "step": 6310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.196734250184121e-05,
+      "loss": 0.2014,
+      "step": 6311
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1954448491488086e-05,
+      "loss": 0.2032,
+      "step": 6312
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.194155435095961e-05,
+      "loss": 0.9547,
+      "step": 6313
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.19286600811146e-05,
+      "loss": 21.0384,
+      "step": 6314
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.191576568281188e-05,
+      "loss": 0.1854,
+      "step": 6315
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.190287115691026e-05,
+      "loss": 0.1922,
+      "step": 6316
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.18899765042686e-05,
+      "loss": 0.2018,
+      "step": 6317
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1877081725745744e-05,
+      "loss": 0.2122,
+      "step": 6318
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.18641868222005e-05,
+      "loss": 20.4844,
+      "step": 6319
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.185129179449178e-05,
+      "loss": 20.057,
+      "step": 6320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1838396643478425e-05,
+      "loss": 0.9467,
+      "step": 6321
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1825501370019326e-05,
+      "loss": 20.4213,
+      "step": 6322
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.181260597497337e-05,
+      "loss": 1.0002,
+      "step": 6323
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1799710459199416e-05,
+      "loss": 19.9908,
+      "step": 6324
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1786814823556415e-05,
+      "loss": 20.5505,
+      "step": 6325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1773919068903245e-05,
+      "loss": 21.0555,
+      "step": 6326
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.176102319609881e-05,
+      "loss": 0.9367,
+      "step": 6327
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.174812720600209e-05,
+      "loss": 21.1838,
+      "step": 6328
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.173523109947197e-05,
+      "loss": 19.966,
+      "step": 6329
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.17223348773674e-05,
+      "loss": 0.9639,
+      "step": 6330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.170943854054733e-05,
+      "loss": 0.1594,
+      "step": 6331
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.169654208987073e-05,
+      "loss": 0.8994,
+      "step": 6332
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1683645526196536e-05,
+      "loss": 0.9501,
+      "step": 6333
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.167074885038373e-05,
+      "loss": 20.2403,
+      "step": 6334
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.165785206329131e-05,
+      "loss": 0.1799,
+      "step": 6335
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.164495516577824e-05,
+      "loss": 0.2087,
+      "step": 6336
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1632058158703523e-05,
+      "loss": 0.9653,
+      "step": 6337
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1619161042926155e-05,
+      "loss": 0.9117,
+      "step": 6338
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.160626381930515e-05,
+      "loss": 20.8849,
+      "step": 6339
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1593366488699515e-05,
+      "loss": 0.1672,
+      "step": 6340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1580469051968264e-05,
+      "loss": 0.248,
+      "step": 6341
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.156757150997045e-05,
+      "loss": 20.0019,
+      "step": 6342
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.155467386356508e-05,
+      "loss": 0.197,
+      "step": 6343
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.154177611361123e-05,
+      "loss": 0.2152,
+      "step": 6344
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.152887826096795e-05,
+      "loss": 18.4506,
+      "step": 6345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.151598030649425e-05,
+      "loss": 0.1757,
+      "step": 6346
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.150308225104923e-05,
+      "loss": 0.9033,
+      "step": 6347
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.149018409549197e-05,
+      "loss": 0.9673,
+      "step": 6348
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.147728584068151e-05,
+      "loss": 20.8254,
+      "step": 6349
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.146438748747698e-05,
+      "loss": 18.7822,
+      "step": 6350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1451489036737435e-05,
+      "loss": 22.1263,
+      "step": 6351
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.143859048932198e-05,
+      "loss": 18.81,
+      "step": 6352
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.142569184608972e-05,
+      "loss": 18.3958,
+      "step": 6353
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.141279310789975e-05,
+      "loss": 0.1819,
+      "step": 6354
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1399894275611225e-05,
+      "loss": 19.4876,
+      "step": 6355
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.138699535008323e-05,
+      "loss": 19.4332,
+      "step": 6356
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1374096332174916e-05,
+      "loss": 20.3319,
+      "step": 6357
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.13611972227454e-05,
+      "loss": 21.7326,
+      "step": 6358
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.134829802265383e-05,
+      "loss": 0.194,
+      "step": 6359
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.133539873275936e-05,
+      "loss": 19.6237,
+      "step": 6360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.132249935392113e-05,
+      "loss": 0.1658,
+      "step": 6361
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.130959988699829e-05,
+      "loss": 21.9497,
+      "step": 6362
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1296700332850046e-05,
+      "loss": 20.921,
+      "step": 6363
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.128380069233552e-05,
+      "loss": 0.9266,
+      "step": 6364
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.127090096631392e-05,
+      "loss": 21.3551,
+      "step": 6365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.125800115564441e-05,
+      "loss": 0.953,
+      "step": 6366
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.124510126118618e-05,
+      "loss": 19.8992,
+      "step": 6367
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.123220128379843e-05,
+      "loss": 0.902,
+      "step": 6368
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.121930122434036e-05,
+      "loss": 0.2129,
+      "step": 6369
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.120640108367116e-05,
+      "loss": 20.2747,
+      "step": 6370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.119350086265004e-05,
+      "loss": 19.3854,
+      "step": 6371
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.118060056213625e-05,
+      "loss": 0.9043,
+      "step": 6372
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.116770018298897e-05,
+      "loss": 0.9467,
+      "step": 6373
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.115479972606742e-05,
+      "loss": 0.1868,
+      "step": 6374
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.114189919223086e-05,
+      "loss": 0.9574,
+      "step": 6375
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1128998582338505e-05,
+      "loss": 20.317,
+      "step": 6376
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1116097897249616e-05,
+      "loss": 19.1564,
+      "step": 6377
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.110319713782342e-05,
+      "loss": 19.2066,
+      "step": 6378
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.109029630491917e-05,
+      "loss": 21.6167,
+      "step": 6379
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.107739539939613e-05,
+      "loss": 20.823,
+      "step": 6380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.106449442211356e-05,
+      "loss": 0.9096,
+      "step": 6381
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1051593373930716e-05,
+      "loss": 0.9632,
+      "step": 6382
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.103869225570687e-05,
+      "loss": 0.9324,
+      "step": 6383
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.10257910683013e-05,
+      "loss": 21.0321,
+      "step": 6384
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.101288981257327e-05,
+      "loss": 19.818,
+      "step": 6385
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.099998848938209e-05,
+      "loss": 18.7373,
+      "step": 6386
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0987087099587035e-05,
+      "loss": 0.1934,
+      "step": 6387
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.097418564404739e-05,
+      "loss": 0.9136,
+      "step": 6388
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0961284123622465e-05,
+      "loss": 0.9131,
+      "step": 6389
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.094838253917156e-05,
+      "loss": 0.9296,
+      "step": 6390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.093548089155397e-05,
+      "loss": 20.744,
+      "step": 6391
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.092257918162901e-05,
+      "loss": 0.1957,
+      "step": 6392
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0909677410255985e-05,
+      "loss": 19.9974,
+      "step": 6393
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.089677557829423e-05,
+      "loss": 0.9693,
+      "step": 6394
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.088387368660306e-05,
+      "loss": 0.9541,
+      "step": 6395
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.08709717360418e-05,
+      "loss": 0.9841,
+      "step": 6396
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.085806972746978e-05,
+      "loss": 0.9199,
+      "step": 6397
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0845167661746315e-05,
+      "loss": 18.8104,
+      "step": 6398
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.083226553973077e-05,
+      "loss": 18.9801,
+      "step": 6399
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.08193633622825e-05,
+      "loss": 0.954,
+      "step": 6400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0806461130260795e-05,
+      "loss": 22.7177,
+      "step": 6401
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.079355884452506e-05,
+      "loss": 0.8927,
+      "step": 6402
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.078065650593461e-05,
+      "loss": 0.2092,
+      "step": 6403
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.076775411534881e-05,
+      "loss": 0.9431,
+      "step": 6404
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0754851673627024e-05,
+      "loss": 20.8378,
+      "step": 6405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.07419491816286e-05,
+      "loss": 20.3839,
+      "step": 6406
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.072904664021294e-05,
+      "loss": 20.5085,
+      "step": 6407
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0716144050239375e-05,
+      "loss": 0.248,
+      "step": 6408
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.070324141256729e-05,
+      "loss": 0.2021,
+      "step": 6409
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.069033872805607e-05,
+      "loss": 19.7256,
+      "step": 6410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0677435997565094e-05,
+      "loss": 23.4105,
+      "step": 6411
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.066453322195373e-05,
+      "loss": 22.0818,
+      "step": 6412
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.065163040208135e-05,
+      "loss": 20.9677,
+      "step": 6413
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.063872753880738e-05,
+      "loss": 0.931,
+      "step": 6414
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.062582463299118e-05,
+      "loss": 17.4394,
+      "step": 6415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0612921685492165e-05,
+      "loss": 0.2101,
+      "step": 6416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.060001869716972e-05,
+      "loss": 19.7805,
+      "step": 6417
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.058711566888323e-05,
+      "loss": 0.9664,
+      "step": 6418
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.057421260149212e-05,
+      "loss": 0.9363,
+      "step": 6419
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.056130949585579e-05,
+      "loss": 21.5323,
+      "step": 6420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.054840635283364e-05,
+      "loss": 20.3207,
+      "step": 6421
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.053550317328509e-05,
+      "loss": 20.9277,
+      "step": 6422
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0522599958069514e-05,
+      "loss": 0.9489,
+      "step": 6423
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0509696708046385e-05,
+      "loss": 19.5774,
+      "step": 6424
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.049679342407508e-05,
+      "loss": 19.7604,
+      "step": 6425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.048389010701501e-05,
+      "loss": 0.9221,
+      "step": 6426
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0470986757725636e-05,
+      "loss": 0.8898,
+      "step": 6427
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.045808337706634e-05,
+      "loss": 0.227,
+      "step": 6428
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0445179965896585e-05,
+      "loss": 0.1907,
+      "step": 6429
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.043227652507576e-05,
+      "loss": 19.5085,
+      "step": 6430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.041937305546332e-05,
+      "loss": 20.3208,
+      "step": 6431
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.040646955791869e-05,
+      "loss": 21.1554,
+      "step": 6432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.039356603330129e-05,
+      "loss": 0.1992,
+      "step": 6433
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0380662482470576e-05,
+      "loss": 19.742,
+      "step": 6434
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0367758906285966e-05,
+      "loss": 0.9656,
+      "step": 6435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.035485530560691e-05,
+      "loss": 20.2401,
+      "step": 6436
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.034195168129283e-05,
+      "loss": 0.1818,
+      "step": 6437
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0329048034203194e-05,
+      "loss": 19.6838,
+      "step": 6438
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0316144365197426e-05,
+      "loss": 20.1576,
+      "step": 6439
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.030324067513499e-05,
+      "loss": 0.1886,
+      "step": 6440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0290336964875295e-05,
+      "loss": 19.4218,
+      "step": 6441
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0277433235277825e-05,
+      "loss": 0.9193,
+      "step": 6442
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.026452948720201e-05,
+      "loss": 20.8729,
+      "step": 6443
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.025162572150731e-05,
+      "loss": 19.4545,
+      "step": 6444
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.023872193905316e-05,
+      "loss": 18.5407,
+      "step": 6445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.022581814069902e-05,
+      "loss": 0.9932,
+      "step": 6446
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.021291432730436e-05,
+      "loss": 21.0775,
+      "step": 6447
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.020001049972859e-05,
+      "loss": 21.1936,
+      "step": 6448
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.018710665883122e-05,
+      "loss": 19.752,
+      "step": 6449
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0174202805471646e-05,
+      "loss": 20.5967,
+      "step": 6450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.016129894050938e-05,
+      "loss": 0.1825,
+      "step": 6451
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.014839506480386e-05,
+      "loss": 0.2007,
+      "step": 6452
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.013549117921452e-05,
+      "loss": 20.5205,
+      "step": 6453
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0122587284600866e-05,
+      "loss": 20.5979,
+      "step": 6454
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0109683381822306e-05,
+      "loss": 19.4165,
+      "step": 6455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.009677947173833e-05,
+      "loss": 20.0145,
+      "step": 6456
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0083875555208394e-05,
+      "loss": 21.71,
+      "step": 6457
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.007097163309196e-05,
+      "loss": 20.3699,
+      "step": 6458
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0058067706248466e-05,
+      "loss": 0.9702,
+      "step": 6459
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0045163775537416e-05,
+      "loss": 20.1408,
+      "step": 6460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.003225984181824e-05,
+      "loss": 0.9732,
+      "step": 6461
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0019355905950405e-05,
+      "loss": 0.8962,
+      "step": 6462
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.000645196879338e-05,
+      "loss": 20.7407,
+      "step": 6463
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.999354803120664e-05,
+      "loss": 0.9945,
+      "step": 6464
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9980644094049606e-05,
+      "loss": 0.1692,
+      "step": 6465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.996774015818177e-05,
+      "loss": 0.2139,
+      "step": 6466
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.995483622446261e-05,
+      "loss": 0.1982,
+      "step": 6467
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.994193229375154e-05,
+      "loss": 0.1977,
+      "step": 6468
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.992902836690806e-05,
+      "loss": 21.3684,
+      "step": 6469
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.991612444479163e-05,
+      "loss": 21.6215,
+      "step": 6470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.990322052826169e-05,
+      "loss": 19.6031,
+      "step": 6471
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9890316618177706e-05,
+      "loss": 20.4264,
+      "step": 6472
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9877412715399166e-05,
+      "loss": 0.2047,
+      "step": 6473
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9864508820785495e-05,
+      "loss": 0.9588,
+      "step": 6474
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.985160493519615e-05,
+      "loss": 0.9815,
+      "step": 6475
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9838701059490624e-05,
+      "loss": 20.0466,
+      "step": 6476
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9825797194528365e-05,
+      "loss": 20.8082,
+      "step": 6477
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.981289334116879e-05,
+      "loss": 0.9006,
+      "step": 6478
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.979998950027141e-05,
+      "loss": 20.3082,
+      "step": 6479
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9787085672695666e-05,
+      "loss": 20.1112,
+      "step": 6480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.977418185930099e-05,
+      "loss": 20.5218,
+      "step": 6481
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.976127806094684e-05,
+      "loss": 0.9759,
+      "step": 6482
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9748374278492705e-05,
+      "loss": 0.9115,
+      "step": 6483
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9735470512798e-05,
+      "loss": 20.6315,
+      "step": 6484
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.972256676472217e-05,
+      "loss": 19.9501,
+      "step": 6485
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.970966303512471e-05,
+      "loss": 0.8931,
+      "step": 6486
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.969675932486503e-05,
+      "loss": 21.1329,
+      "step": 6487
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.968385563480258e-05,
+      "loss": 0.1885,
+      "step": 6488
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9670951965796824e-05,
+      "loss": 19.7965,
+      "step": 6489
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.965804831870718e-05,
+      "loss": 20.8522,
+      "step": 6490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.964514469439311e-05,
+      "loss": 20.8847,
+      "step": 6491
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.963224109371406e-05,
+      "loss": 20.2293,
+      "step": 6492
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.961933751752944e-05,
+      "loss": 18.9248,
+      "step": 6493
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.960643396669872e-05,
+      "loss": 18.8374,
+      "step": 6494
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9593530442081336e-05,
+      "loss": 17.8521,
+      "step": 6495
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.95806269445367e-05,
+      "loss": 0.9408,
+      "step": 6496
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.956772347492425e-05,
+      "loss": 0.2345,
+      "step": 6497
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.955482003410343e-05,
+      "loss": 20.4515,
+      "step": 6498
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.954191662293367e-05,
+      "loss": 0.1935,
+      "step": 6499
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.952901324227437e-05,
+      "loss": 0.9532,
+      "step": 6500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.951610989298498e-05,
+      "loss": 0.1932,
+      "step": 6501
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.950320657592494e-05,
+      "loss": 0.1705,
+      "step": 6502
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.949030329195362e-05,
+      "loss": 0.1814,
+      "step": 6503
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9477400041930484e-05,
+      "loss": 19.6201,
+      "step": 6504
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9464496826714934e-05,
+      "loss": 21.7356,
+      "step": 6505
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.945159364716637e-05,
+      "loss": 21.0176,
+      "step": 6506
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.943869050414421e-05,
+      "loss": 0.9613,
+      "step": 6507
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.942578739850788e-05,
+      "loss": 18.6805,
+      "step": 6508
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.941288433111678e-05,
+      "loss": 0.9457,
+      "step": 6509
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.939998130283029e-05,
+      "loss": 0.2223,
+      "step": 6510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.938707831450785e-05,
+      "loss": 0.9417,
+      "step": 6511
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.937417536700882e-05,
+      "loss": 0.9012,
+      "step": 6512
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.936127246119263e-05,
+      "loss": 0.8935,
+      "step": 6513
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9348369597918665e-05,
+      "loss": 0.9218,
+      "step": 6514
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.933546677804629e-05,
+      "loss": 19.9089,
+      "step": 6515
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9322564002434924e-05,
+      "loss": 0.9231,
+      "step": 6516
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9309661271943926e-05,
+      "loss": 20.7142,
+      "step": 6517
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9296758587432715e-05,
+      "loss": 0.1942,
+      "step": 6518
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928385594976063e-05,
+      "loss": 0.9516,
+      "step": 6519
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927095335978707e-05,
+      "loss": 0.9554,
+      "step": 6520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.925805081837141e-05,
+      "loss": 20.5504,
+      "step": 6521
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9245148326372994e-05,
+      "loss": 0.2247,
+      "step": 6522
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.92322458846512e-05,
+      "loss": 20.3347,
+      "step": 6523
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.921934349406542e-05,
+      "loss": 1.0055,
+      "step": 6524
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.920644115547495e-05,
+      "loss": 0.9271,
+      "step": 6525
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.919353886973921e-05,
+      "loss": 19.0996,
+      "step": 6526
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9180636637717526e-05,
+      "loss": 22.6438,
+      "step": 6527
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.916773446026924e-05,
+      "loss": 19.9263,
+      "step": 6528
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.915483233825369e-05,
+      "loss": 20.0629,
+      "step": 6529
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.914193027253025e-05,
+      "loss": 0.2163,
+      "step": 6530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.912902826395822e-05,
+      "loss": 21.5908,
+      "step": 6531
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9116126313396946e-05,
+      "loss": 0.9185,
+      "step": 6532
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.910322442170578e-05,
+      "loss": 20.1666,
+      "step": 6533
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9090322589744027e-05,
+      "loss": 0.204,
+      "step": 6534
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9077420818371e-05,
+      "loss": 18.9582,
+      "step": 6535
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.906451910844606e-05,
+      "loss": 18.628,
+      "step": 6536
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9051617460828454e-05,
+      "loss": 21.5685,
+      "step": 6537
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9038715876377546e-05,
+      "loss": 0.9776,
+      "step": 6538
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9025814355952605e-05,
+      "loss": 22.674,
+      "step": 6539
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9012912900412976e-05,
+      "loss": 0.914,
+      "step": 6540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.900001151061792e-05,
+      "loss": 0.8966,
+      "step": 6541
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.898711018742672e-05,
+      "loss": 0.1841,
+      "step": 6542
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.897420893169872e-05,
+      "loss": 0.9571,
+      "step": 6543
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8961307744293144e-05,
+      "loss": 18.4401,
+      "step": 6544
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8948406626069295e-05,
+      "loss": 20.5421,
+      "step": 6545
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.893550557788646e-05,
+      "loss": 0.9637,
+      "step": 6546
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.892260460060387e-05,
+      "loss": 18.4517,
+      "step": 6547
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.890970369508083e-05,
+      "loss": 0.1846,
+      "step": 6548
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.88968028621766e-05,
+      "loss": 0.2273,
+      "step": 6549
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.888390210275039e-05,
+      "loss": 20.0656,
+      "step": 6550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.88710014176615e-05,
+      "loss": 0.9288,
+      "step": 6551
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.885810080776916e-05,
+      "loss": 22.6799,
+      "step": 6552
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.88452002739326e-05,
+      "loss": 20.0289,
+      "step": 6553
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.883229981701105e-05,
+      "loss": 20.7065,
+      "step": 6554
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8819399437863775e-05,
+      "loss": 0.2167,
+      "step": 6555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.880649913734996e-05,
+      "loss": 0.8945,
+      "step": 6556
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.879359891632885e-05,
+      "loss": 22.7292,
+      "step": 6557
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.878069877565966e-05,
+      "loss": 20.5521,
+      "step": 6558
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8767798716201585e-05,
+      "loss": 0.9014,
+      "step": 6559
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8754898738813833e-05,
+      "loss": 21.6159,
+      "step": 6560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8741998844355594e-05,
+      "loss": 19.6918,
+      "step": 6561
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.87290990336861e-05,
+      "loss": 18.8255,
+      "step": 6562
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.871619930766449e-05,
+      "loss": 19.718,
+      "step": 6563
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.870329966714996e-05,
+      "loss": 20.5762,
+      "step": 6564
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8690400113001714e-05,
+      "loss": 20.0421,
+      "step": 6565
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.867750064607889e-05,
+      "loss": 19.8221,
+      "step": 6566
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.866460126724065e-05,
+      "loss": 20.8219,
+      "step": 6567
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.865170197734619e-05,
+      "loss": 0.9482,
+      "step": 6568
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.863880277725461e-05,
+      "loss": 0.9418,
+      "step": 6569
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8625903667825095e-05,
+      "loss": 20.5592,
+      "step": 6570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.861300464991679e-05,
+      "loss": 21.0041,
+      "step": 6571
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8600105724388786e-05,
+      "loss": 18.2222,
+      "step": 6572
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.858720689210025e-05,
+      "loss": 18.8731,
+      "step": 6573
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.85743081539103e-05,
+      "loss": 0.9365,
+      "step": 6574
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.856140951067804e-05,
+      "loss": 19.2275,
+      "step": 6575
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8548510963262576e-05,
+      "loss": 0.1864,
+      "step": 6576
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8535612512523044e-05,
+      "loss": 18.6424,
+      "step": 6577
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.85227141593185e-05,
+      "loss": 21.9859,
+      "step": 6578
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.850981590450804e-05,
+      "loss": 18.2057,
+      "step": 6579
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.849691774895078e-05,
+      "loss": 0.15,
+      "step": 6580
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.848401969350577e-05,
+      "loss": 20.5652,
+      "step": 6581
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.847112173903207e-05,
+      "loss": 21.8033,
+      "step": 6582
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.845822388638877e-05,
+      "loss": 20.5952,
+      "step": 6583
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8445326136434924e-05,
+      "loss": 20.2914,
+      "step": 6584
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.843242849002957e-05,
+      "loss": 0.9172,
+      "step": 6585
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.841953094803174e-05,
+      "loss": 22.0747,
+      "step": 6586
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.840663351130051e-05,
+      "loss": 20.0356,
+      "step": 6587
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.839373618069487e-05,
+      "loss": 0.9496,
+      "step": 6588
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.838083895707385e-05,
+      "loss": 20.6698,
+      "step": 6589
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.836794184129649e-05,
+      "loss": 22.3205,
+      "step": 6590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.835504483422177e-05,
+      "loss": 0.9237,
+      "step": 6591
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8342147936708706e-05,
+      "loss": 0.9435,
+      "step": 6592
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.832925114961629e-05,
+      "loss": 19.5677,
+      "step": 6593
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.831635447380348e-05,
+      "loss": 20.5907,
+      "step": 6594
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.830345791012929e-05,
+      "loss": 0.9403,
+      "step": 6595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8290561459452695e-05,
+      "loss": 0.9509,
+      "step": 6596
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8277665122632615e-05,
+      "loss": 20.8175,
+      "step": 6597
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.826476890052804e-05,
+      "loss": 0.1556,
+      "step": 6598
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.825187279399793e-05,
+      "loss": 20.514,
+      "step": 6599
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8238976803901194e-05,
+      "loss": 19.2288,
+      "step": 6600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.822608093109676e-05,
+      "loss": 19.8594,
+      "step": 6601
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.821318517644359e-05,
+      "loss": 17.9813,
+      "step": 6602
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.820028954080059e-05,
+      "loss": 18.8115,
+      "step": 6603
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.818739402502664e-05,
+      "loss": 0.9315,
+      "step": 6604
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.817449862998067e-05,
+      "loss": 18.5071,
+      "step": 6605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8161603356521586e-05,
+      "loss": 0.188,
+      "step": 6606
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.814870820550822e-05,
+      "loss": 20.4589,
+      "step": 6607
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8135813177799496e-05,
+      "loss": 0.1987,
+      "step": 6608
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.812291827425428e-05,
+      "loss": 18.8788,
+      "step": 6609
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.811002349573141e-05,
+      "loss": 21.07,
+      "step": 6610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.809712884308973e-05,
+      "loss": 0.208,
+      "step": 6611
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.808423431718814e-05,
+      "loss": 19.003,
+      "step": 6612
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.807133991888541e-05,
+      "loss": 0.9504,
+      "step": 6613
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8058445649040395e-05,
+      "loss": 0.1932,
+      "step": 6614
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.804555150851194e-05,
+      "loss": 19.4619,
+      "step": 6615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.803265749815879e-05,
+      "loss": 19.6779,
+      "step": 6616
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.801976361883981e-05,
+      "loss": 21.3579,
+      "step": 6617
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8006869871413765e-05,
+      "loss": 19.3569,
+      "step": 6618
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7993976256739416e-05,
+      "loss": 20.9289,
+      "step": 6619
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7981082775675576e-05,
+      "loss": 21.1272,
+      "step": 6620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7968189429081e-05,
+      "loss": 19.8505,
+      "step": 6621
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.795529621781442e-05,
+      "loss": 19.1881,
+      "step": 6622
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.794240314273462e-05,
+      "loss": 18.8438,
+      "step": 6623
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7929510204700314e-05,
+      "loss": 0.1732,
+      "step": 6624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.791661740457026e-05,
+      "loss": 0.9767,
+      "step": 6625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7903724743203126e-05,
+      "loss": 19.7238,
+      "step": 6626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.789083222145768e-05,
+      "loss": 20.8774,
+      "step": 6627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.78779398401926e-05,
+      "loss": 0.8797,
+      "step": 6628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.786504760026655e-05,
+      "loss": 21.7819,
+      "step": 6629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.785215550253826e-05,
+      "loss": 19.5802,
+      "step": 6630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.78392635478664e-05,
+      "loss": 0.2105,
+      "step": 6631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.78263717371096e-05,
+      "loss": 0.9662,
+      "step": 6632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.781348007112652e-05,
+      "loss": 19.3167,
+      "step": 6633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.780058855077584e-05,
+      "loss": 20.5186,
+      "step": 6634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.778769717691616e-05,
+      "loss": 19.27,
+      "step": 6635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.777480595040611e-05,
+      "loss": 20.4893,
+      "step": 6636
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7761914872104336e-05,
+      "loss": 0.9428,
+      "step": 6637
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.774902394286941e-05,
+      "loss": 0.9669,
+      "step": 6638
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.773613316355992e-05,
+      "loss": 20.4473,
+      "step": 6639
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.772324253503452e-05,
+      "loss": 0.2132,
+      "step": 6640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7710352058151683e-05,
+      "loss": 0.2203,
+      "step": 6641
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7697461733770055e-05,
+      "loss": 0.2195,
+      "step": 6642
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.768457156274817e-05,
+      "loss": 19.669,
+      "step": 6643
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.767168154594454e-05,
+      "loss": 19.1368,
+      "step": 6644
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7658791684217745e-05,
+      "loss": 0.18,
+      "step": 6645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.764590197842628e-05,
+      "loss": 19.9651,
+      "step": 6646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7633012429428705e-05,
+      "loss": 20.4549,
+      "step": 6647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.762012303808345e-05,
+      "loss": 22.0757,
+      "step": 6648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.760723380524906e-05,
+      "loss": 0.2259,
+      "step": 6649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7594344731784025e-05,
+      "loss": 0.1891,
+      "step": 6650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.758145581854677e-05,
+      "loss": 0.2221,
+      "step": 6651
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7568567066395796e-05,
+      "loss": 0.927,
+      "step": 6652
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.755567847618956e-05,
+      "loss": 0.9159,
+      "step": 6653
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7542790048786456e-05,
+      "loss": 0.915,
+      "step": 6654
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.752990178504495e-05,
+      "loss": 0.1959,
+      "step": 6655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.751701368582347e-05,
+      "loss": 0.9737,
+      "step": 6656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.750412575198039e-05,
+      "loss": 20.984,
+      "step": 6657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7491237984374106e-05,
+      "loss": 0.9267,
+      "step": 6658
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7478350383863046e-05,
+      "loss": 0.9842,
+      "step": 6659
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7465462951305547e-05,
+      "loss": 20.3368,
+      "step": 6660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7452575687559966e-05,
+      "loss": 20.9586,
+      "step": 6661
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7439688593484686e-05,
+      "loss": 19.0672,
+      "step": 6662
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.742680166993803e-05,
+      "loss": 19.3046,
+      "step": 6663
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.741391491777833e-05,
+      "loss": 20.8495,
+      "step": 6664
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.74010283378639e-05,
+      "loss": 18.1462,
+      "step": 6665
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.738814193105303e-05,
+      "loss": 22.0502,
+      "step": 6666
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7375255698204045e-05,
+      "loss": 22.2636,
+      "step": 6667
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7362369640175204e-05,
+      "loss": 0.2318,
+      "step": 6668
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.734948375782481e-05,
+      "loss": 19.6147,
+      "step": 6669
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.733659805201109e-05,
+      "loss": 21.2422,
+      "step": 6670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7323712523592286e-05,
+      "loss": 18.9982,
+      "step": 6671
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.73108271734267e-05,
+      "loss": 0.9609,
+      "step": 6672
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7297942002372466e-05,
+      "loss": 19.5121,
+      "step": 6673
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.728505701128785e-05,
+      "loss": 0.8939,
+      "step": 6674
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7272172201031054e-05,
+      "loss": 21.2198,
+      "step": 6675
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.725928757246022e-05,
+      "loss": 19.1652,
+      "step": 6676
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7246403126433576e-05,
+      "loss": 0.2148,
+      "step": 6677
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.723351886380927e-05,
+      "loss": 21.4359,
+      "step": 6678
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.722063478544544e-05,
+      "loss": 0.9304,
+      "step": 6679
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7207750892200236e-05,
+      "loss": 20.8848,
+      "step": 6680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.71948671849318e-05,
+      "loss": 18.3203,
+      "step": 6681
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.718198366449822e-05,
+      "loss": 0.9923,
+      "step": 6682
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.71691003317576e-05,
+      "loss": 0.1908,
+      "step": 6683
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7156217187568054e-05,
+      "loss": 21.0706,
+      "step": 6684
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.714333423278764e-05,
+      "loss": 0.1609,
+      "step": 6685
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.713045146827441e-05,
+      "loss": 0.9516,
+      "step": 6686
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.711756889488645e-05,
+      "loss": 19.6308,
+      "step": 6687
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.710468651348179e-05,
+      "loss": 19.894,
+      "step": 6688
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.709180432491844e-05,
+      "loss": 0.9778,
+      "step": 6689
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.707892233005441e-05,
+      "loss": 18.3093,
+      "step": 6690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7066040529747737e-05,
+      "loss": 19.099,
+      "step": 6691
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.705315892485637e-05,
+      "loss": 19.2178,
+      "step": 6692
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.704027751623829e-05,
+      "loss": 0.9404,
+      "step": 6693
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.702739630475149e-05,
+      "loss": 20.2467,
+      "step": 6694
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.701451529125388e-05,
+      "loss": 21.106,
+      "step": 6695
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7001634476603403e-05,
+      "loss": 19.777,
+      "step": 6696
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.698875386165801e-05,
+      "loss": 0.2342,
+      "step": 6697
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.697587344727557e-05,
+      "loss": 0.953,
+      "step": 6698
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6962993234314004e-05,
+      "loss": 0.9551,
+      "step": 6699
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.695011322363119e-05,
+      "loss": 22.1602,
+      "step": 6700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.693723341608498e-05,
+      "loss": 0.9525,
+      "step": 6701
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6924353812533254e-05,
+      "loss": 0.9275,
+      "step": 6702
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6911474413833845e-05,
+      "loss": 0.9194,
+      "step": 6703
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6898595220844574e-05,
+      "loss": 21.0858,
+      "step": 6704
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.688571623442324e-05,
+      "loss": 0.9758,
+      "step": 6705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6872837455427685e-05,
+      "loss": 0.1892,
+      "step": 6706
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.685995888471567e-05,
+      "loss": 0.9216,
+      "step": 6707
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.684708052314495e-05,
+      "loss": 20.2005,
+      "step": 6708
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.683420237157331e-05,
+      "loss": 0.2086,
+      "step": 6709
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.682132443085851e-05,
+      "loss": 18.8332,
+      "step": 6710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6808446701858225e-05,
+      "loss": 0.9735,
+      "step": 6711
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.679556918543023e-05,
+      "loss": 21.5523,
+      "step": 6712
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6782691882432206e-05,
+      "loss": 0.9222,
+      "step": 6713
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6769814793721824e-05,
+      "loss": 22.0012,
+      "step": 6714
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.675693792015676e-05,
+      "loss": 19.7395,
+      "step": 6715
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.674406126259471e-05,
+      "loss": 0.9199,
+      "step": 6716
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.673118482189329e-05,
+      "loss": 19.6008,
+      "step": 6717
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6718308598910114e-05,
+      "loss": 20.1153,
+      "step": 6718
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6705432594502864e-05,
+      "loss": 0.2199,
+      "step": 6719
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.669255680952905e-05,
+      "loss": 0.8787,
+      "step": 6720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6679681244846326e-05,
+      "loss": 20.9865,
+      "step": 6721
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.666680590131225e-05,
+      "loss": 20.3228,
+      "step": 6722
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.665393077978436e-05,
+      "loss": 22.0879,
+      "step": 6723
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6641055881120216e-05,
+      "loss": 0.894,
+      "step": 6724
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.662818120617736e-05,
+      "loss": 0.979,
+      "step": 6725
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.661530675581326e-05,
+      "loss": 0.1673,
+      "step": 6726
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6602432530885456e-05,
+      "loss": 0.9555,
+      "step": 6727
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.658955853225142e-05,
+      "loss": 0.203,
+      "step": 6728
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.657668476076862e-05,
+      "loss": 19.6822,
+      "step": 6729
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6563811217294485e-05,
+      "loss": 18.9224,
+      "step": 6730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.655093790268649e-05,
+      "loss": 0.9174,
+      "step": 6731
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.653806481780205e-05,
+      "loss": 0.8748,
+      "step": 6732
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.652519196349855e-05,
+      "loss": 19.0796,
+      "step": 6733
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6512319340633396e-05,
+      "loss": 0.9254,
+      "step": 6734
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.649944695006398e-05,
+      "loss": 19.4744,
+      "step": 6735
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.648657479264764e-05,
+      "loss": 0.186,
+      "step": 6736
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.647370286924171e-05,
+      "loss": 0.9368,
+      "step": 6737
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.646083118070356e-05,
+      "loss": 0.9341,
+      "step": 6738
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.644795972789048e-05,
+      "loss": 0.9505,
+      "step": 6739
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.643508851165975e-05,
+      "loss": 20.7488,
+      "step": 6740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.64222175328687e-05,
+      "loss": 0.9406,
+      "step": 6741
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6409346792374564e-05,
+      "loss": 19.807,
+      "step": 6742
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.639647629103459e-05,
+      "loss": 21.6743,
+      "step": 6743
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6383606029706046e-05,
+      "loss": 19.5024,
+      "step": 6744
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6370736009246095e-05,
+      "loss": 19.739,
+      "step": 6745
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.635786623051199e-05,
+      "loss": 0.2005,
+      "step": 6746
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.634499669436092e-05,
+      "loss": 0.1731,
+      "step": 6747
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6332127401649997e-05,
+      "loss": 0.2046,
+      "step": 6748
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6319258353236436e-05,
+      "loss": 19.2681,
+      "step": 6749
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.630638954997737e-05,
+      "loss": 0.9309,
+      "step": 6750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6293520992729886e-05,
+      "loss": 0.8819,
+      "step": 6751
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.628065268235111e-05,
+      "loss": 0.9237,
+      "step": 6752
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6267784619698134e-05,
+      "loss": 0.9815,
+      "step": 6753
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.625491680562804e-05,
+      "loss": 0.9741,
+      "step": 6754
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.624204924099785e-05,
+      "loss": 0.1997,
+      "step": 6755
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.622918192666464e-05,
+      "loss": 21.2555,
+      "step": 6756
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.621631486348542e-05,
+      "loss": 0.9777,
+      "step": 6757
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6203448052317176e-05,
+      "loss": 19.7644,
+      "step": 6758
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.619058149401693e-05,
+      "loss": 18.4709,
+      "step": 6759
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.617771518944165e-05,
+      "loss": 22.2358,
+      "step": 6760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.616484913944828e-05,
+      "loss": 0.1708,
+      "step": 6761
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.615198334489373e-05,
+      "loss": 0.1938,
+      "step": 6762
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.613911780663499e-05,
+      "loss": 0.1587,
+      "step": 6763
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.612625252552891e-05,
+      "loss": 0.1843,
+      "step": 6764
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.611338750243238e-05,
+      "loss": 18.5726,
+      "step": 6765
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.610052273820231e-05,
+      "loss": 21.4353,
+      "step": 6766
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.608765823369552e-05,
+      "loss": 20.1806,
+      "step": 6767
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.607479398976884e-05,
+      "loss": 0.2038,
+      "step": 6768
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.606193000727913e-05,
+      "loss": 0.9419,
+      "step": 6769
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.604906628708313e-05,
+      "loss": 19.4862,
+      "step": 6770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.603620283003767e-05,
+      "loss": 0.1768,
+      "step": 6771
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.602333963699951e-05,
+      "loss": 22.1346,
+      "step": 6772
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.601047670882537e-05,
+      "loss": 0.179,
+      "step": 6773
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.599761404637201e-05,
+      "loss": 21.2472,
+      "step": 6774
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.598475165049613e-05,
+      "loss": 20.4445,
+      "step": 6775
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.597188952205445e-05,
+      "loss": 21.3749,
+      "step": 6776
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.59590276619036e-05,
+      "loss": 0.9835,
+      "step": 6777
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.594616607090028e-05,
+      "loss": 0.9562,
+      "step": 6778
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.593330474990113e-05,
+      "loss": 0.9068,
+      "step": 6779
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.592044369976273e-05,
+      "loss": 0.1893,
+      "step": 6780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.590758292134174e-05,
+      "loss": 19.0101,
+      "step": 6781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5894722415494726e-05,
+      "loss": 18.6389,
+      "step": 6782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.588186218307826e-05,
+      "loss": 20.3513,
+      "step": 6783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.586900222494887e-05,
+      "loss": 20.9913,
+      "step": 6784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.585614254196312e-05,
+      "loss": 18.5913,
+      "step": 6785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.584328313497751e-05,
+      "loss": 20.7875,
+      "step": 6786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.583042400484853e-05,
+      "loss": 21.4969,
+      "step": 6787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.581756515243269e-05,
+      "loss": 0.2016,
+      "step": 6788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.580470657858641e-05,
+      "loss": 21.8791,
+      "step": 6789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.579184828416614e-05,
+      "loss": 19.8132,
+      "step": 6790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5778990270028346e-05,
+      "loss": 21.3835,
+      "step": 6791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.576613253702935e-05,
+      "loss": 20.4632,
+      "step": 6792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.57532750860256e-05,
+      "loss": 20.1148,
+      "step": 6793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.574041791787343e-05,
+      "loss": 20.3583,
+      "step": 6794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5727561033429225e-05,
+      "loss": 19.8573,
+      "step": 6795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.571470443354927e-05,
+      "loss": 0.9511,
+      "step": 6796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5701848119089886e-05,
+      "loss": 19.7135,
+      "step": 6797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.568899209090739e-05,
+      "loss": 1.0268,
+      "step": 6798
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.567613634985802e-05,
+      "loss": 0.9501,
+      "step": 6799
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.566328089679804e-05,
+      "loss": 20.3384,
+      "step": 6800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5650425732583706e-05,
+      "loss": 0.8684,
+      "step": 6801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.563757085807118e-05,
+      "loss": 19.4766,
+      "step": 6802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5624716274116695e-05,
+      "loss": 19.08,
+      "step": 6803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.561186198157643e-05,
+      "loss": 0.1609,
+      "step": 6804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.559900798130651e-05,
+      "loss": 20.8061,
+      "step": 6805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.558615427416309e-05,
+      "loss": 0.9249,
+      "step": 6806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.55733008610023e-05,
+      "loss": 0.9681,
+      "step": 6807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.556044774268021e-05,
+      "loss": 0.9177,
+      "step": 6808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.55475949200529e-05,
+      "loss": 0.9179,
+      "step": 6809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.553474239397646e-05,
+      "loss": 0.9741,
+      "step": 6810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.552189016530689e-05,
+      "loss": 22.3075,
+      "step": 6811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.550903823490021e-05,
+      "loss": 19.7501,
+      "step": 6812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.549618660361246e-05,
+      "loss": 18.5625,
+      "step": 6813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.548333527229957e-05,
+      "loss": 18.8776,
+      "step": 6814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.547048424181751e-05,
+      "loss": 19.5443,
+      "step": 6815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.545763351302224e-05,
+      "loss": 0.2056,
+      "step": 6816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.544478308676967e-05,
+      "loss": 0.1896,
+      "step": 6817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.543193296391569e-05,
+      "loss": 0.9125,
+      "step": 6818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.541908314531616e-05,
+      "loss": 0.9267,
+      "step": 6819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5406233631826984e-05,
+      "loss": 0.9739,
+      "step": 6820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5393384424303965e-05,
+      "loss": 0.1987,
+      "step": 6821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.538053552360291e-05,
+      "loss": 20.5706,
+      "step": 6822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.536768693057967e-05,
+      "loss": 19.1224,
+      "step": 6823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5354838646089944e-05,
+      "loss": 0.1969,
+      "step": 6824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.534199067098955e-05,
+      "loss": 21.5308,
+      "step": 6825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.53291430061342e-05,
+      "loss": 19.7195,
+      "step": 6826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.531629565237959e-05,
+      "loss": 22.0485,
+      "step": 6827
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.530344861058145e-05,
+      "loss": 0.8779,
+      "step": 6828
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.529060188159544e-05,
+      "loss": 0.9417,
+      "step": 6829
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.527775546627718e-05,
+      "loss": 0.9606,
+      "step": 6830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.526490936548234e-05,
+      "loss": 0.1764,
+      "step": 6831
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.525206358006654e-05,
+      "loss": 0.9945,
+      "step": 6832
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.523921811088532e-05,
+      "loss": 18.9743,
+      "step": 6833
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.522637295879427e-05,
+      "loss": 0.9568,
+      "step": 6834
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.521352812464896e-05,
+      "loss": 0.1935,
+      "step": 6835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5200683609304886e-05,
+      "loss": 0.9459,
+      "step": 6836
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5187839413617546e-05,
+      "loss": 20.3514,
+      "step": 6837
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.517499553844246e-05,
+      "loss": 0.2075,
+      "step": 6838
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5162151984635074e-05,
+      "loss": 18.8875,
+      "step": 6839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.514930875305082e-05,
+      "loss": 19.8881,
+      "step": 6840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.51364658445451e-05,
+      "loss": 1.0238,
+      "step": 6841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.512362325997336e-05,
+      "loss": 21.5789,
+      "step": 6842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.511078100019093e-05,
+      "loss": 18.5565,
+      "step": 6843
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.509793906605318e-05,
+      "loss": 21.2285,
+      "step": 6844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.508509745841546e-05,
+      "loss": 19.6961,
+      "step": 6845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.507225617813305e-05,
+      "loss": 20.1199,
+      "step": 6846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5059415226061254e-05,
+      "loss": 0.9243,
+      "step": 6847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5046574603055365e-05,
+      "loss": 19.0202,
+      "step": 6848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5033734309970564e-05,
+      "loss": 19.8077,
+      "step": 6849
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.502089434766213e-05,
+      "loss": 19.9225,
+      "step": 6850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5008054716985256e-05,
+      "loss": 19.0311,
+      "step": 6851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4995215418795085e-05,
+      "loss": 19.0927,
+      "step": 6852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4982376453946816e-05,
+      "loss": 18.4313,
+      "step": 6853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.496953782329557e-05,
+      "loss": 0.2247,
+      "step": 6854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.495669952769646e-05,
+      "loss": 0.9562,
+      "step": 6855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.494386156800455e-05,
+      "loss": 0.9667,
+      "step": 6856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4931023945074954e-05,
+      "loss": 0.9804,
+      "step": 6857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4918186659762685e-05,
+      "loss": 18.4118,
+      "step": 6858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.490534971292277e-05,
+      "loss": 22.063,
+      "step": 6859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4892513105410214e-05,
+      "loss": 20.3506,
+      "step": 6860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.487967683808001e-05,
+      "loss": 0.202,
+      "step": 6861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4866840911787076e-05,
+      "loss": 20.2363,
+      "step": 6862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.485400532738638e-05,
+      "loss": 0.2342,
+      "step": 6863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.484117008573282e-05,
+      "loss": 20.7481,
+      "step": 6864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.482833518768127e-05,
+      "loss": 0.9833,
+      "step": 6865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4815500634086606e-05,
+      "loss": 20.7825,
+      "step": 6866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.480266642580368e-05,
+      "loss": 0.9523,
+      "step": 6867
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4789832563687284e-05,
+      "loss": 17.4321,
+      "step": 6868
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.477699904859222e-05,
+      "loss": 1.024,
+      "step": 6869
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4764165881373274e-05,
+      "loss": 20.58,
+      "step": 6870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4751333062885186e-05,
+      "loss": 18.5221,
+      "step": 6871
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.473850059398268e-05,
+      "loss": 19.0071,
+      "step": 6872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.472566847552047e-05,
+      "loss": 20.8556,
+      "step": 6873
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.47128367083532e-05,
+      "loss": 21.6557,
+      "step": 6874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4700005293335556e-05,
+      "loss": 19.6149,
+      "step": 6875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.468717423132218e-05,
+      "loss": 0.2021,
+      "step": 6876
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.467434352316764e-05,
+      "loss": 0.9724,
+      "step": 6877
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4661513169726546e-05,
+      "loss": 20.8791,
+      "step": 6878
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4648683171853456e-05,
+      "loss": 19.9187,
+      "step": 6879
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.463585353040294e-05,
+      "loss": 0.2134,
+      "step": 6880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.462302424622944e-05,
+      "loss": 19.1683,
+      "step": 6881
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.46101953201875e-05,
+      "loss": 21.1243,
+      "step": 6882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.459736675313159e-05,
+      "loss": 22.4993,
+      "step": 6883
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.45845385459161e-05,
+      "loss": 0.9278,
+      "step": 6884
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.45717106993955e-05,
+      "loss": 0.9871,
+      "step": 6885
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.455888321442417e-05,
+      "loss": 0.9524,
+      "step": 6886
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.454605609185647e-05,
+      "loss": 0.9442,
+      "step": 6887
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.453322933254674e-05,
+      "loss": 18.4157,
+      "step": 6888
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.452040293734934e-05,
+      "loss": 19.1123,
+      "step": 6889
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4507576907118525e-05,
+      "loss": 19.2453,
+      "step": 6890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4494751242708576e-05,
+      "loss": 19.5351,
+      "step": 6891
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4481925944973775e-05,
+      "loss": 19.7723,
+      "step": 6892
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4469101014768304e-05,
+      "loss": 21.3043,
+      "step": 6893
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.445627645294638e-05,
+      "loss": 0.2201,
+      "step": 6894
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.44434522603622e-05,
+      "loss": 0.9334,
+      "step": 6895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4430628437869865e-05,
+      "loss": 0.9524,
+      "step": 6896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.441780498632354e-05,
+      "loss": 0.2238,
+      "step": 6897
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4404981906577324e-05,
+      "loss": 19.9239,
+      "step": 6898
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4392159199485264e-05,
+      "loss": 20.8846,
+      "step": 6899
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4379336865901455e-05,
+      "loss": 18.6429,
+      "step": 6900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.436651490667988e-05,
+      "loss": 0.8997,
+      "step": 6901
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.435369332267459e-05,
+      "loss": 0.2497,
+      "step": 6902
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.434087211473952e-05,
+      "loss": 20.3758,
+      "step": 6903
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.432805128372865e-05,
+      "loss": 20.6557,
+      "step": 6904
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.431523083049589e-05,
+      "loss": 19.4648,
+      "step": 6905
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.430241075589513e-05,
+      "loss": 18.4169,
+      "step": 6906
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.428959106078028e-05,
+      "loss": 21.5045,
+      "step": 6907
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4276771746005176e-05,
+      "loss": 21.6276,
+      "step": 6908
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.426395281242362e-05,
+      "loss": 20.0429,
+      "step": 6909
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.425113426088945e-05,
+      "loss": 19.843,
+      "step": 6910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4238316092256426e-05,
+      "loss": 21.3365,
+      "step": 6911
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4225498307378295e-05,
+      "loss": 0.9016,
+      "step": 6912
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.421268090710877e-05,
+      "loss": 19.4567,
+      "step": 6913
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.419986389230158e-05,
+      "loss": 0.2277,
+      "step": 6914
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4187047263810374e-05,
+      "loss": 18.7815,
+      "step": 6915
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4174231022488785e-05,
+      "loss": 19.0979,
+      "step": 6916
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.416141516919048e-05,
+      "loss": 19.2926,
+      "step": 6917
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4148599704769024e-05,
+      "loss": 0.2192,
+      "step": 6918
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4135784630077975e-05,
+      "loss": 0.9215,
+      "step": 6919
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.412296994597092e-05,
+      "loss": 20.9845,
+      "step": 6920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.411015565330132e-05,
+      "loss": 19.465,
+      "step": 6921
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.40973417529227e-05,
+      "loss": 0.9136,
+      "step": 6922
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4084528245688524e-05,
+      "loss": 18.5893,
+      "step": 6923
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.407171513245224e-05,
+      "loss": 0.1842,
+      "step": 6924
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.405890241406723e-05,
+      "loss": 0.2042,
+      "step": 6925
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.404609009138689e-05,
+      "loss": 19.173,
+      "step": 6926
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.403327816526463e-05,
+      "loss": 21.5946,
+      "step": 6927
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4020466636553707e-05,
+      "loss": 0.1932,
+      "step": 6928
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.400765550610747e-05,
+      "loss": 0.1868,
+      "step": 6929
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.39948447747792e-05,
+      "loss": 20.2599,
+      "step": 6930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3982034443422134e-05,
+      "loss": 18.7433,
+      "step": 6931
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.396922451288951e-05,
+      "loss": 0.9266,
+      "step": 6932
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.395641498403455e-05,
+      "loss": 18.8798,
+      "step": 6933
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.394360585771038e-05,
+      "loss": 0.9699,
+      "step": 6934
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.393079713477019e-05,
+      "loss": 21.6273,
+      "step": 6935
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3917988816067085e-05,
+      "loss": 0.9329,
+      "step": 6936
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.390518090245415e-05,
+      "loss": 19.6868,
+      "step": 6937
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.389237339478446e-05,
+      "loss": 17.6873,
+      "step": 6938
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.387956629391106e-05,
+      "loss": 19.8355,
+      "step": 6939
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.386675960068696e-05,
+      "loss": 20.7554,
+      "step": 6940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3853953315965124e-05,
+      "loss": 19.72,
+      "step": 6941
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3841147440598544e-05,
+      "loss": 18.3044,
+      "step": 6942
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3828341975440134e-05,
+      "loss": 0.9446,
+      "step": 6943
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3815536921342795e-05,
+      "loss": 0.1929,
+      "step": 6944
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3802732279159394e-05,
+      "loss": 21.7221,
+      "step": 6945
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3789928049742805e-05,
+      "loss": 18.2937,
+      "step": 6946
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3777124233945835e-05,
+      "loss": 0.2086,
+      "step": 6947
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.376432083262126e-05,
+      "loss": 0.9495,
+      "step": 6948
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3751517846621895e-05,
+      "loss": 0.9735,
+      "step": 6949
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3738715276800436e-05,
+      "loss": 0.9851,
+      "step": 6950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.372591312400959e-05,
+      "loss": 18.3614,
+      "step": 6951
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3713111389102084e-05,
+      "loss": 21.8012,
+      "step": 6952
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.370031007293052e-05,
+      "loss": 20.0933,
+      "step": 6953
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.368750917634755e-05,
+      "loss": 0.1962,
+      "step": 6954
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.367470870020579e-05,
+      "loss": 0.882,
+      "step": 6955
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.366190864535777e-05,
+      "loss": 19.5481,
+      "step": 6956
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.364910901265606e-05,
+      "loss": 20.1763,
+      "step": 6957
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.363630980295319e-05,
+      "loss": 0.935,
+      "step": 6958
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.362351101710162e-05,
+      "loss": 0.1703,
+      "step": 6959
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.36107126559538e-05,
+      "loss": 0.9644,
+      "step": 6960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3597914720362195e-05,
+      "loss": 0.204,
+      "step": 6961
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.358511721117919e-05,
+      "loss": 21.2346,
+      "step": 6962
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.357232012925714e-05,
+      "loss": 0.9252,
+      "step": 6963
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3559523475448414e-05,
+      "loss": 0.9409,
+      "step": 6964
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.354672725060534e-05,
+      "loss": 19.3262,
+      "step": 6965
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.353393145558016e-05,
+      "loss": 0.2173,
+      "step": 6966
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3521136091225175e-05,
+      "loss": 0.9212,
+      "step": 6967
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.350834115839261e-05,
+      "loss": 0.9592,
+      "step": 6968
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.349554665793465e-05,
+      "loss": 19.3632,
+      "step": 6969
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.348275259070346e-05,
+      "loss": 19.7891,
+      "step": 6970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.346995895755123e-05,
+      "loss": 0.1851,
+      "step": 6971
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.345716575933004e-05,
+      "loss": 0.1914,
+      "step": 6972
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3444372996891965e-05,
+      "loss": 18.5464,
+      "step": 6973
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.34315806710891e-05,
+      "loss": 0.921,
+      "step": 6974
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.341878878277344e-05,
+      "loss": 0.2063,
+      "step": 6975
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.340599733279701e-05,
+      "loss": 0.2106,
+      "step": 6976
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.339320632201177e-05,
+      "loss": 18.8193,
+      "step": 6977
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.338041575126964e-05,
+      "loss": 0.9327,
+      "step": 6978
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.336762562142256e-05,
+      "loss": 0.22,
+      "step": 6979
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.335483593332241e-05,
+      "loss": 0.9531,
+      "step": 6980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3342046687821015e-05,
+      "loss": 20.0464,
+      "step": 6981
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.332925788577023e-05,
+      "loss": 20.8937,
+      "step": 6982
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.331646952802184e-05,
+      "loss": 19.5588,
+      "step": 6983
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.33036816154276e-05,
+      "loss": 0.2003,
+      "step": 6984
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.329089414883924e-05,
+      "loss": 0.222,
+      "step": 6985
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.327810712910848e-05,
+      "loss": 0.9382,
+      "step": 6986
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3265320557087005e-05,
+      "loss": 21.2092,
+      "step": 6987
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.325253443362642e-05,
+      "loss": 21.3693,
+      "step": 6988
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.323974875957838e-05,
+      "loss": 0.9407,
+      "step": 6989
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.322696353579445e-05,
+      "loss": 0.92,
+      "step": 6990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.321417876312619e-05,
+      "loss": 19.0544,
+      "step": 6991
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3201394442425115e-05,
+      "loss": 0.9101,
+      "step": 6992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.318861057454275e-05,
+      "loss": 0.2073,
+      "step": 6993
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3175827160330524e-05,
+      "loss": 0.23,
+      "step": 6994
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.316304420063988e-05,
+      "loss": 0.9619,
+      "step": 6995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.315026169632225e-05,
+      "loss": 0.2328,
+      "step": 6996
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.313747964822898e-05,
+      "loss": 17.6493,
+      "step": 6997
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.312469805721141e-05,
+      "loss": 21.9887,
+      "step": 6998
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.31119169241209e-05,
+      "loss": 19.923,
+      "step": 6999
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3099136249808665e-05,
+      "loss": 18.6523,
+      "step": 7000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3086356035126e-05,
+      "loss": 19.621,
+      "step": 7001
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.307357628092413e-05,
+      "loss": 19.2792,
+      "step": 7002
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.306079698805421e-05,
+      "loss": 0.8805,
+      "step": 7003
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3048018157367433e-05,
+      "loss": 0.9304,
+      "step": 7004
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.303523978971492e-05,
+      "loss": 18.8542,
+      "step": 7005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3022461885947765e-05,
+      "loss": 22.789,
+      "step": 7006
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.300968444691704e-05,
+      "loss": 0.968,
+      "step": 7007
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.299690747347378e-05,
+      "loss": 17.3611,
+      "step": 7008
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2984130966469e-05,
+      "loss": 18.6462,
+      "step": 7009
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.297135492675365e-05,
+      "loss": 19.6353,
+      "step": 7010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.29585793551787e-05,
+      "loss": 20.4654,
+      "step": 7011
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.294580425259506e-05,
+      "loss": 0.9478,
+      "step": 7012
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.293302961985358e-05,
+      "loss": 0.9441,
+      "step": 7013
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.292025545780514e-05,
+      "loss": 0.1769,
+      "step": 7014
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.290748176730057e-05,
+      "loss": 20.6042,
+      "step": 7015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.289470854919062e-05,
+      "loss": 20.1306,
+      "step": 7016
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2881935804326064e-05,
+      "loss": 20.1004,
+      "step": 7017
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2869163533557646e-05,
+      "loss": 0.9739,
+      "step": 7018
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.285639173773602e-05,
+      "loss": 0.9654,
+      "step": 7019
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.284362041771186e-05,
+      "loss": 20.5896,
+      "step": 7020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.283084957433582e-05,
+      "loss": 19.9778,
+      "step": 7021
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2818079208458466e-05,
+      "loss": 21.0412,
+      "step": 7022
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.280530932093038e-05,
+      "loss": 20.9463,
+      "step": 7023
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.27925399126021e-05,
+      "loss": 0.1977,
+      "step": 7024
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.27797709843241e-05,
+      "loss": 19.6815,
+      "step": 7025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2767002536946875e-05,
+      "loss": 0.967,
+      "step": 7026
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.275423457132088e-05,
+      "loss": 0.2028,
+      "step": 7027
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.274146708829646e-05,
+      "loss": 0.901,
+      "step": 7028
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.272870008872405e-05,
+      "loss": 0.1564,
+      "step": 7029
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2715933573453956e-05,
+      "loss": 0.2006,
+      "step": 7030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.270316754333653e-05,
+      "loss": 0.9778,
+      "step": 7031
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.269040199922198e-05,
+      "loss": 20.6185,
+      "step": 7032
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2677636941960605e-05,
+      "loss": 21.4445,
+      "step": 7033
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.266487237240261e-05,
+      "loss": 0.2065,
+      "step": 7034
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.265210829139815e-05,
+      "loss": 0.1917,
+      "step": 7035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2639344699797395e-05,
+      "loss": 0.9301,
+      "step": 7036
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.262658159845046e-05,
+      "loss": 20.0519,
+      "step": 7037
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.26138189882074e-05,
+      "loss": 18.7587,
+      "step": 7038
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.260105686991829e-05,
+      "loss": 0.94,
+      "step": 7039
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.258829524443315e-05,
+      "loss": 19.649,
+      "step": 7040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.257553411260194e-05,
+      "loss": 0.9695,
+      "step": 7041
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.256277347527462e-05,
+      "loss": 0.1964,
+      "step": 7042
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.255001333330113e-05,
+      "loss": 0.9732,
+      "step": 7043
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.253725368753132e-05,
+      "loss": 20.6202,
+      "step": 7044
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.252449453881505e-05,
+      "loss": 0.1845,
+      "step": 7045
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2511735888002166e-05,
+      "loss": 20.0485,
+      "step": 7046
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2498977735942426e-05,
+      "loss": 0.976,
+      "step": 7047
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.248622008348558e-05,
+      "loss": 21.3128,
+      "step": 7048
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2473462931481355e-05,
+      "loss": 20.343,
+      "step": 7049
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.246070628077946e-05,
+      "loss": 0.2032,
+      "step": 7050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.244795013222951e-05,
+      "loss": 19.4912,
+      "step": 7051
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.243519448668113e-05,
+      "loss": 18.9688,
+      "step": 7052
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2422439344983925e-05,
+      "loss": 18.3353,
+      "step": 7053
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.240968470798744e-05,
+      "loss": 19.7974,
+      "step": 7054
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2396930576541164e-05,
+      "loss": 20.5101,
+      "step": 7055
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.238417695149465e-05,
+      "loss": 19.3132,
+      "step": 7056
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.237142383369726e-05,
+      "loss": 19.5415,
+      "step": 7057
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.235867122399848e-05,
+      "loss": 19.5269,
+      "step": 7058
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.234591912324767e-05,
+      "loss": 0.9873,
+      "step": 7059
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.233316753229415e-05,
+      "loss": 19.4027,
+      "step": 7060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.232041645198729e-05,
+      "loss": 0.1875,
+      "step": 7061
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.230766588317634e-05,
+      "loss": 18.4281,
+      "step": 7062
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.229491582671055e-05,
+      "loss": 18.6282,
+      "step": 7063
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.228216628343912e-05,
+      "loss": 0.9962,
+      "step": 7064
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2269417254211256e-05,
+      "loss": 0.2312,
+      "step": 7065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.225666873987609e-05,
+      "loss": 20.4311,
+      "step": 7066
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.224392074128271e-05,
+      "loss": 0.2232,
+      "step": 7067
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.223117325928024e-05,
+      "loss": 0.24,
+      "step": 7068
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.221842629471768e-05,
+      "loss": 0.9938,
+      "step": 7069
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.220567984844405e-05,
+      "loss": 0.9504,
+      "step": 7070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.219293392130832e-05,
+      "loss": 19.4781,
+      "step": 7071
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2180188514159455e-05,
+      "loss": 20.1584,
+      "step": 7072
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.216744362784632e-05,
+      "loss": 19.2899,
+      "step": 7073
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.215469926321779e-05,
+      "loss": 19.3399,
+      "step": 7074
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.214195542112273e-05,
+      "loss": 0.944,
+      "step": 7075
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2129212102409904e-05,
+      "loss": 0.9216,
+      "step": 7076
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.211646930792808e-05,
+      "loss": 0.9273,
+      "step": 7077
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.210372703852602e-05,
+      "loss": 19.944,
+      "step": 7078
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.209098529505237e-05,
+      "loss": 19.567,
+      "step": 7079
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.207824407835583e-05,
+      "loss": 22.2289,
+      "step": 7080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.206550338928501e-05,
+      "loss": 0.916,
+      "step": 7081
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.205276322868848e-05,
+      "loss": 20.2979,
+      "step": 7082
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2040023597414825e-05,
+      "loss": 20.287,
+      "step": 7083
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.202728449631255e-05,
+      "loss": 19.262,
+      "step": 7084
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.201454592623012e-05,
+      "loss": 18.0031,
+      "step": 7085
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2001807888016014e-05,
+      "loss": 0.9625,
+      "step": 7086
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.198907038251864e-05,
+      "loss": 0.9285,
+      "step": 7087
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.197633341058636e-05,
+      "loss": 19.2101,
+      "step": 7088
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.19635969730675e-05,
+      "loss": 0.9831,
+      "step": 7089
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1950861070810415e-05,
+      "loss": 18.9194,
+      "step": 7090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.193812570466334e-05,
+      "loss": 22.6276,
+      "step": 7091
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.19253908754745e-05,
+      "loss": 0.1991,
+      "step": 7092
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1912656584092124e-05,
+      "loss": 18.6215,
+      "step": 7093
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.189992283136437e-05,
+      "loss": 18.809,
+      "step": 7094
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1887189618139353e-05,
+      "loss": 0.948,
+      "step": 7095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1874456945265154e-05,
+      "loss": 21.5913,
+      "step": 7096
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.186172481358986e-05,
+      "loss": 0.9832,
+      "step": 7097
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.184899322396147e-05,
+      "loss": 0.9834,
+      "step": 7098
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.183626217722795e-05,
+      "loss": 0.9625,
+      "step": 7099
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.182353167423729e-05,
+      "loss": 0.216,
+      "step": 7100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.181080171583737e-05,
+      "loss": 20.0276,
+      "step": 7101
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.179807230287605e-05,
+      "loss": 20.8495,
+      "step": 7102
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.178534343620123e-05,
+      "loss": 22.6137,
+      "step": 7103
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1772615116660636e-05,
+      "loss": 0.1977,
+      "step": 7104
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.175988734510207e-05,
+      "loss": 0.9627,
+      "step": 7105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.174716012237328e-05,
+      "loss": 19.2943,
+      "step": 7106
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.173443344932191e-05,
+      "loss": 19.5413,
+      "step": 7107
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1721707326795655e-05,
+      "loss": 0.1756,
+      "step": 7108
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.170898175564213e-05,
+      "loss": 0.2226,
+      "step": 7109
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.169625673670888e-05,
+      "loss": 19.3246,
+      "step": 7110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1683532270843504e-05,
+      "loss": 0.2108,
+      "step": 7111
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.167080835889348e-05,
+      "loss": 20.0442,
+      "step": 7112
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.165808500170628e-05,
+      "loss": 0.9335,
+      "step": 7113
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.164536220012934e-05,
+      "loss": 19.1079,
+      "step": 7114
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1632639955010056e-05,
+      "loss": 0.9608,
+      "step": 7115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.161991826719581e-05,
+      "loss": 19.9548,
+      "step": 7116
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1607197137533894e-05,
+      "loss": 0.9145,
+      "step": 7117
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1594476566871607e-05,
+      "loss": 20.2548,
+      "step": 7118
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.158175655605622e-05,
+      "loss": 0.9442,
+      "step": 7119
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.156903710593491e-05,
+      "loss": 0.9553,
+      "step": 7120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1556318217354864e-05,
+      "loss": 0.9641,
+      "step": 7121
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.154359989116323e-05,
+      "loss": 0.184,
+      "step": 7122
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.153088212820709e-05,
+      "loss": 21.0642,
+      "step": 7123
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.151816492933351e-05,
+      "loss": 19.4893,
+      "step": 7124
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1505448295389535e-05,
+      "loss": 0.9442,
+      "step": 7125
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.149273222722213e-05,
+      "loss": 0.892,
+      "step": 7126
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.148001672567825e-05,
+      "loss": 20.2534,
+      "step": 7127
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.146730179160482e-05,
+      "loss": 0.9341,
+      "step": 7128
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1454587425848666e-05,
+      "loss": 20.8279,
+      "step": 7129
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.144187362925668e-05,
+      "loss": 0.9649,
+      "step": 7130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.142916040267565e-05,
+      "loss": 0.9268,
+      "step": 7131
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.14164477469523e-05,
+      "loss": 19.5591,
+      "step": 7132
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.140373566293338e-05,
+      "loss": 19.1414,
+      "step": 7133
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.139102415146556e-05,
+      "loss": 18.5412,
+      "step": 7134
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.137831321339554e-05,
+      "loss": 0.9525,
+      "step": 7135
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.136560284956984e-05,
+      "loss": 20.3817,
+      "step": 7136
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.135289306083508e-05,
+      "loss": 20.8573,
+      "step": 7137
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.13401838480378e-05,
+      "loss": 19.5357,
+      "step": 7138
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1327475212024454e-05,
+      "loss": 0.9598,
+      "step": 7139
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.131476715364152e-05,
+      "loss": 0.203,
+      "step": 7140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1302059673735434e-05,
+      "loss": 20.4559,
+      "step": 7141
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1289352773152525e-05,
+      "loss": 0.9316,
+      "step": 7142
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.127664645273917e-05,
+      "loss": 21.7762,
+      "step": 7143
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.126394071334165e-05,
+      "loss": 0.9305,
+      "step": 7144
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.125123555580624e-05,
+      "loss": 19.5142,
+      "step": 7145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1238530980979136e-05,
+      "loss": 0.9658,
+      "step": 7146
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.122582698970656e-05,
+      "loss": 18.7286,
+      "step": 7147
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.121312358283463e-05,
+      "loss": 20.4125,
+      "step": 7148
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.120042076120945e-05,
+      "loss": 20.3319,
+      "step": 7149
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1187718525677125e-05,
+      "loss": 20.2427,
+      "step": 7150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1175016877083625e-05,
+      "loss": 18.4088,
+      "step": 7151
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.116231581627498e-05,
+      "loss": 21.7683,
+      "step": 7152
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1149615344097124e-05,
+      "loss": 20.0863,
+      "step": 7153
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.113691546139597e-05,
+      "loss": 0.2059,
+      "step": 7154
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.112421616901738e-05,
+      "loss": 21.5657,
+      "step": 7155
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.111151746780719e-05,
+      "loss": 18.9635,
+      "step": 7156
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.109881935861123e-05,
+      "loss": 0.21,
+      "step": 7157
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.10861218422752e-05,
+      "loss": 0.2169,
+      "step": 7158
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.107342491964484e-05,
+      "loss": 20.3209,
+      "step": 7159
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.106072859156582e-05,
+      "loss": 0.2152,
+      "step": 7160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.104803285888376e-05,
+      "loss": 0.9626,
+      "step": 7161
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.103533772244428e-05,
+      "loss": 0.9609,
+      "step": 7162
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.102264318309292e-05,
+      "loss": 0.8961,
+      "step": 7163
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1009949241675186e-05,
+      "loss": 20.4747,
+      "step": 7164
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.099725589903658e-05,
+      "loss": 0.1528,
+      "step": 7165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0984563156022526e-05,
+      "loss": 20.5531,
+      "step": 7166
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0971871013478416e-05,
+      "loss": 20.3263,
+      "step": 7167
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.095917947224958e-05,
+      "loss": 21.5798,
+      "step": 7168
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0946488533181395e-05,
+      "loss": 18.581,
+      "step": 7169
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.093379819711908e-05,
+      "loss": 19.5136,
+      "step": 7170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.092110846490789e-05,
+      "loss": 19.7222,
+      "step": 7171
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.090841933739303e-05,
+      "loss": 0.9286,
+      "step": 7172
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.089573081541964e-05,
+      "loss": 0.882,
+      "step": 7173
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.088304289983284e-05,
+      "loss": 20.6329,
+      "step": 7174
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0870355591477724e-05,
+      "loss": 0.2279,
+      "step": 7175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.085766889119928e-05,
+      "loss": 0.1913,
+      "step": 7176
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0844982799842536e-05,
+      "loss": 0.2118,
+      "step": 7177
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.083229731825243e-05,
+      "loss": 19.8296,
+      "step": 7178
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0819612447273895e-05,
+      "loss": 0.9622,
+      "step": 7179
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.080692818775178e-05,
+      "loss": 17.5062,
+      "step": 7180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0794244540530914e-05,
+      "loss": 21.2601,
+      "step": 7181
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.078156150645611e-05,
+      "loss": 18.8352,
+      "step": 7182
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.07688790863721e-05,
+      "loss": 0.9253,
+      "step": 7183
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0756197281123594e-05,
+      "loss": 18.6284,
+      "step": 7184
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.074351609155527e-05,
+      "loss": 0.1674,
+      "step": 7185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0730835518511716e-05,
+      "loss": 21.0902,
+      "step": 7186
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.071815556283755e-05,
+      "loss": 20.4813,
+      "step": 7187
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.070547622537734e-05,
+      "loss": 0.945,
+      "step": 7188
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.069279750697553e-05,
+      "loss": 0.2165,
+      "step": 7189
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.068011940847662e-05,
+      "loss": 22.3111,
+      "step": 7190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0667441930725036e-05,
+      "loss": 19.0621,
+      "step": 7191
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0654765074565124e-05,
+      "loss": 18.9808,
+      "step": 7192
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0642088840841234e-05,
+      "loss": 18.5445,
+      "step": 7193
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.062941323039769e-05,
+      "loss": 0.9077,
+      "step": 7194
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0616738244078714e-05,
+      "loss": 19.8019,
+      "step": 7195
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.060406388272851e-05,
+      "loss": 0.2182,
+      "step": 7196
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059139014719129e-05,
+      "loss": 17.5926,
+      "step": 7197
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057871703831116e-05,
+      "loss": 0.9375,
+      "step": 7198
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.056604455693221e-05,
+      "loss": 0.9558,
+      "step": 7199
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055337270389847e-05,
+      "loss": 19.678,
+      "step": 7200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0540701480053984e-05,
+      "loss": 18.4587,
+      "step": 7201
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0528030886242676e-05,
+      "loss": 18.589,
+      "step": 7202
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.051536092330847e-05,
+      "loss": 18.6139,
+      "step": 7203
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0502691592095285e-05,
+      "loss": 0.2041,
+      "step": 7204
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0490022893446904e-05,
+      "loss": 20.5465,
+      "step": 7205
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.047735482820714e-05,
+      "loss": 20.001,
+      "step": 7206
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0464687397219784e-05,
+      "loss": 0.927,
+      "step": 7207
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0452020601328476e-05,
+      "loss": 19.7458,
+      "step": 7208
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.043935444137693e-05,
+      "loss": 19.9742,
+      "step": 7209
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.042668891820878e-05,
+      "loss": 0.9637,
+      "step": 7210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0414024032667556e-05,
+      "loss": 19.3228,
+      "step": 7211
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.040135978559684e-05,
+      "loss": 19.6106,
+      "step": 7212
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0388696177840135e-05,
+      "loss": 0.9293,
+      "step": 7213
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.037603321024085e-05,
+      "loss": 19.1922,
+      "step": 7214
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0363370883642455e-05,
+      "loss": 0.9569,
+      "step": 7215
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0350709198888294e-05,
+      "loss": 19.9695,
+      "step": 7216
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.033804815682168e-05,
+      "loss": 20.7652,
+      "step": 7217
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.032538775828591e-05,
+      "loss": 19.2901,
+      "step": 7218
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.031272800412423e-05,
+      "loss": 18.2127,
+      "step": 7219
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.030006889517985e-05,
+      "loss": 19.2945,
+      "step": 7220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0287410432295886e-05,
+      "loss": 0.2102,
+      "step": 7221
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.027475261631548e-05,
+      "loss": 18.091,
+      "step": 7222
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.026209544808171e-05,
+      "loss": 18.8531,
+      "step": 7223
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0249438928437576e-05,
+      "loss": 19.2422,
+      "step": 7224
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.023678305822606e-05,
+      "loss": 0.1686,
+      "step": 7225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.022412783829014e-05,
+      "loss": 0.9306,
+      "step": 7226
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.021147326947267e-05,
+      "loss": 18.0426,
+      "step": 7227
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.019881935261651e-05,
+      "loss": 0.2077,
+      "step": 7228
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.018616608856451e-05,
+      "loss": 19.9672,
+      "step": 7229
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.017351347815939e-05,
+      "loss": 0.1826,
+      "step": 7230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.016086152224388e-05,
+      "loss": 0.1797,
+      "step": 7231
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.014821022166068e-05,
+      "loss": 21.4887,
+      "step": 7232
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0135559577252395e-05,
+      "loss": 20.689,
+      "step": 7233
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.012290958986165e-05,
+      "loss": 19.539,
+      "step": 7234
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0110260260330974e-05,
+      "loss": 0.9266,
+      "step": 7235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0097611589502856e-05,
+      "loss": 0.2245,
+      "step": 7236
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.008496357821979e-05,
+      "loss": 20.7214,
+      "step": 7237
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0072316227324185e-05,
+      "loss": 20.2334,
+      "step": 7238
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.005966953765839e-05,
+      "loss": 0.972,
+      "step": 7239
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.004702351006474e-05,
+      "loss": 19.4759,
+      "step": 7240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0034378145385545e-05,
+      "loss": 19.8036,
+      "step": 7241
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.002173344446303e-05,
+      "loss": 20.0476,
+      "step": 7242
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.000908940813937e-05,
+      "loss": 20.739,
+      "step": 7243
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.999644603725673e-05,
+      "loss": 0.1486,
+      "step": 7244
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.998380333265725e-05,
+      "loss": 20.3375,
+      "step": 7245
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.997116129518293e-05,
+      "loss": 0.2047,
+      "step": 7246
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.995851992567584e-05,
+      "loss": 0.2408,
+      "step": 7247
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.994587922497795e-05,
+      "loss": 20.0409,
+      "step": 7248
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.993323919393117e-05,
+      "loss": 0.181,
+      "step": 7249
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.992059983337736e-05,
+      "loss": 20.9884,
+      "step": 7250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9907961144158426e-05,
+      "loss": 21.2482,
+      "step": 7251
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.989532312711612e-05,
+      "loss": 20.7419,
+      "step": 7252
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.988268578309218e-05,
+      "loss": 0.9347,
+      "step": 7253
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9870049112928374e-05,
+      "loss": 0.9714,
+      "step": 7254
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.985741311746628e-05,
+      "loss": 0.225,
+      "step": 7255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.984477779754758e-05,
+      "loss": 0.1894,
+      "step": 7256
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.983214315401382e-05,
+      "loss": 19.9,
+      "step": 7257
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9819509187706504e-05,
+      "loss": 19.8905,
+      "step": 7258
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.980687589946715e-05,
+      "loss": 21.0031,
+      "step": 7259
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.979424329013719e-05,
+      "loss": 0.1956,
+      "step": 7260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9781611360557974e-05,
+      "loss": 19.8805,
+      "step": 7261
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.976898011157089e-05,
+      "loss": 21.0199,
+      "step": 7262
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.975634954401723e-05,
+      "loss": 0.9593,
+      "step": 7263
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.974371965873825e-05,
+      "loss": 0.1932,
+      "step": 7264
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.973109045657513e-05,
+      "loss": 20.5813,
+      "step": 7265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.971846193836906e-05,
+      "loss": 0.2509,
+      "step": 7266
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9705834104961173e-05,
+      "loss": 18.5666,
+      "step": 7267
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9693206957192494e-05,
+      "loss": 0.223,
+      "step": 7268
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9680580495904085e-05,
+      "loss": 0.9429,
+      "step": 7269
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9667954721936936e-05,
+      "loss": 19.79,
+      "step": 7270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.965532963613195e-05,
+      "loss": 0.1758,
+      "step": 7271
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.964270523933002e-05,
+      "loss": 21.2328,
+      "step": 7272
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.963008153237202e-05,
+      "loss": 0.1983,
+      "step": 7273
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.961745851609873e-05,
+      "loss": 18.8962,
+      "step": 7274
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.960483619135088e-05,
+      "loss": 0.9882,
+      "step": 7275
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.959221455896921e-05,
+      "loss": 21.2998,
+      "step": 7276
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.957959361979436e-05,
+      "loss": 0.8995,
+      "step": 7277
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.956697337466694e-05,
+      "loss": 0.1985,
+      "step": 7278
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.955435382442755e-05,
+      "loss": 0.9462,
+      "step": 7279
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.954173496991665e-05,
+      "loss": 0.9818,
+      "step": 7280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.952911681197477e-05,
+      "loss": 0.9497,
+      "step": 7281
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.951649935144233e-05,
+      "loss": 0.8648,
+      "step": 7282
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9503882589159666e-05,
+      "loss": 19.9267,
+      "step": 7283
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9491266525967155e-05,
+      "loss": 0.183,
+      "step": 7284
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.947865116270508e-05,
+      "loss": 0.1889,
+      "step": 7285
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.94660365002137e-05,
+      "loss": 0.9235,
+      "step": 7286
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9453422539333156e-05,
+      "loss": 19.9831,
+      "step": 7287
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9440809280903645e-05,
+      "loss": 0.9546,
+      "step": 7288
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.942819672576526e-05,
+      "loss": 21.249,
+      "step": 7289
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.941558487475803e-05,
+      "loss": 0.9772,
+      "step": 7290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.940297372872199e-05,
+      "loss": 0.9526,
+      "step": 7291
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.93903632884971e-05,
+      "loss": 20.973,
+      "step": 7292
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9377753554923246e-05,
+      "loss": 18.65,
+      "step": 7293
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9365144528840325e-05,
+      "loss": 20.9477,
+      "step": 7294
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9352536211088164e-05,
+      "loss": 0.926,
+      "step": 7295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9339928602506505e-05,
+      "loss": 20.8482,
+      "step": 7296
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.932732170393507e-05,
+      "loss": 20.7411,
+      "step": 7297
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.931471551621358e-05,
+      "loss": 21.3705,
+      "step": 7298
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.930211004018162e-05,
+      "loss": 18.969,
+      "step": 7299
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.928950527667878e-05,
+      "loss": 0.9446,
+      "step": 7300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9276901226544646e-05,
+      "loss": 19.5101,
+      "step": 7301
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.926429789061864e-05,
+      "loss": 18.646,
+      "step": 7302
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.925169526974024e-05,
+      "loss": 24.3453,
+      "step": 7303
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9239093364748836e-05,
+      "loss": 0.9525,
+      "step": 7304
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9226492176483746e-05,
+      "loss": 0.1916,
+      "step": 7305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9213891705784304e-05,
+      "loss": 0.2007,
+      "step": 7306
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.920129195348973e-05,
+      "loss": 0.9642,
+      "step": 7307
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.918869292043926e-05,
+      "loss": 20.5823,
+      "step": 7308
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.917609460747202e-05,
+      "loss": 18.7974,
+      "step": 7309
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9163497015427117e-05,
+      "loss": 20.4034,
+      "step": 7310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.915090014514365e-05,
+      "loss": 17.6634,
+      "step": 7311
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9138303997460557e-05,
+      "loss": 0.1892,
+      "step": 7312
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9125708573216854e-05,
+      "loss": 18.3655,
+      "step": 7313
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.911311387325145e-05,
+      "loss": 21.2618,
+      "step": 7314
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.910051989840318e-05,
+      "loss": 17.8571,
+      "step": 7315
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.908792664951089e-05,
+      "loss": 0.1769,
+      "step": 7316
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.907533412741335e-05,
+      "loss": 20.6656,
+      "step": 7317
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.906274233294925e-05,
+      "loss": 18.9139,
+      "step": 7318
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9050151266957294e-05,
+      "loss": 21.0129,
+      "step": 7319
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9037560930276104e-05,
+      "loss": 17.3846,
+      "step": 7320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.902497132374423e-05,
+      "loss": 18.474,
+      "step": 7321
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.901238244820021e-05,
+      "loss": 18.6022,
+      "step": 7322
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.899979430448254e-05,
+      "loss": 18.0885,
+      "step": 7323
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8987206893429626e-05,
+      "loss": 18.918,
+      "step": 7324
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8974620215879844e-05,
+      "loss": 0.9199,
+      "step": 7325
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8962034272671544e-05,
+      "loss": 0.9899,
+      "step": 7326
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.894944906464301e-05,
+      "loss": 20.1843,
+      "step": 7327
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.893686459263246e-05,
+      "loss": 0.204,
+      "step": 7328
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.892428085747807e-05,
+      "loss": 18.6279,
+      "step": 7329
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.891169786001802e-05,
+      "loss": 17.8446,
+      "step": 7330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.889911560109034e-05,
+      "loss": 0.9813,
+      "step": 7331
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8886534081533095e-05,
+      "loss": 19.618,
+      "step": 7332
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.887395330218429e-05,
+      "loss": 19.2331,
+      "step": 7333
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8861373263881825e-05,
+      "loss": 0.8836,
+      "step": 7334
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.884879396746361e-05,
+      "loss": 0.2297,
+      "step": 7335
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.88362154137675e-05,
+      "loss": 19.9497,
+      "step": 7336
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.882363760363123e-05,
+      "loss": 0.9242,
+      "step": 7337
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8811060537892595e-05,
+      "loss": 0.9655,
+      "step": 7338
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8798484217389286e-05,
+      "loss": 20.0028,
+      "step": 7339
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8785908642958905e-05,
+      "loss": 0.2096,
+      "step": 7340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8773333815439066e-05,
+      "loss": 0.8357,
+      "step": 7341
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.876075973566732e-05,
+      "loss": 21.3712,
+      "step": 7342
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8748186404481154e-05,
+      "loss": 0.9597,
+      "step": 7343
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.873561382271798e-05,
+      "loss": 0.8977,
+      "step": 7344
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.872304199121524e-05,
+      "loss": 20.2944,
+      "step": 7345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8710470910810245e-05,
+      "loss": 0.9181,
+      "step": 7346
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.869790058234028e-05,
+      "loss": 18.4299,
+      "step": 7347
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.868533100664262e-05,
+      "loss": 19.2844,
+      "step": 7348
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.867276218455445e-05,
+      "loss": 0.2087,
+      "step": 7349
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.866019411691287e-05,
+      "loss": 0.8647,
+      "step": 7350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8647626804555014e-05,
+      "loss": 0.9227,
+      "step": 7351
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.863506024831792e-05,
+      "loss": 18.6249,
+      "step": 7352
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8622494449038574e-05,
+      "loss": 20.6321,
+      "step": 7353
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.860992940755389e-05,
+      "loss": 0.1688,
+      "step": 7354
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.859736512470081e-05,
+      "loss": 19.7469,
+      "step": 7355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.858480160131612e-05,
+      "loss": 20.135,
+      "step": 7356
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.857223883823664e-05,
+      "loss": 21.1583,
+      "step": 7357
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.855967683629913e-05,
+      "loss": 19.2008,
+      "step": 7358
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8547115596340214e-05,
+      "loss": 19.6927,
+      "step": 7359
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8534555119196576e-05,
+      "loss": 19.5101,
+      "step": 7360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.85219954057048e-05,
+      "loss": 18.4813,
+      "step": 7361
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8509436456701394e-05,
+      "loss": 17.5746,
+      "step": 7362
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.849687827302287e-05,
+      "loss": 0.2126,
+      "step": 7363
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8484320855505664e-05,
+      "loss": 18.7908,
+      "step": 7364
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.847176420498612e-05,
+      "loss": 0.1872,
+      "step": 7365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.845920832230061e-05,
+      "loss": 0.1994,
+      "step": 7366
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.844665320828541e-05,
+      "loss": 21.2981,
+      "step": 7367
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.843409886377672e-05,
+      "loss": 18.4091,
+      "step": 7368
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.842154528961073e-05,
+      "loss": 18.9107,
+      "step": 7369
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.840899248662358e-05,
+      "loss": 0.2131,
+      "step": 7370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.839644045565136e-05,
+      "loss": 19.622,
+      "step": 7371
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8383889197530035e-05,
+      "loss": 0.9747,
+      "step": 7372
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.837133871309563e-05,
+      "loss": 21.0276,
+      "step": 7373
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.835878900318406e-05,
+      "loss": 18.891,
+      "step": 7374
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.834624006863117e-05,
+      "loss": 19.1165,
+      "step": 7375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.833369191027277e-05,
+      "loss": 0.1847,
+      "step": 7376
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.832114452894468e-05,
+      "loss": 18.4405,
+      "step": 7377
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.830859792548256e-05,
+      "loss": 0.8882,
+      "step": 7378
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.829605210072208e-05,
+      "loss": 18.97,
+      "step": 7379
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8283507055498886e-05,
+      "loss": 20.6175,
+      "step": 7380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.82709627906485e-05,
+      "loss": 0.9698,
+      "step": 7381
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8258419307006426e-05,
+      "loss": 17.3484,
+      "step": 7382
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8245876605408166e-05,
+      "loss": 0.2331,
+      "step": 7383
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.823333468668905e-05,
+      "loss": 0.9817,
+      "step": 7384
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8220793551684484e-05,
+      "loss": 19.6333,
+      "step": 7385
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.820825320122976e-05,
+      "loss": 0.9526,
+      "step": 7386
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8195713636160077e-05,
+      "loss": 20.2422,
+      "step": 7387
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.818317485731069e-05,
+      "loss": 19.3643,
+      "step": 7388
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.817063686551671e-05,
+      "loss": 21.5191,
+      "step": 7389
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.815809966161322e-05,
+      "loss": 0.1862,
+      "step": 7390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.814556324643526e-05,
+      "loss": 0.909,
+      "step": 7391
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8133027620817826e-05,
+      "loss": 18.7369,
+      "step": 7392
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.812049278559584e-05,
+      "loss": 19.0478,
+      "step": 7393
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.810795874160417e-05,
+      "loss": 20.1769,
+      "step": 7394
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.809542548967766e-05,
+      "loss": 17.3375,
+      "step": 7395
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.808289303065108e-05,
+      "loss": 20.719,
+      "step": 7396
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.807036136535913e-05,
+      "loss": 21.0063,
+      "step": 7397
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8057830494636506e-05,
+      "loss": 18.7159,
+      "step": 7398
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8045300419317806e-05,
+      "loss": 19.8902,
+      "step": 7399
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.803277114023759e-05,
+      "loss": 21.6021,
+      "step": 7400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.802024265823036e-05,
+      "loss": 19.4013,
+      "step": 7401
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.80077149741306e-05,
+      "loss": 20.9502,
+      "step": 7402
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.799518808877268e-05,
+      "loss": 0.9656,
+      "step": 7403
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.798266200299094e-05,
+      "loss": 21.7192,
+      "step": 7404
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.797013671761973e-05,
+      "loss": 0.9652,
+      "step": 7405
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.795761223349325e-05,
+      "loss": 0.9356,
+      "step": 7406
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7945088551445693e-05,
+      "loss": 0.1887,
+      "step": 7407
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.793256567231121e-05,
+      "loss": 0.9268,
+      "step": 7408
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.792004359692385e-05,
+      "loss": 19.8306,
+      "step": 7409
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7907522326117676e-05,
+      "loss": 20.167,
+      "step": 7410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.789500186072663e-05,
+      "loss": 0.2229,
+      "step": 7411
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.788248220158468e-05,
+      "loss": 19.2626,
+      "step": 7412
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.786996334952565e-05,
+      "loss": 20.3433,
+      "step": 7413
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.785744530538337e-05,
+      "loss": 0.2016,
+      "step": 7414
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.784492806999162e-05,
+      "loss": 0.9507,
+      "step": 7415
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7832411644184064e-05,
+      "loss": 20.0532,
+      "step": 7416
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.781989602879438e-05,
+      "loss": 19.5485,
+      "step": 7417
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.780738122465617e-05,
+      "loss": 18.3366,
+      "step": 7418
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7794867232602955e-05,
+      "loss": 19.6118,
+      "step": 7419
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7782354053468243e-05,
+      "loss": 19.9208,
+      "step": 7420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7769841688085484e-05,
+      "loss": 19.7009,
+      "step": 7421
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.775733013728803e-05,
+      "loss": 0.2379,
+      "step": 7422
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.77448194019092e-05,
+      "loss": 18.7567,
+      "step": 7423
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.773230948278232e-05,
+      "loss": 0.9536,
+      "step": 7424
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.771980038074057e-05,
+      "loss": 0.1702,
+      "step": 7425
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.77072920966171e-05,
+      "loss": 19.4968,
+      "step": 7426
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.769478463124507e-05,
+      "loss": 21.9122,
+      "step": 7427
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7682277985457484e-05,
+      "loss": 18.8479,
+      "step": 7428
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.766977216008736e-05,
+      "loss": 0.8985,
+      "step": 7429
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.765726715596768e-05,
+      "loss": 0.9213,
+      "step": 7430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.764476297393127e-05,
+      "loss": 0.9762,
+      "step": 7431
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7632259614811014e-05,
+      "loss": 20.1806,
+      "step": 7432
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.761975707943967e-05,
+      "loss": 18.9574,
+      "step": 7433
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.760725536864999e-05,
+      "loss": 0.1915,
+      "step": 7434
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.759475448327462e-05,
+      "loss": 20.5606,
+      "step": 7435
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.758225442414619e-05,
+      "loss": 18.0114,
+      "step": 7436
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7569755192097265e-05,
+      "loss": 0.9355,
+      "step": 7437
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.755725678796034e-05,
+      "loss": 21.3899,
+      "step": 7438
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.754475921256788e-05,
+      "loss": 1.0282,
+      "step": 7439
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7532262466752284e-05,
+      "loss": 19.1723,
+      "step": 7440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.751976655134586e-05,
+      "loss": 17.5798,
+      "step": 7441
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.750727146718093e-05,
+      "loss": 0.9275,
+      "step": 7442
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7494777215089715e-05,
+      "loss": 0.9811,
+      "step": 7443
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.748228379590438e-05,
+      "loss": 19.5421,
+      "step": 7444
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.746979121045706e-05,
+      "loss": 21.3538,
+      "step": 7445
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.745729945957981e-05,
+      "loss": 0.9155,
+      "step": 7446
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.744480854410465e-05,
+      "loss": 0.929,
+      "step": 7447
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.74323184648635e-05,
+      "loss": 0.9666,
+      "step": 7448
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.74198292226883e-05,
+      "loss": 20.33,
+      "step": 7449
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.740734081841086e-05,
+      "loss": 19.4179,
+      "step": 7450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7394853252862985e-05,
+      "loss": 0.9525,
+      "step": 7451
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.73823665268764e-05,
+      "loss": 19.3546,
+      "step": 7452
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.736988064128277e-05,
+      "loss": 19.2119,
+      "step": 7453
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7357395596913715e-05,
+      "loss": 18.6921,
+      "step": 7454
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7344911394600814e-05,
+      "loss": 0.2074,
+      "step": 7455
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.733242803517556e-05,
+      "loss": 0.9342,
+      "step": 7456
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.73199455194694e-05,
+      "loss": 20.0468,
+      "step": 7457
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7307463848313716e-05,
+      "loss": 20.24,
+      "step": 7458
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.729498302253989e-05,
+      "loss": 20.8296,
+      "step": 7459
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.728250304297915e-05,
+      "loss": 19.2691,
+      "step": 7460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7270023910462735e-05,
+      "loss": 0.8981,
+      "step": 7461
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.725754562582185e-05,
+      "loss": 17.0392,
+      "step": 7462
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.724506818988756e-05,
+      "loss": 21.2277,
+      "step": 7463
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.723259160349094e-05,
+      "loss": 20.1017,
+      "step": 7464
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7220115867463e-05,
+      "loss": 0.2188,
+      "step": 7465
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7207640982634644e-05,
+      "loss": 0.1979,
+      "step": 7466
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7195166949836794e-05,
+      "loss": 20.8653,
+      "step": 7467
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.718269376990028e-05,
+      "loss": 18.6915,
+      "step": 7468
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.717022144365584e-05,
+      "loss": 0.9293,
+      "step": 7469
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.715774997193421e-05,
+      "loss": 20.7251,
+      "step": 7470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.714527935556606e-05,
+      "loss": 0.9371,
+      "step": 7471
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7132809595381975e-05,
+      "loss": 18.2427,
+      "step": 7472
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.712034069221248e-05,
+      "loss": 0.9553,
+      "step": 7473
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7107872646888116e-05,
+      "loss": 0.2053,
+      "step": 7474
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7095405460239266e-05,
+      "loss": 0.1924,
+      "step": 7475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.70829391330963e-05,
+      "loss": 20.1815,
+      "step": 7476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7070473666289564e-05,
+      "loss": 0.2166,
+      "step": 7477
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7058009060649315e-05,
+      "loss": 0.8763,
+      "step": 7478
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.704554531700573e-05,
+      "loss": 0.9046,
+      "step": 7479
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.703308243618895e-05,
+      "loss": 19.5447,
+      "step": 7480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7020620419029094e-05,
+      "loss": 19.314,
+      "step": 7481
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.700815926635616e-05,
+      "loss": 0.1875,
+      "step": 7482
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.699569897900012e-05,
+      "loss": 20.3323,
+      "step": 7483
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.698323955779092e-05,
+      "loss": 0.938,
+      "step": 7484
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.697078100355837e-05,
+      "loss": 0.2184,
+      "step": 7485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6958323317132284e-05,
+      "loss": 0.206,
+      "step": 7486
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.694586649934244e-05,
+      "loss": 18.6766,
+      "step": 7487
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.693341055101844e-05,
+      "loss": 21.5077,
+      "step": 7488
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.692095547298998e-05,
+      "loss": 0.187,
+      "step": 7489
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.69085012660866e-05,
+      "loss": 21.0565,
+      "step": 7490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.689604793113779e-05,
+      "loss": 19.5731,
+      "step": 7491
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.688359546897303e-05,
+      "loss": 20.7016,
+      "step": 7492
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.687114388042171e-05,
+      "loss": 0.1929,
+      "step": 7493
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.685869316631314e-05,
+      "loss": 0.1975,
+      "step": 7494
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6846243327476596e-05,
+      "loss": 19.4915,
+      "step": 7495
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.683379436474133e-05,
+      "loss": 0.1985,
+      "step": 7496
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.682134627893647e-05,
+      "loss": 19.4506,
+      "step": 7497
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.680889907089112e-05,
+      "loss": 0.2113,
+      "step": 7498
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.679645274143433e-05,
+      "loss": 19.6842,
+      "step": 7499
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.678400729139509e-05,
+      "loss": 19.3106,
+      "step": 7500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.677156272160229e-05,
+      "loss": 0.1852,
+      "step": 7501
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.675911903288484e-05,
+      "loss": 0.9535,
+      "step": 7502
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.674667622607153e-05,
+      "loss": 19.1902,
+      "step": 7503
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6734234301991096e-05,
+      "loss": 19.9526,
+      "step": 7504
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6721793261472226e-05,
+      "loss": 19.6242,
+      "step": 7505
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.670935310534359e-05,
+      "loss": 0.9554,
+      "step": 7506
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.669691383443372e-05,
+      "loss": 0.1875,
+      "step": 7507
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6684475449571135e-05,
+      "loss": 0.1895,
+      "step": 7508
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.667203795158432e-05,
+      "loss": 0.1903,
+      "step": 7509
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.665960134130163e-05,
+      "loss": 0.2023,
+      "step": 7510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.664716561955143e-05,
+      "loss": 20.2348,
+      "step": 7511
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.663473078716199e-05,
+      "loss": 0.1592,
+      "step": 7512
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.662229684496149e-05,
+      "loss": 18.9659,
+      "step": 7513
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6609863793778145e-05,
+      "loss": 0.9677,
+      "step": 7514
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.659743163444003e-05,
+      "loss": 21.1093,
+      "step": 7515
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.658500036777517e-05,
+      "loss": 19.4899,
+      "step": 7516
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.657256999461157e-05,
+      "loss": 0.1786,
+      "step": 7517
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.656014051577713e-05,
+      "loss": 21.3689,
+      "step": 7518
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.654771193209975e-05,
+      "loss": 0.2301,
+      "step": 7519
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.653528424440716e-05,
+      "loss": 0.2204,
+      "step": 7520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.652285745352717e-05,
+      "loss": 20.0309,
+      "step": 7521
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6510431560287446e-05,
+      "loss": 18.6046,
+      "step": 7522
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6498006565515576e-05,
+      "loss": 19.1276,
+      "step": 7523
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.648558247003915e-05,
+      "loss": 21.0695,
+      "step": 7524
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6473159274685686e-05,
+      "loss": 19.304,
+      "step": 7525
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.646073698028259e-05,
+      "loss": 0.1617,
+      "step": 7526
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.644831558765726e-05,
+      "loss": 1.0174,
+      "step": 7527
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.643589509763703e-05,
+      "loss": 0.2067,
+      "step": 7528
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.642347551104914e-05,
+      "loss": 19.2892,
+      "step": 7529
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.641105682872081e-05,
+      "loss": 18.1125,
+      "step": 7530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.639863905147919e-05,
+      "loss": 0.9575,
+      "step": 7531
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.638622218015133e-05,
+      "loss": 0.9567,
+      "step": 7532
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6373806215564265e-05,
+      "loss": 19.9162,
+      "step": 7533
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6361391158544994e-05,
+      "loss": 20.3716,
+      "step": 7534
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6348977009920346e-05,
+      "loss": 0.9461,
+      "step": 7535
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.633656377051722e-05,
+      "loss": 0.1968,
+      "step": 7536
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.632415144116237e-05,
+      "loss": 21.6245,
+      "step": 7537
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.631174002268251e-05,
+      "loss": 21.1916,
+      "step": 7538
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.629932951590431e-05,
+      "loss": 19.9541,
+      "step": 7539
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.628691992165435e-05,
+      "loss": 18.7719,
+      "step": 7540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.627451124075921e-05,
+      "loss": 20.0005,
+      "step": 7541
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6262103474045324e-05,
+      "loss": 19.1031,
+      "step": 7542
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.624969662233911e-05,
+      "loss": 19.0226,
+      "step": 7543
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6237290686466955e-05,
+      "loss": 0.9484,
+      "step": 7544
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.62248856672551e-05,
+      "loss": 20.2795,
+      "step": 7545
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.621248156552982e-05,
+      "loss": 18.4915,
+      "step": 7546
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.620007838211728e-05,
+      "loss": 0.9378,
+      "step": 7547
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.618767611784356e-05,
+      "loss": 0.9231,
+      "step": 7548
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6175274773534734e-05,
+      "loss": 0.8946,
+      "step": 7549
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.61628743500168e-05,
+      "loss": 0.9409,
+      "step": 7550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.615047484811565e-05,
+      "loss": 0.9341,
+      "step": 7551
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.613807626865716e-05,
+      "loss": 18.6861,
+      "step": 7552
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.612567861246716e-05,
+      "loss": 0.2222,
+      "step": 7553
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6113281880371364e-05,
+      "loss": 18.1778,
+      "step": 7554
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.610088607319544e-05,
+      "loss": 0.2038,
+      "step": 7555
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.608849119176505e-05,
+      "loss": 0.9255,
+      "step": 7556
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6076097236905713e-05,
+      "loss": 0.9361,
+      "step": 7557
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6063704209442926e-05,
+      "loss": 0.1873,
+      "step": 7558
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.605131211020217e-05,
+      "loss": 0.217,
+      "step": 7559
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.603892094000874e-05,
+      "loss": 18.8582,
+      "step": 7560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6026530699687995e-05,
+      "loss": 19.3409,
+      "step": 7561
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.601414139006516e-05,
+      "loss": 0.9679,
+      "step": 7562
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6001753011965455e-05,
+      "loss": 19.1307,
+      "step": 7563
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5989365566213975e-05,
+      "loss": 19.0514,
+      "step": 7564
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.597697905363577e-05,
+      "loss": 21.0811,
+      "step": 7565
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5964593475055886e-05,
+      "loss": 0.2154,
+      "step": 7566
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.595220883129921e-05,
+      "loss": 19.7796,
+      "step": 7567
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5939825123190635e-05,
+      "loss": 0.1956,
+      "step": 7568
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.592744235155499e-05,
+      "loss": 0.9673,
+      "step": 7569
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5915060517216995e-05,
+      "loss": 0.9211,
+      "step": 7570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.590267962100136e-05,
+      "loss": 0.2045,
+      "step": 7571
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.589029966373271e-05,
+      "loss": 0.2537,
+      "step": 7572
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.587792064623559e-05,
+      "loss": 20.5792,
+      "step": 7573
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.586554256933451e-05,
+      "loss": 0.8907,
+      "step": 7574
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5853165433853916e-05,
+      "loss": 0.9118,
+      "step": 7575
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5840789240618175e-05,
+      "loss": 19.9913,
+      "step": 7576
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.582841399045159e-05,
+      "loss": 0.2625,
+      "step": 7577
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.581603968417843e-05,
+      "loss": 0.1654,
+      "step": 7578
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.580366632262287e-05,
+      "loss": 17.7079,
+      "step": 7579
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5791293906609015e-05,
+      "loss": 0.2063,
+      "step": 7580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.577892243696096e-05,
+      "loss": 19.447,
+      "step": 7581
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5766551914502696e-05,
+      "loss": 0.9202,
+      "step": 7582
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5754182340058136e-05,
+      "loss": 0.9385,
+      "step": 7583
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5741813714451155e-05,
+      "loss": 0.2021,
+      "step": 7584
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5729446038505585e-05,
+      "loss": 0.9383,
+      "step": 7585
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.571707931304514e-05,
+      "loss": 0.2153,
+      "step": 7586
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5704713538893514e-05,
+      "loss": 0.2039,
+      "step": 7587
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.569234871687435e-05,
+      "loss": 19.7161,
+      "step": 7588
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5679984847811154e-05,
+      "loss": 0.1919,
+      "step": 7589
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.566762193252744e-05,
+      "loss": 0.2306,
+      "step": 7590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.565525997184667e-05,
+      "loss": 0.9911,
+      "step": 7591
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.564289896659214e-05,
+      "loss": 0.9244,
+      "step": 7592
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.56305389175872e-05,
+      "loss": 0.938,
+      "step": 7593
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.561817982565508e-05,
+      "loss": 20.3394,
+      "step": 7594
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.560582169161894e-05,
+      "loss": 21.3398,
+      "step": 7595
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5593464516301884e-05,
+      "loss": 19.8287,
+      "step": 7596
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.558110830052699e-05,
+      "loss": 20.8971,
+      "step": 7597
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.55687530451172e-05,
+      "loss": 19.1019,
+      "step": 7598
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.555639875089544e-05,
+      "loss": 19.473,
+      "step": 7599
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.554404541868459e-05,
+      "loss": 17.8173,
+      "step": 7600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.553169304930741e-05,
+      "loss": 18.418,
+      "step": 7601
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5519341643586634e-05,
+      "loss": 19.9489,
+      "step": 7602
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.550699120234493e-05,
+      "loss": 18.6516,
+      "step": 7603
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.54946417264049e-05,
+      "loss": 0.2163,
+      "step": 7604
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5482293216589054e-05,
+      "loss": 19.5031,
+      "step": 7605
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5469945673719876e-05,
+      "loss": 0.9566,
+      "step": 7606
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5457599098619776e-05,
+      "loss": 0.2416,
+      "step": 7607
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.544525349211107e-05,
+      "loss": 0.1997,
+      "step": 7608
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.543290885501604e-05,
+      "loss": 19.4554,
+      "step": 7609
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.542056518815692e-05,
+      "loss": 0.9821,
+      "step": 7610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.540822249235583e-05,
+      "loss": 0.936,
+      "step": 7611
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5395880768434845e-05,
+      "loss": 20.6249,
+      "step": 7612
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.538354001721602e-05,
+      "loss": 19.6441,
+      "step": 7613
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5371200239521264e-05,
+      "loss": 21.8877,
+      "step": 7614
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5358861436172485e-05,
+      "loss": 20.282,
+      "step": 7615
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.534652360799151e-05,
+      "loss": 21.1369,
+      "step": 7616
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5334186755800065e-05,
+      "loss": 18.2672,
+      "step": 7617
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.532185088041987e-05,
+      "loss": 0.9263,
+      "step": 7618
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.530951598267256e-05,
+      "loss": 18.8695,
+      "step": 7619
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.529718206337965e-05,
+      "loss": 0.2033,
+      "step": 7620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5284849123362684e-05,
+      "loss": 19.4275,
+      "step": 7621
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.527251716344309e-05,
+      "loss": 18.3083,
+      "step": 7622
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5260186184442204e-05,
+      "loss": 18.92,
+      "step": 7623
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.524785618718133e-05,
+      "loss": 22.2073,
+      "step": 7624
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.523552717248172e-05,
+      "loss": 18.3861,
+      "step": 7625
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5223199141164555e-05,
+      "loss": 0.9547,
+      "step": 7626
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.52108720940509e-05,
+      "loss": 19.3482,
+      "step": 7627
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.519854603196183e-05,
+      "loss": 18.4938,
+      "step": 7628
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5186220955718306e-05,
+      "loss": 21.4023,
+      "step": 7629
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.517389686614123e-05,
+      "loss": 19.48,
+      "step": 7630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.516157376405144e-05,
+      "loss": 22.1292,
+      "step": 7631
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.514925165026973e-05,
+      "loss": 17.1948,
+      "step": 7632
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5136930525616796e-05,
+      "loss": 20.0115,
+      "step": 7633
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5124610390913274e-05,
+      "loss": 0.9645,
+      "step": 7634
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.511229124697977e-05,
+      "loss": 0.9298,
+      "step": 7635
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5099973094636774e-05,
+      "loss": 0.1518,
+      "step": 7636
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.508765593470472e-05,
+      "loss": 21.4792,
+      "step": 7637
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.507533976800405e-05,
+      "loss": 21.0253,
+      "step": 7638
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.506302459535498e-05,
+      "loss": 0.9592,
+      "step": 7639
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.505071041757784e-05,
+      "loss": 0.1903,
+      "step": 7640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.503839723549278e-05,
+      "loss": 19.4479,
+      "step": 7641
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5026085049919906e-05,
+      "loss": 0.8913,
+      "step": 7642
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5013773861679285e-05,
+      "loss": 0.1801,
+      "step": 7643
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.50014636715909e-05,
+      "loss": 19.8123,
+      "step": 7644
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.498915448047464e-05,
+      "loss": 21.0108,
+      "step": 7645
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.497684628915039e-05,
+      "loss": 19.5825,
+      "step": 7646
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.496453909843791e-05,
+      "loss": 18.6198,
+      "step": 7647
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.495223290915693e-05,
+      "loss": 0.9399,
+      "step": 7648
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.493992772212708e-05,
+      "loss": 19.3232,
+      "step": 7649
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.492762353816795e-05,
+      "loss": 19.4736,
+      "step": 7650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.491532035809908e-05,
+      "loss": 18.5904,
+      "step": 7651
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4903018182739874e-05,
+      "loss": 0.9319,
+      "step": 7652
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.489071701290975e-05,
+      "loss": 20.5836,
+      "step": 7653
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.487841684942802e-05,
+      "loss": 19.4781,
+      "step": 7654
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.486611769311392e-05,
+      "loss": 0.9468,
+      "step": 7655
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.485381954478663e-05,
+      "loss": 0.9587,
+      "step": 7656
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.484152240526527e-05,
+      "loss": 0.2012,
+      "step": 7657
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4829226275368886e-05,
+      "loss": 20.6085,
+      "step": 7658
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4816931155916445e-05,
+      "loss": 20.2656,
+      "step": 7659
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4804637047726895e-05,
+      "loss": 0.9403,
+      "step": 7660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.479234395161904e-05,
+      "loss": 20.8902,
+      "step": 7661
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4780051868411675e-05,
+      "loss": 19.9705,
+      "step": 7662
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4767760798923515e-05,
+      "loss": 19.61,
+      "step": 7663
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4755470743973175e-05,
+      "loss": 0.1793,
+      "step": 7664
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.474318170437927e-05,
+      "loss": 0.925,
+      "step": 7665
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.473089368096026e-05,
+      "loss": 19.6172,
+      "step": 7666
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4718606674534647e-05,
+      "loss": 0.2163,
+      "step": 7667
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.470632068592075e-05,
+      "loss": 20.404,
+      "step": 7668
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4694035715936885e-05,
+      "loss": 20.1153,
+      "step": 7669
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4681751765401316e-05,
+      "loss": 20.0682,
+      "step": 7670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.466946883513217e-05,
+      "loss": 18.8906,
+      "step": 7671
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.465718692594756e-05,
+      "loss": 0.2048,
+      "step": 7672
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4644906038665544e-05,
+      "loss": 0.2177,
+      "step": 7673
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.463262617410405e-05,
+      "loss": 0.9773,
+      "step": 7674
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.462034733308098e-05,
+      "loss": 20.5784,
+      "step": 7675
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4608069516414196e-05,
+      "loss": 18.8908,
+      "step": 7676
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.459579272492141e-05,
+      "loss": 20.8349,
+      "step": 7677
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.458351695942035e-05,
+      "loss": 0.9653,
+      "step": 7678
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4571242220728634e-05,
+      "loss": 0.975,
+      "step": 7679
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4558968509663795e-05,
+      "loss": 0.9149,
+      "step": 7680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.454669582704332e-05,
+      "loss": 18.1821,
+      "step": 7681
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4534424173684666e-05,
+      "loss": 0.9705,
+      "step": 7682
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.452215355040514e-05,
+      "loss": 19.182,
+      "step": 7683
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4509883958022026e-05,
+      "loss": 20.5,
+      "step": 7684
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.449761539735257e-05,
+      "loss": 20.1435,
+      "step": 7685
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.448534786921388e-05,
+      "loss": 19.8146,
+      "step": 7686
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.447308137442304e-05,
+      "loss": 0.164,
+      "step": 7687
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4460815913797055e-05,
+      "loss": 19.9758,
+      "step": 7688
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.444855148815288e-05,
+      "loss": 19.1432,
+      "step": 7689
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4436288098307364e-05,
+      "loss": 0.951,
+      "step": 7690
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4424025745077295e-05,
+      "loss": 19.135,
+      "step": 7691
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.441176442927943e-05,
+      "loss": 0.1778,
+      "step": 7692
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4399504151730414e-05,
+      "loss": 0.2215,
+      "step": 7693
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.438724491324682e-05,
+      "loss": 0.1831,
+      "step": 7694
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.437498671464523e-05,
+      "loss": 17.8647,
+      "step": 7695
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4362729556742025e-05,
+      "loss": 0.2208,
+      "step": 7696
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4350473440353634e-05,
+      "loss": 19.6463,
+      "step": 7697
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.433821836629637e-05,
+      "loss": 18.5471,
+      "step": 7698
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.432596433538644e-05,
+      "loss": 1.0034,
+      "step": 7699
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.431371134844006e-05,
+      "loss": 0.1858,
+      "step": 7700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.430145940627333e-05,
+      "loss": 17.7941,
+      "step": 7701
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4289208509702265e-05,
+      "loss": 18.9198,
+      "step": 7702
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.427695865954284e-05,
+      "loss": 0.893,
+      "step": 7703
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4264709856610974e-05,
+      "loss": 19.0404,
+      "step": 7704
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.425246210172246e-05,
+      "loss": 0.9168,
+      "step": 7705
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4240215395693067e-05,
+      "loss": 20.482,
+      "step": 7706
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.422796973933851e-05,
+      "loss": 18.0255,
+      "step": 7707
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.421572513347437e-05,
+      "loss": 18.694,
+      "step": 7708
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4203481578916194e-05,
+      "loss": 19.4435,
+      "step": 7709
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.419123907647949e-05,
+      "loss": 0.2063,
+      "step": 7710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.417899762697966e-05,
+      "loss": 1.0001,
+      "step": 7711
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4166757231232014e-05,
+      "loss": 16.763,
+      "step": 7712
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.415451789005183e-05,
+      "loss": 19.2105,
+      "step": 7713
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4142279604254334e-05,
+      "loss": 0.2248,
+      "step": 7714
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.413004237465462e-05,
+      "loss": 19.7676,
+      "step": 7715
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4117806202067736e-05,
+      "loss": 0.9741,
+      "step": 7716
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4105571087308715e-05,
+      "loss": 20.0929,
+      "step": 7717
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.409333703119243e-05,
+      "loss": 0.97,
+      "step": 7718
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4081104034533735e-05,
+      "loss": 18.5718,
+      "step": 7719
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.406887209814742e-05,
+      "loss": 0.9074,
+      "step": 7720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4056641222848155e-05,
+      "loss": 19.7827,
+      "step": 7721
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.404441140945061e-05,
+      "loss": 19.6648,
+      "step": 7722
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.403218265876934e-05,
+      "loss": 20.4132,
+      "step": 7723
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.401995497161881e-05,
+      "loss": 21.0964,
+      "step": 7724
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.400772834881347e-05,
+      "loss": 19.3165,
+      "step": 7725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3995502791167664e-05,
+      "loss": 19.9516,
+      "step": 7726
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.398327829949567e-05,
+      "loss": 19.3441,
+      "step": 7727
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3971054874611674e-05,
+      "loss": 20.5655,
+      "step": 7728
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.395883251732985e-05,
+      "loss": 20.7825,
+      "step": 7729
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.394661122846423e-05,
+      "loss": 0.9471,
+      "step": 7730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.393439100882883e-05,
+      "loss": 0.2247,
+      "step": 7731
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.392217185923756e-05,
+      "loss": 0.9657,
+      "step": 7732
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.39099537805043e-05,
+      "loss": 0.9682,
+      "step": 7733
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.389773677344279e-05,
+      "loss": 19.7553,
+      "step": 7734
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.388552083886676e-05,
+      "loss": 0.1893,
+      "step": 7735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3873305977589856e-05,
+      "loss": 20.2893,
+      "step": 7736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3861092190425636e-05,
+      "loss": 17.6199,
+      "step": 7737
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.384887947818758e-05,
+      "loss": 22.2211,
+      "step": 7738
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.383666784168914e-05,
+      "loss": 0.9134,
+      "step": 7739
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3824457281743646e-05,
+      "loss": 0.9357,
+      "step": 7740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.381224779916439e-05,
+      "loss": 20.1997,
+      "step": 7741
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3800039394764586e-05,
+      "loss": 20.4324,
+      "step": 7742
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.378783206935734e-05,
+      "loss": 19.3281,
+      "step": 7743
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.377562582375575e-05,
+      "loss": 20.4741,
+      "step": 7744
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.376342065877279e-05,
+      "loss": 21.0111,
+      "step": 7745
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3751216575221376e-05,
+      "loss": 19.8237,
+      "step": 7746
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.373901357391438e-05,
+      "loss": 19.746,
+      "step": 7747
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3726811655664576e-05,
+      "loss": 19.3487,
+      "step": 7748
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3714610821284634e-05,
+      "loss": 19.7663,
+      "step": 7749
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3702411071587214e-05,
+      "loss": 18.4537,
+      "step": 7750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.369021240738489e-05,
+      "loss": 0.1957,
+      "step": 7751
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.367801482949012e-05,
+      "loss": 0.9244,
+      "step": 7752
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.366581833871532e-05,
+      "loss": 20.0789,
+      "step": 7753
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3653622935872845e-05,
+      "loss": 0.2354,
+      "step": 7754
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.364142862177499e-05,
+      "loss": 0.922,
+      "step": 7755
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.362923539723389e-05,
+      "loss": 20.6612,
+      "step": 7756
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3617043263061726e-05,
+      "loss": 21.5522,
+      "step": 7757
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.360485222007054e-05,
+      "loss": 18.6656,
+      "step": 7758
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.359266226907229e-05,
+      "loss": 0.1946,
+      "step": 7759
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.35804734108789e-05,
+      "loss": 18.9444,
+      "step": 7760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.356828564630221e-05,
+      "loss": 20.6429,
+      "step": 7761
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.355609897615397e-05,
+      "loss": 0.9737,
+      "step": 7762
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.354391340124586e-05,
+      "loss": 20.5049,
+      "step": 7763
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3531728922389524e-05,
+      "loss": 18.1717,
+      "step": 7764
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.351954554039649e-05,
+      "loss": 0.9861,
+      "step": 7765
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3507363256078215e-05,
+      "loss": 18.4407,
+      "step": 7766
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3495182070246124e-05,
+      "loss": 0.9122,
+      "step": 7767
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3483001983711506e-05,
+      "loss": 0.9337,
+      "step": 7768
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.347082299728563e-05,
+      "loss": 18.9063,
+      "step": 7769
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3458645111779694e-05,
+      "loss": 0.2074,
+      "step": 7770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.344646832800475e-05,
+      "loss": 20.7019,
+      "step": 7771
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.343429264677188e-05,
+      "loss": 19.4685,
+      "step": 7772
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3422118068892e-05,
+      "loss": 19.4384,
+      "step": 7773
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.340994459517605e-05,
+      "loss": 0.9333,
+      "step": 7774
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.339777222643477e-05,
+      "loss": 20.0772,
+      "step": 7775
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.338560096347893e-05,
+      "loss": 19.6645,
+      "step": 7776
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.337343080711921e-05,
+      "loss": 0.2007,
+      "step": 7777
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3361261758166155e-05,
+      "loss": 18.8958,
+      "step": 7778
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.334909381743032e-05,
+      "loss": 20.7853,
+      "step": 7779
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3336926985722145e-05,
+      "loss": 20.5576,
+      "step": 7780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.332476126385196e-05,
+      "loss": 0.9757,
+      "step": 7781
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.331259665263009e-05,
+      "loss": 0.9126,
+      "step": 7782
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.330043315286676e-05,
+      "loss": 21.0658,
+      "step": 7783
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3288270765372096e-05,
+      "loss": 20.5574,
+      "step": 7784
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.327610949095616e-05,
+      "loss": 0.175,
+      "step": 7785
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3263949330428995e-05,
+      "loss": 19.1556,
+      "step": 7786
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.325179028460048e-05,
+      "loss": 0.962,
+      "step": 7787
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.323963235428047e-05,
+      "loss": 20.223,
+      "step": 7788
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.322747554027878e-05,
+      "loss": 0.9022,
+      "step": 7789
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.321531984340504e-05,
+      "loss": 20.7926,
+      "step": 7790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.320316526446894e-05,
+      "loss": 19.2029,
+      "step": 7791
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.319101180428e-05,
+      "loss": 17.7951,
+      "step": 7792
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.317885946364769e-05,
+      "loss": 17.5081,
+      "step": 7793
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.316670824338144e-05,
+      "loss": 0.9992,
+      "step": 7794
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3154558144290546e-05,
+      "loss": 0.2254,
+      "step": 7795
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3142409167184294e-05,
+      "loss": 0.2362,
+      "step": 7796
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.313026131287184e-05,
+      "loss": 0.917,
+      "step": 7797
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.31181145821623e-05,
+      "loss": 0.1927,
+      "step": 7798
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.31059689758647e-05,
+      "loss": 18.7698,
+      "step": 7799
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.309382449478797e-05,
+      "loss": 18.9378,
+      "step": 7800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3081681139741016e-05,
+      "loss": 0.9654,
+      "step": 7801
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.306953891153265e-05,
+      "loss": 20.8017,
+      "step": 7802
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.305739781097157e-05,
+      "loss": 19.9775,
+      "step": 7803
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.304525783886645e-05,
+      "loss": 20.02,
+      "step": 7804
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.303311899602587e-05,
+      "loss": 18.9565,
+      "step": 7805
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.302098128325831e-05,
+      "loss": 0.8495,
+      "step": 7806
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3008844701372224e-05,
+      "loss": 18.7,
+      "step": 7807
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.299670925117596e-05,
+      "loss": 0.1915,
+      "step": 7808
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.298457493347779e-05,
+      "loss": 21.1892,
+      "step": 7809
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2972441749085894e-05,
+      "loss": 19.2967,
+      "step": 7810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.296030969880846e-05,
+      "loss": 18.5797,
+      "step": 7811
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.294817878345348e-05,
+      "loss": 0.9506,
+      "step": 7812
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2936049003828926e-05,
+      "loss": 0.1943,
+      "step": 7813
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2923920360742774e-05,
+      "loss": 20.7801,
+      "step": 7814
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.291179285500275e-05,
+      "loss": 20.0282,
+      "step": 7815
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.289966648741666e-05,
+      "loss": 20.6018,
+      "step": 7816
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.288754125879215e-05,
+      "loss": 0.9395,
+      "step": 7817
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.287541716993686e-05,
+      "loss": 0.1877,
+      "step": 7818
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.286329422165827e-05,
+      "loss": 20.1106,
+      "step": 7819
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.285117241476382e-05,
+      "loss": 0.9192,
+      "step": 7820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.283905175006091e-05,
+      "loss": 19.5645,
+      "step": 7821
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.282693222835681e-05,
+      "loss": 17.4928,
+      "step": 7822
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.281481385045874e-05,
+      "loss": 0.9206,
+      "step": 7823
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.280269661717385e-05,
+      "loss": 19.9827,
+      "step": 7824
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.279058052930919e-05,
+      "loss": 19.4061,
+      "step": 7825
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.277846558767176e-05,
+      "loss": 19.7272,
+      "step": 7826
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2766351793068476e-05,
+      "loss": 0.9339,
+      "step": 7827
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.275423914630613e-05,
+      "loss": 0.205,
+      "step": 7828
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.274212764819154e-05,
+      "loss": 0.96,
+      "step": 7829
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.273001729953136e-05,
+      "loss": 20.7075,
+      "step": 7830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.271790810113219e-05,
+      "loss": 0.9597,
+      "step": 7831
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.270580005380056e-05,
+      "loss": 19.6865,
+      "step": 7832
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2693693158342946e-05,
+      "loss": 17.2582,
+      "step": 7833
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2681587415565694e-05,
+      "loss": 0.1973,
+      "step": 7834
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.26694828262751e-05,
+      "loss": 19.3405,
+      "step": 7835
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.265737939127742e-05,
+      "loss": 0.2159,
+      "step": 7836
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.264527711137878e-05,
+      "loss": 0.221,
+      "step": 7837
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.263317598738524e-05,
+      "loss": 0.2261,
+      "step": 7838
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.262107602010278e-05,
+      "loss": 0.2115,
+      "step": 7839
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2608977210337366e-05,
+      "loss": 20.3985,
+      "step": 7840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.259687955889478e-05,
+      "loss": 21.4779,
+      "step": 7841
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2584783066580795e-05,
+      "loss": 20.0831,
+      "step": 7842
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.257268773420112e-05,
+      "loss": 20.8653,
+      "step": 7843
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.256059356256134e-05,
+      "loss": 20.1251,
+      "step": 7844
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.254850055246696e-05,
+      "loss": 18.038,
+      "step": 7845
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2536408704723495e-05,
+      "loss": 18.1827,
+      "step": 7846
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.252431802013624e-05,
+      "loss": 0.9073,
+      "step": 7847
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.251222849951054e-05,
+      "loss": 0.9388,
+      "step": 7848
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.250014014365162e-05,
+      "loss": 0.9099,
+      "step": 7849
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.248805295336458e-05,
+      "loss": 0.927,
+      "step": 7850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2475966929454504e-05,
+      "loss": 0.1797,
+      "step": 7851
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.24638820727264e-05,
+      "loss": 19.226,
+      "step": 7852
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2451798383985135e-05,
+      "loss": 0.9198,
+      "step": 7853
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.243971586403556e-05,
+      "loss": 0.9175,
+      "step": 7854
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.242763451368244e-05,
+      "loss": 16.7973,
+      "step": 7855
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.241555433373042e-05,
+      "loss": 0.9241,
+      "step": 7856
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.240347532498411e-05,
+      "loss": 19.2391,
+      "step": 7857
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2391397488248034e-05,
+      "loss": 18.208,
+      "step": 7858
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.237932082432663e-05,
+      "loss": 20.3877,
+      "step": 7859
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2367245334024245e-05,
+      "loss": 19.0787,
+      "step": 7860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.235517101814518e-05,
+      "loss": 0.9105,
+      "step": 7861
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.234309787749364e-05,
+      "loss": 21.0705,
+      "step": 7862
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2331025912873734e-05,
+      "loss": 0.1977,
+      "step": 7863
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2318955125089524e-05,
+      "loss": 19.7312,
+      "step": 7864
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.230688551494499e-05,
+      "loss": 20.1752,
+      "step": 7865
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2294817083244014e-05,
+      "loss": 19.2987,
+      "step": 7866
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2282749830790395e-05,
+      "loss": 0.2084,
+      "step": 7867
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2270683758387896e-05,
+      "loss": 20.8877,
+      "step": 7868
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.225861886684016e-05,
+      "loss": 18.5848,
+      "step": 7869
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.224655515695076e-05,
+      "loss": 0.1908,
+      "step": 7870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2234492629523203e-05,
+      "loss": 0.9205,
+      "step": 7871
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.22224312853609e-05,
+      "loss": 22.1354,
+      "step": 7872
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.22103711252672e-05,
+      "loss": 19.5355,
+      "step": 7873
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2198312150045374e-05,
+      "loss": 0.9276,
+      "step": 7874
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.218625436049858e-05,
+      "loss": 0.2031,
+      "step": 7875
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2174197757429946e-05,
+      "loss": 0.9561,
+      "step": 7876
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2162142341642496e-05,
+      "loss": 19.8526,
+      "step": 7877
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.215008811393917e-05,
+      "loss": 20.0384,
+      "step": 7878
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2138035075122815e-05,
+      "loss": 19.84,
+      "step": 7879
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.212598322599627e-05,
+      "loss": 1.0082,
+      "step": 7880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.211393256736221e-05,
+      "loss": 0.2349,
+      "step": 7881
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.210188310002325e-05,
+      "loss": 0.1831,
+      "step": 7882
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.208983482478198e-05,
+      "loss": 0.2015,
+      "step": 7883
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.207778774244085e-05,
+      "loss": 0.2177,
+      "step": 7884
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.206574185380224e-05,
+      "loss": 0.2274,
+      "step": 7885
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.205369715966848e-05,
+      "loss": 20.3665,
+      "step": 7886
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.204165366084182e-05,
+      "loss": 19.7329,
+      "step": 7887
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.202961135812437e-05,
+      "loss": 0.1909,
+      "step": 7888
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.201757025231821e-05,
+      "loss": 0.9545,
+      "step": 7889
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.200553034422537e-05,
+      "loss": 18.7921,
+      "step": 7890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1993491634647737e-05,
+      "loss": 19.2377,
+      "step": 7891
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.198145412438713e-05,
+      "loss": 0.9491,
+      "step": 7892
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1969417814245365e-05,
+      "loss": 21.1848,
+      "step": 7893
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1957382705024036e-05,
+      "loss": 0.9292,
+      "step": 7894
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1945348797524785e-05,
+      "loss": 19.1176,
+      "step": 7895
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.193331609254913e-05,
+      "loss": 0.9506,
+      "step": 7896
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1921284590898456e-05,
+      "loss": 20.0148,
+      "step": 7897
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.190925429337418e-05,
+      "loss": 18.599,
+      "step": 7898
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1897225200777544e-05,
+      "loss": 0.1543,
+      "step": 7899
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.188519731390973e-05,
+      "loss": 0.1917,
+      "step": 7900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.187317063357187e-05,
+      "loss": 18.7209,
+      "step": 7901
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1861145160565e-05,
+      "loss": 0.9065,
+      "step": 7902
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.184912089569008e-05,
+      "loss": 20.1185,
+      "step": 7903
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1837097839747934e-05,
+      "loss": 20.3616,
+      "step": 7904
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.18250759935394e-05,
+      "loss": 18.7846,
+      "step": 7905
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1813055357865185e-05,
+      "loss": 19.6202,
+      "step": 7906
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.180103593352588e-05,
+      "loss": 0.1814,
+      "step": 7907
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.178901772132209e-05,
+      "loss": 19.3584,
+      "step": 7908
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.177700072205425e-05,
+      "loss": 17.5015,
+      "step": 7909
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.176498493652276e-05,
+      "loss": 20.2203,
+      "step": 7910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1752970365527904e-05,
+      "loss": 19.5452,
+      "step": 7911
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.174095700986994e-05,
+      "loss": 21.5829,
+      "step": 7912
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1728944870349e-05,
+      "loss": 0.9493,
+      "step": 7913
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.171693394776514e-05,
+      "loss": 0.1735,
+      "step": 7914
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.170492424291836e-05,
+      "loss": 0.2223,
+      "step": 7915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.169291575660855e-05,
+      "loss": 20.0451,
+      "step": 7916
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.168090848963552e-05,
+      "loss": 17.7837,
+      "step": 7917
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.166890244279905e-05,
+      "loss": 0.9719,
+      "step": 7918
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.165689761689875e-05,
+      "loss": 20.2307,
+      "step": 7919
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.164489401273423e-05,
+      "loss": 0.9317,
+      "step": 7920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.163289163110498e-05,
+      "loss": 0.2178,
+      "step": 7921
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1620890472810396e-05,
+      "loss": 20.1955,
+      "step": 7922
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.160889053864983e-05,
+      "loss": 19.1942,
+      "step": 7923
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1596891829422507e-05,
+      "loss": 19.6323,
+      "step": 7924
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.158489434592766e-05,
+      "loss": 18.2857,
+      "step": 7925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1572898088964296e-05,
+      "loss": 19.1225,
+      "step": 7926
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.156090305933148e-05,
+      "loss": 20.1873,
+      "step": 7927
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.154890925782812e-05,
+      "loss": 0.8665,
+      "step": 7928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.153691668525304e-05,
+      "loss": 0.9146,
+      "step": 7929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.152492534240502e-05,
+      "loss": 0.9767,
+      "step": 7930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.151293523008275e-05,
+      "loss": 0.9476,
+      "step": 7931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.150094634908478e-05,
+      "loss": 0.9712,
+      "step": 7932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.148895870020968e-05,
+      "loss": 0.1988,
+      "step": 7933
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.147697228425587e-05,
+      "loss": 18.7024,
+      "step": 7934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.146498710202168e-05,
+      "loss": 18.7317,
+      "step": 7935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.145300315430538e-05,
+      "loss": 21.3028,
+      "step": 7936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.144102044190519e-05,
+      "loss": 20.0364,
+      "step": 7937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.142903896561917e-05,
+      "loss": 17.8981,
+      "step": 7938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1417058726245364e-05,
+      "loss": 21.3281,
+      "step": 7939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.140507972458173e-05,
+      "loss": 1.0038,
+      "step": 7940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1393101961426094e-05,
+      "loss": 0.9646,
+      "step": 7941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.138112543757624e-05,
+      "loss": 20.0419,
+      "step": 7942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.136915015382985e-05,
+      "loss": 20.7956,
+      "step": 7943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.135717611098458e-05,
+      "loss": 20.232,
+      "step": 7944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1345203309837904e-05,
+      "loss": 0.937,
+      "step": 7945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.133323175118727e-05,
+      "loss": 18.5579,
+      "step": 7946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.132126143583007e-05,
+      "loss": 16.4633,
+      "step": 7947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.130929236456357e-05,
+      "loss": 20.5594,
+      "step": 7948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.129732453818495e-05,
+      "loss": 0.1509,
+      "step": 7949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.128535795749137e-05,
+      "loss": 19.5797,
+      "step": 7950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1273392623279785e-05,
+      "loss": 18.2345,
+      "step": 7951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.126142853634719e-05,
+      "loss": 0.9205,
+      "step": 7952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1249465697490457e-05,
+      "loss": 19.4079,
+      "step": 7953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1237504107506334e-05,
+      "loss": 17.6085,
+      "step": 7954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.122554376719153e-05,
+      "loss": 0.9461,
+      "step": 7955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.121358467734269e-05,
+      "loss": 20.2074,
+      "step": 7956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.120162683875629e-05,
+      "loss": 0.9013,
+      "step": 7957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.118967025222882e-05,
+      "loss": 19.3192,
+      "step": 7958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.117771491855664e-05,
+      "loss": 18.7267,
+      "step": 7959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.116576083853602e-05,
+      "loss": 19.9116,
+      "step": 7960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1153808012963136e-05,
+      "loss": 18.5105,
+      "step": 7961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.114185644263415e-05,
+      "loss": 0.2202,
+      "step": 7962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.112990612834505e-05,
+      "loss": 18.7379,
+      "step": 7963
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.11179570708918e-05,
+      "loss": 19.124,
+      "step": 7964
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1106009271070257e-05,
+      "loss": 0.8861,
+      "step": 7965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.109406272967623e-05,
+      "loss": 0.9122,
+      "step": 7966
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1082117447505367e-05,
+      "loss": 19.568,
+      "step": 7967
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1070173425353294e-05,
+      "loss": 0.1902,
+      "step": 7968
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1058230664015564e-05,
+      "loss": 0.9575,
+      "step": 7969
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.104628916428759e-05,
+      "loss": 19.7426,
+      "step": 7970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.103434892696474e-05,
+      "loss": 0.1681,
+      "step": 7971
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1022409952842305e-05,
+      "loss": 19.5918,
+      "step": 7972
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.101047224271545e-05,
+      "loss": 0.9284,
+      "step": 7973
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.09985357973793e-05,
+      "loss": 0.898,
+      "step": 7974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.098660061762888e-05,
+      "loss": 19.4755,
+      "step": 7975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.09746667042591e-05,
+      "loss": 19.2991,
+      "step": 7976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0962734058064854e-05,
+      "loss": 19.3183,
+      "step": 7977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0950802679840896e-05,
+      "loss": 0.1895,
+      "step": 7978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.093887257038189e-05,
+      "loss": 20.9942,
+      "step": 7979
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.092694373048246e-05,
+      "loss": 18.9686,
+      "step": 7980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.091501616093713e-05,
+      "loss": 17.545,
+      "step": 7981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.090308986254031e-05,
+      "loss": 0.1925,
+      "step": 7982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.089116483608634e-05,
+      "loss": 0.212,
+      "step": 7983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.087924108236952e-05,
+      "loss": 0.9243,
+      "step": 7984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.086731860218399e-05,
+      "loss": 20.9061,
+      "step": 7985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.085539739632385e-05,
+      "loss": 18.9226,
+      "step": 7986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0843477465583126e-05,
+      "loss": 17.4682,
+      "step": 7987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.083155881075574e-05,
+      "loss": 20.7689,
+      "step": 7988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.081964143263549e-05,
+      "loss": 18.2928,
+      "step": 7989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.080772533201618e-05,
+      "loss": 0.1975,
+      "step": 7990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.079581050969146e-05,
+      "loss": 19.953,
+      "step": 7991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0783896966454893e-05,
+      "loss": 19.5529,
+      "step": 7992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0771984703099985e-05,
+      "loss": 17.8544,
+      "step": 7993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.076007372042018e-05,
+      "loss": 0.906,
+      "step": 7994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.074816401920876e-05,
+      "loss": 0.9402,
+      "step": 7995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.073625560025898e-05,
+      "loss": 0.9251,
+      "step": 7996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.072434846436403e-05,
+      "loss": 0.9646,
+      "step": 7997
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.071244261231693e-05,
+      "loss": 19.3714,
+      "step": 7998
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.070053804491068e-05,
+      "loss": 20.204,
+      "step": 7999
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0688634762938204e-05,
+      "loss": 17.7804,
+      "step": 8000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0676732767192284e-05,
+      "loss": 20.3597,
+      "step": 8001
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.066483205846566e-05,
+      "loss": 18.0435,
+      "step": 8002
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0652932637551e-05,
+      "loss": 21.7418,
+      "step": 8003
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.064103450524081e-05,
+      "loss": 19.211,
+      "step": 8004
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0629137662327604e-05,
+      "loss": 0.9352,
+      "step": 8005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.061724210960375e-05,
+      "loss": 0.9228,
+      "step": 8006
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.060534784786154e-05,
+      "loss": 22.0909,
+      "step": 8007
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.059345487789319e-05,
+      "loss": 0.9501,
+      "step": 8008
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0581563200490846e-05,
+      "loss": 20.8265,
+      "step": 8009
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0569672816446537e-05,
+      "loss": 20.0565,
+      "step": 8010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.05577837265522e-05,
+      "loss": 18.3687,
+      "step": 8011
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0545895931599736e-05,
+      "loss": 0.9387,
+      "step": 8012
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0534009432380904e-05,
+      "loss": 0.9926,
+      "step": 8013
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.052212422968741e-05,
+      "loss": 19.9295,
+      "step": 8014
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0510240324310847e-05,
+      "loss": 21.965,
+      "step": 8015
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.049835771704277e-05,
+      "loss": 0.9583,
+      "step": 8016
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.048647640867459e-05,
+      "loss": 18.3531,
+      "step": 8017
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0474596399997655e-05,
+      "loss": 0.8832,
+      "step": 8018
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0462717691803262e-05,
+      "loss": 0.1921,
+      "step": 8019
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0450840284882546e-05,
+      "loss": 18.5779,
+      "step": 8020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0438964180026613e-05,
+      "loss": 0.9691,
+      "step": 8021
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0427089378026497e-05,
+      "loss": 19.7004,
+      "step": 8022
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0415215879673058e-05,
+      "loss": 0.1936,
+      "step": 8023
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0403343685757168e-05,
+      "loss": 20.2398,
+      "step": 8024
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0391472797069566e-05,
+      "loss": 0.1704,
+      "step": 8025
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0379603214400883e-05,
+      "loss": 0.9683,
+      "step": 8026
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0367734938541715e-05,
+      "loss": 0.2129,
+      "step": 8027
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0355867970282514e-05,
+      "loss": 0.9476,
+      "step": 8028
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0344002310413745e-05,
+      "loss": 18.9868,
+      "step": 8029
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0332137959725628e-05,
+      "loss": 18.1843,
+      "step": 8030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0320274919008427e-05,
+      "loss": 0.1809,
+      "step": 8031
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0308413189052287e-05,
+      "loss": 0.221,
+      "step": 8032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0296552770647224e-05,
+      "loss": 19.94,
+      "step": 8033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0284693664583215e-05,
+      "loss": 19.6306,
+      "step": 8034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.027283587165014e-05,
+      "loss": 0.2362,
+      "step": 8035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.026097939263775e-05,
+      "loss": 20.0756,
+      "step": 8036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0249124228335773e-05,
+      "loss": 0.2177,
+      "step": 8037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.023727037953382e-05,
+      "loss": 0.2151,
+      "step": 8038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0225417847021394e-05,
+      "loss": 0.9384,
+      "step": 8039
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.021356663158792e-05,
+      "loss": 19.6234,
+      "step": 8040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0201716734022777e-05,
+      "loss": 19.0541,
+      "step": 8041
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.01898681551152e-05,
+      "loss": 0.216,
+      "step": 8042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0178020895654345e-05,
+      "loss": 18.6692,
+      "step": 8043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0166174956429348e-05,
+      "loss": 0.2143,
+      "step": 8044
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0154330338229152e-05,
+      "loss": 20.4744,
+      "step": 8045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.014248704184268e-05,
+      "loss": 19.4555,
+      "step": 8046
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.013064506805876e-05,
+      "loss": 0.9289,
+      "step": 8047
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0118804417666103e-05,
+      "loss": 17.8638,
+      "step": 8048
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0106965091453364e-05,
+      "loss": 0.9951,
+      "step": 8049
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0095127090209087e-05,
+      "loss": 19.8387,
+      "step": 8050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0083290414721764e-05,
+      "loss": 20.863,
+      "step": 8051
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0071455065779747e-05,
+      "loss": 0.9212,
+      "step": 8052
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0059621044171325e-05,
+      "loss": 0.184,
+      "step": 8053
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0047788350684736e-05,
+      "loss": 18.6841,
+      "step": 8054
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0035956986108027e-05,
+      "loss": 0.9314,
+      "step": 8055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.002412695122927e-05,
+      "loss": 0.9704,
+      "step": 8056
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.00122982468364e-05,
+      "loss": 20.8845,
+      "step": 8057
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0000470873717236e-05,
+      "loss": 0.1981,
+      "step": 8058
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9988644832659557e-05,
+      "loss": 0.9565,
+      "step": 8059
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.997682012445104e-05,
+      "loss": 19.0149,
+      "step": 8060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9964996749879238e-05,
+      "loss": 0.1936,
+      "step": 8061
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9953174709731646e-05,
+      "loss": 0.9545,
+      "step": 8062
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9941354004795703e-05,
+      "loss": 0.9282,
+      "step": 8063
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.992953463585868e-05,
+      "loss": 19.5931,
+      "step": 8064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.991771660370782e-05,
+      "loss": 18.4561,
+      "step": 8065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.990589990913027e-05,
+      "loss": 0.201,
+      "step": 8066
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.989408455291306e-05,
+      "loss": 19.6379,
+      "step": 8067
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9882270535843137e-05,
+      "loss": 0.9182,
+      "step": 8068
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9870457858707413e-05,
+      "loss": 0.9689,
+      "step": 8069
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9858646522292616e-05,
+      "loss": 0.2211,
+      "step": 8070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9846836527385463e-05,
+      "loss": 18.3208,
+      "step": 8071
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9835027874772546e-05,
+      "loss": 0.215,
+      "step": 8072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9823220565240394e-05,
+      "loss": 0.9499,
+      "step": 8073
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.981141459957541e-05,
+      "loss": 19.5594,
+      "step": 8074
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.979960997856392e-05,
+      "loss": 0.9522,
+      "step": 8075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9787806702992195e-05,
+      "loss": 0.9433,
+      "step": 8076
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.977600477364636e-05,
+      "loss": 0.9618,
+      "step": 8077
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9764204191312494e-05,
+      "loss": 0.2058,
+      "step": 8078
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.975240495677657e-05,
+      "loss": 20.0869,
+      "step": 8079
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.974060707082445e-05,
+      "loss": 19.0066,
+      "step": 8080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.972881053424196e-05,
+      "loss": 17.1078,
+      "step": 8081
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9717015347814792e-05,
+      "loss": 0.9532,
+      "step": 8082
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.970522151232854e-05,
+      "loss": 17.5218,
+      "step": 8083
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9693429028568753e-05,
+      "loss": 0.9251,
+      "step": 8084
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9681637897320868e-05,
+      "loss": 17.7186,
+      "step": 8085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9669848119370215e-05,
+      "loss": 0.8834,
+      "step": 8086
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9658059695502034e-05,
+      "loss": 17.9003,
+      "step": 8087
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.964627262650153e-05,
+      "loss": 18.5434,
+      "step": 8088
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9634486913153746e-05,
+      "loss": 0.8913,
+      "step": 8089
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9622702556243652e-05,
+      "loss": 0.9172,
+      "step": 8090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9610919556556195e-05,
+      "loss": 20.4486,
+      "step": 8091
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9599137914876123e-05,
+      "loss": 20.2629,
+      "step": 8092
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9587357631988165e-05,
+      "loss": 20.2829,
+      "step": 8093
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9575578708676954e-05,
+      "loss": 0.9578,
+      "step": 8094
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.956380114572702e-05,
+      "loss": 19.0691,
+      "step": 8095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9552024943922796e-05,
+      "loss": 20.9407,
+      "step": 8096
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.954025010404861e-05,
+      "loss": 20.474,
+      "step": 8097
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9528476626888767e-05,
+      "loss": 19.3313,
+      "step": 8098
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9516704513227413e-05,
+      "loss": 20.4935,
+      "step": 8099
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9504933763848596e-05,
+      "loss": 18.2796,
+      "step": 8100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9493164379536374e-05,
+      "loss": 17.7776,
+      "step": 8101
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9481396361074564e-05,
+      "loss": 19.4707,
+      "step": 8102
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9469629709247015e-05,
+      "loss": 19.9505,
+      "step": 8103
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9457864424837438e-05,
+      "loss": 21.8939,
+      "step": 8104
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9446100508629437e-05,
+      "loss": 0.1875,
+      "step": 8105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9434337961406554e-05,
+      "loss": 0.9361,
+      "step": 8106
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.942257678395225e-05,
+      "loss": 20.5494,
+      "step": 8107
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.941081697704984e-05,
+      "loss": 18.6997,
+      "step": 8108
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9399058541482593e-05,
+      "loss": 0.8937,
+      "step": 8109
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.938730147803369e-05,
+      "loss": 0.9172,
+      "step": 8110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9375545787486192e-05,
+      "loss": 0.2241,
+      "step": 8111
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.936379147062307e-05,
+      "loss": 0.1794,
+      "step": 8112
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9352038528227242e-05,
+      "loss": 17.9215,
+      "step": 8113
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9340286961081515e-05,
+      "loss": 20.2445,
+      "step": 8114
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9328536769968552e-05,
+      "loss": 19.2855,
+      "step": 8115
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9316787955671016e-05,
+      "loss": 0.9271,
+      "step": 8116
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9305040518971423e-05,
+      "loss": 21.9319,
+      "step": 8117
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9293294460652187e-05,
+      "loss": 19.9494,
+      "step": 8118
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9281549781495655e-05,
+      "loss": 0.9057,
+      "step": 8119
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9269806482284103e-05,
+      "loss": 0.9128,
+      "step": 8120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.925806456379966e-05,
+      "loss": 0.9276,
+      "step": 8121
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.924632402682439e-05,
+      "loss": 0.9172,
+      "step": 8122
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9234584872140307e-05,
+      "loss": 0.1561,
+      "step": 8123
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.922284710052926e-05,
+      "loss": 20.37,
+      "step": 8124
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9211110712773027e-05,
+      "loss": 18.9139,
+      "step": 8125
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9199375709653353e-05,
+      "loss": 0.9434,
+      "step": 8126
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9187642091951806e-05,
+      "loss": 18.6516,
+      "step": 8127
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.917590986044988e-05,
+      "loss": 19.1382,
+      "step": 8128
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.916417901592907e-05,
+      "loss": 17.925,
+      "step": 8129
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9152449559170636e-05,
+      "loss": 18.5698,
+      "step": 8130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9140721490955815e-05,
+      "loss": 0.9091,
+      "step": 8131
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.912899481206582e-05,
+      "loss": 16.8841,
+      "step": 8132
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9117269523281632e-05,
+      "loss": 0.9331,
+      "step": 8133
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9105545625384233e-05,
+      "loss": 0.1779,
+      "step": 8134
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.909382311915447e-05,
+      "loss": 0.9658,
+      "step": 8135
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.908210200537318e-05,
+      "loss": 20.5386,
+      "step": 8136
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9070382284820974e-05,
+      "loss": 0.9061,
+      "step": 8137
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9058663958278442e-05,
+      "loss": 0.1892,
+      "step": 8138
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.904694702652615e-05,
+      "loss": 17.5516,
+      "step": 8139
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9035231490344426e-05,
+      "loss": 18.032,
+      "step": 8140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9023517350513595e-05,
+      "loss": 0.1932,
+      "step": 8141
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9011804607813912e-05,
+      "loss": 0.9256,
+      "step": 8142
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9000093263025456e-05,
+      "loss": 0.9755,
+      "step": 8143
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8988383316928275e-05,
+      "loss": 19.4934,
+      "step": 8144
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8976674770302305e-05,
+      "loss": 19.4277,
+      "step": 8145
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8964967623927374e-05,
+      "loss": 17.8662,
+      "step": 8146
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.895326187858326e-05,
+      "loss": 19.2611,
+      "step": 8147
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.89415575350496e-05,
+      "loss": 19.3367,
+      "step": 8148
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8929854594105965e-05,
+      "loss": 19.2189,
+      "step": 8149
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.891815305653182e-05,
+      "loss": 18.9187,
+      "step": 8150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8906452923106552e-05,
+      "loss": 0.175,
+      "step": 8151
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.889475419460943e-05,
+      "loss": 17.6875,
+      "step": 8152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.888305687181966e-05,
+      "loss": 19.4188,
+      "step": 8153
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8871360955516326e-05,
+      "loss": 0.2038,
+      "step": 8154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8859666446478427e-05,
+      "loss": 18.6638,
+      "step": 8155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.884797334548488e-05,
+      "loss": 0.1973,
+      "step": 8156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.88362816533145e-05,
+      "loss": 0.8997,
+      "step": 8157
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8824591370746013e-05,
+      "loss": 0.9546,
+      "step": 8158
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.881290249855803e-05,
+      "loss": 18.0617,
+      "step": 8159
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8801215037529095e-05,
+      "loss": 0.9607,
+      "step": 8160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8789528988437653e-05,
+      "loss": 18.3477,
+      "step": 8161
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8777844352062032e-05,
+      "loss": 19.3208,
+      "step": 8162
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8766161129180492e-05,
+      "loss": 0.9936,
+      "step": 8163
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8754479320571193e-05,
+      "loss": 0.1765,
+      "step": 8164
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.87427989270122e-05,
+      "loss": 0.2021,
+      "step": 8165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.873111994928147e-05,
+      "loss": 19.2506,
+      "step": 8166
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8719442388156903e-05,
+      "loss": 0.2067,
+      "step": 8167
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8707766244416222e-05,
+      "loss": 19.6967,
+      "step": 8168
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8696091518837177e-05,
+      "loss": 19.4477,
+      "step": 8169
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.868441821219734e-05,
+      "loss": 0.932,
+      "step": 8170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.867274632527417e-05,
+      "loss": 18.189,
+      "step": 8171
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.866107585884511e-05,
+      "loss": 20.2196,
+      "step": 8172
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8649406813687473e-05,
+      "loss": 0.9557,
+      "step": 8173
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8637739190578437e-05,
+      "loss": 17.6814,
+      "step": 8174
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8626072990295116e-05,
+      "loss": 19.9673,
+      "step": 8175
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8614408213614598e-05,
+      "loss": 19.5272,
+      "step": 8176
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8602744861313747e-05,
+      "loss": 0.9436,
+      "step": 8177
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8591082934169387e-05,
+      "loss": 0.9693,
+      "step": 8178
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8579422432958312e-05,
+      "loss": 0.9543,
+      "step": 8179
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.856776335845716e-05,
+      "loss": 20.5423,
+      "step": 8180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8556105711442437e-05,
+      "loss": 21.0917,
+      "step": 8181
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.85444494926906e-05,
+      "loss": 19.8794,
+      "step": 8182
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8532794702978067e-05,
+      "loss": 20.7553,
+      "step": 8183
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.852114134308104e-05,
+      "loss": 0.2069,
+      "step": 8184
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.850948941377568e-05,
+      "loss": 0.9532,
+      "step": 8185
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8497838915838132e-05,
+      "loss": 0.9469,
+      "step": 8186
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.848618985004431e-05,
+      "loss": 18.4031,
+      "step": 8187
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8474542217170088e-05,
+      "loss": 20.4871,
+      "step": 8188
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8462896017991312e-05,
+      "loss": 1.0003,
+      "step": 8189
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.845125125328363e-05,
+      "loss": 0.9292,
+      "step": 8190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8439607923822637e-05,
+      "loss": 20.7888,
+      "step": 8191
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8427966030383844e-05,
+      "loss": 0.9356,
+      "step": 8192
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8416325573742652e-05,
+      "loss": 18.4181,
+      "step": 8193
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8404686554674377e-05,
+      "loss": 20.6552,
+      "step": 8194
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8393048973954216e-05,
+      "loss": 18.3451,
+      "step": 8195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.83814128323573e-05,
+      "loss": 19.9111,
+      "step": 8196
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8369778130658648e-05,
+      "loss": 19.0436,
+      "step": 8197
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.835814486963318e-05,
+      "loss": 20.2306,
+      "step": 8198
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8346513050055723e-05,
+      "loss": 20.6057,
+      "step": 8199
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.833488267270102e-05,
+      "loss": 20.5024,
+      "step": 8200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8323253738343704e-05,
+      "loss": 19.863,
+      "step": 8201
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8311626247758306e-05,
+      "loss": 19.4285,
+      "step": 8202
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8300000201719284e-05,
+      "loss": 0.8984,
+      "step": 8203
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8288375601000983e-05,
+      "loss": 18.6303,
+      "step": 8204
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.827675244637765e-05,
+      "loss": 0.9441,
+      "step": 8205
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8265130738623447e-05,
+      "loss": 0.2125,
+      "step": 8206
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8253510478512436e-05,
+      "loss": 18.651,
+      "step": 8207
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8241891666818577e-05,
+      "loss": 18.8326,
+      "step": 8208
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8230274304315728e-05,
+      "loss": 20.036,
+      "step": 8209
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8218658391777675e-05,
+      "loss": 19.4213,
+      "step": 8210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8207043929978083e-05,
+      "loss": 19.4605,
+      "step": 8211
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8195430919690535e-05,
+      "loss": 18.3938,
+      "step": 8212
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.81838193616885e-05,
+      "loss": 0.2073,
+      "step": 8213
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8172209256745386e-05,
+      "loss": 18.688,
+      "step": 8214
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8160600605634428e-05,
+      "loss": 0.9717,
+      "step": 8215
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8148993409128865e-05,
+      "loss": 19.9008,
+      "step": 8216
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.81373876680018e-05,
+      "loss": 0.9536,
+      "step": 8217
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8125783383026165e-05,
+      "loss": 0.2086,
+      "step": 8218
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8114180554974918e-05,
+      "loss": 20.2042,
+      "step": 8219
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8102579184620843e-05,
+      "loss": 0.2254,
+      "step": 8220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8090979272736662e-05,
+      "loss": 0.9127,
+      "step": 8221
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8079380820094924e-05,
+      "loss": 18.2294,
+      "step": 8222
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8067783827468207e-05,
+      "loss": 0.9259,
+      "step": 8223
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8056188295628916e-05,
+      "loss": 19.6867,
+      "step": 8224
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8044594225349307e-05,
+      "loss": 0.927,
+      "step": 8225
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.803300161740166e-05,
+      "loss": 19.5636,
+      "step": 8226
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8021410472558102e-05,
+      "loss": 0.2259,
+      "step": 8227
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.800982079159059e-05,
+      "loss": 0.9939,
+      "step": 8228
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.799823257527112e-05,
+      "loss": 19.2487,
+      "step": 8229
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.798664582437151e-05,
+      "loss": 20.0961,
+      "step": 8230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7975060539663457e-05,
+      "loss": 17.7642,
+      "step": 8231
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7963476721918597e-05,
+      "loss": 0.9843,
+      "step": 8232
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7951894371908517e-05,
+      "loss": 0.1861,
+      "step": 8233
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.794031349040461e-05,
+      "loss": 22.571,
+      "step": 8234
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.79287340781782e-05,
+      "loss": 0.9582,
+      "step": 8235
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.79171561360006e-05,
+      "loss": 0.1958,
+      "step": 8236
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.790557966464289e-05,
+      "loss": 0.9442,
+      "step": 8237
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.789400466487614e-05,
+      "loss": 19.3719,
+      "step": 8238
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.78824311374713e-05,
+      "loss": 0.2134,
+      "step": 8239
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7870859083199217e-05,
+      "loss": 20.7905,
+      "step": 8240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7859288502830643e-05,
+      "loss": 20.9804,
+      "step": 8241
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7847719397136218e-05,
+      "loss": 20.5461,
+      "step": 8242
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7836151766886546e-05,
+      "loss": 0.9671,
+      "step": 8243
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7824585612852037e-05,
+      "loss": 19.0493,
+      "step": 8244
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.781302093580305e-05,
+      "loss": 19.1922,
+      "step": 8245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7801457736509907e-05,
+      "loss": 19.4766,
+      "step": 8246
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7789896015742696e-05,
+      "loss": 18.5703,
+      "step": 8247
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7778335774271513e-05,
+      "loss": 0.9478,
+      "step": 8248
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.776677701286633e-05,
+      "loss": 20.0655,
+      "step": 8249
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7755219732297e-05,
+      "loss": 21.9199,
+      "step": 8250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7743663933333298e-05,
+      "loss": 20.3874,
+      "step": 8251
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.77321096167449e-05,
+      "loss": 18.3684,
+      "step": 8252
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7720556783301365e-05,
+      "loss": 19.7585,
+      "step": 8253
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7709005433772178e-05,
+      "loss": 19.6912,
+      "step": 8254
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.76974555689267e-05,
+      "loss": 21.5506,
+      "step": 8255
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7685907189534214e-05,
+      "loss": 0.9231,
+      "step": 8256
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.76743602963639e-05,
+      "loss": 20.159,
+      "step": 8257
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7662814890184818e-05,
+      "loss": 0.9326,
+      "step": 8258
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7651270971765957e-05,
+      "loss": 19.1426,
+      "step": 8259
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7639728541876198e-05,
+      "loss": 0.9432,
+      "step": 8260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7628187601284337e-05,
+      "loss": 0.1944,
+      "step": 8261
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.761664815075899e-05,
+      "loss": 20.4432,
+      "step": 8262
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7605110191068805e-05,
+      "loss": 19.9321,
+      "step": 8263
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7593573722982237e-05,
+      "loss": 0.9679,
+      "step": 8264
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.758203874726768e-05,
+      "loss": 0.2033,
+      "step": 8265
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.75705052646934e-05,
+      "loss": 0.9693,
+      "step": 8266
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7558973276027588e-05,
+      "loss": 19.8989,
+      "step": 8267
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7547442782038335e-05,
+      "loss": 0.9047,
+      "step": 8268
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7535913783493622e-05,
+      "loss": 19.7244,
+      "step": 8269
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.752438628116133e-05,
+      "loss": 16.3233,
+      "step": 8270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.751286027580926e-05,
+      "loss": 18.7916,
+      "step": 8271
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.750133576820505e-05,
+      "loss": 17.88,
+      "step": 8272
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.748981275911633e-05,
+      "loss": 20.0951,
+      "step": 8273
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7478291249310605e-05,
+      "loss": 0.9806,
+      "step": 8274
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7466771239555185e-05,
+      "loss": 0.1683,
+      "step": 8275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.745525273061742e-05,
+      "loss": 0.991,
+      "step": 8276
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7443735723264502e-05,
+      "loss": 0.9065,
+      "step": 8277
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7432220218263473e-05,
+      "loss": 19.2969,
+      "step": 8278
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.742070621638132e-05,
+      "loss": 17.0721,
+      "step": 8279
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7409193718384986e-05,
+      "loss": 18.7424,
+      "step": 8280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7397682725041196e-05,
+      "loss": 0.2245,
+      "step": 8281
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.738617323711664e-05,
+      "loss": 20.2318,
+      "step": 8282
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.737466525537797e-05,
+      "loss": 0.1864,
+      "step": 8283
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7363158780591596e-05,
+      "loss": 20.1736,
+      "step": 8284
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.735165381352393e-05,
+      "loss": 18.7996,
+      "step": 8285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.734015035494124e-05,
+      "loss": 19.5806,
+      "step": 8286
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.732864840560976e-05,
+      "loss": 18.53,
+      "step": 8287
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.731714796629552e-05,
+      "loss": 19.464,
+      "step": 8288
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7305649037764502e-05,
+      "loss": 19.2503,
+      "step": 8289
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7294151620782647e-05,
+      "loss": 18.2732,
+      "step": 8290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.728265571611568e-05,
+      "loss": 0.9247,
+      "step": 8291
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.727116132452928e-05,
+      "loss": 0.9149,
+      "step": 8292
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7259668446789078e-05,
+      "loss": 18.3275,
+      "step": 8293
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7248177083660507e-05,
+      "loss": 0.9234,
+      "step": 8294
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7236687235908953e-05,
+      "loss": 0.9723,
+      "step": 8295
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.72251989042997e-05,
+      "loss": 19.2128,
+      "step": 8296
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7213712089597925e-05,
+      "loss": 0.197,
+      "step": 8297
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7202226792568698e-05,
+      "loss": 19.4199,
+      "step": 8298
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7190743013977e-05,
+      "loss": 19.4457,
+      "step": 8299
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.71792607545877e-05,
+      "loss": 0.2285,
+      "step": 8300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7167780015165565e-05,
+      "loss": 0.9154,
+      "step": 8301
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7156300796475275e-05,
+      "loss": 0.9329,
+      "step": 8302
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.714482309928139e-05,
+      "loss": 0.2057,
+      "step": 8303
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7133346924348386e-05,
+      "loss": 19.4251,
+      "step": 8304
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7121872272440625e-05,
+      "loss": 19.8654,
+      "step": 8305
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7110399144322374e-05,
+      "loss": 19.1243,
+      "step": 8306
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7098927540757797e-05,
+      "loss": 18.614,
+      "step": 8307
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.708745746251095e-05,
+      "loss": 0.1953,
+      "step": 8308
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.70759889103458e-05,
+      "loss": 0.2147,
+      "step": 8309
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7064521885026206e-05,
+      "loss": 18.735,
+      "step": 8310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7053056387315922e-05,
+      "loss": 18.154,
+      "step": 8311
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7041592417978606e-05,
+      "loss": 19.5758,
+      "step": 8312
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.703012997777781e-05,
+      "loss": 0.2045,
+      "step": 8313
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7018669067476987e-05,
+      "loss": 0.932,
+      "step": 8314
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.700720968783948e-05,
+      "loss": 20.2315,
+      "step": 8315
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6995751839628547e-05,
+      "loss": 19.4989,
+      "step": 8316
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.698429552360733e-05,
+      "loss": 0.9545,
+      "step": 8317
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.697284074053889e-05,
+      "loss": 19.6993,
+      "step": 8318
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.696138749118611e-05,
+      "loss": 0.922,
+      "step": 8319
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6949935776311896e-05,
+      "loss": 20.0993,
+      "step": 8320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6938485596678975e-05,
+      "loss": 0.9108,
+      "step": 8321
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.692703695304993e-05,
+      "loss": 19.0917,
+      "step": 8322
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.691558984618735e-05,
+      "loss": 20.6778,
+      "step": 8323
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.690414427685366e-05,
+      "loss": 0.905,
+      "step": 8324
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6892700245811164e-05,
+      "loss": 0.9356,
+      "step": 8325
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6881257753822076e-05,
+      "loss": 19.0474,
+      "step": 8326
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6869816801648558e-05,
+      "loss": 18.7429,
+      "step": 8327
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6858377390052632e-05,
+      "loss": 0.215,
+      "step": 8328
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.684693951979616e-05,
+      "loss": 19.6148,
+      "step": 8329
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6835503191641014e-05,
+      "loss": 0.179,
+      "step": 8330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6824068406348914e-05,
+      "loss": 19.6959,
+      "step": 8331
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6812635164681386e-05,
+      "loss": 0.2092,
+      "step": 8332
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6801203467400033e-05,
+      "loss": 0.9501,
+      "step": 8333
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.678977331526623e-05,
+      "loss": 0.9146,
+      "step": 8334
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.677834470904125e-05,
+      "loss": 19.2976,
+      "step": 8335
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.676691764948629e-05,
+      "loss": 20.5317,
+      "step": 8336
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6755492137362503e-05,
+      "loss": 20.9509,
+      "step": 8337
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6744068173430825e-05,
+      "loss": 0.9116,
+      "step": 8338
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6732645758452148e-05,
+      "loss": 0.1626,
+      "step": 8339
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.67212248931873e-05,
+      "loss": 0.9716,
+      "step": 8340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.670980557839693e-05,
+      "loss": 19.0986,
+      "step": 8341
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.669838781484162e-05,
+      "loss": 21.0639,
+      "step": 8342
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.668697160328184e-05,
+      "loss": 0.1786,
+      "step": 8343
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6675556944477985e-05,
+      "loss": 19.5289,
+      "step": 8344
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6664143839190302e-05,
+      "loss": 0.9837,
+      "step": 8345
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6652732288178976e-05,
+      "loss": 18.3117,
+      "step": 8346
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6641322292204058e-05,
+      "loss": 0.186,
+      "step": 8347
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.66299138520255e-05,
+      "loss": 21.2872,
+      "step": 8348
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6618506968403156e-05,
+      "loss": 0.2073,
+      "step": 8349
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.660710164209682e-05,
+      "loss": 0.8977,
+      "step": 8350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6595697873866092e-05,
+      "loss": 20.2459,
+      "step": 8351
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6584295664470525e-05,
+      "loss": 0.9612,
+      "step": 8352
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6572895014669575e-05,
+      "loss": 19.6898,
+      "step": 8353
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6561495925222558e-05,
+      "loss": 0.9376,
+      "step": 8354
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6550098396888722e-05,
+      "loss": 18.3344,
+      "step": 8355
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6538702430427188e-05,
+      "loss": 18.8333,
+      "step": 8356
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6527308026596975e-05,
+      "loss": 0.9001,
+      "step": 8357
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6515915186157014e-05,
+      "loss": 0.917,
+      "step": 8358
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.650452390986612e-05,
+      "loss": 18.7049,
+      "step": 8359
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6493134198483006e-05,
+      "loss": 0.1582,
+      "step": 8360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6481746052766266e-05,
+      "loss": 0.9474,
+      "step": 8361
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6470359473474417e-05,
+      "loss": 0.9166,
+      "step": 8362
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6458974461365854e-05,
+      "loss": 0.9326,
+      "step": 8363
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6447591017198868e-05,
+      "loss": 20.788,
+      "step": 8364
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6436209141731676e-05,
+      "loss": 18.8277,
+      "step": 8365
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6424828835722304e-05,
+      "loss": 19.6032,
+      "step": 8366
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6413450099928783e-05,
+      "loss": 0.9496,
+      "step": 8367
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6402072935109e-05,
+      "loss": 0.2102,
+      "step": 8368
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6390697342020665e-05,
+      "loss": 20.1405,
+      "step": 8369
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.63793233214215e-05,
+      "loss": 0.9429,
+      "step": 8370
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.636795087406906e-05,
+      "loss": 0.9309,
+      "step": 8371
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.635658000072078e-05,
+      "loss": 0.9879,
+      "step": 8372
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6345210702134037e-05,
+      "loss": 19.9705,
+      "step": 8373
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.633384297906607e-05,
+      "loss": 18.8629,
+      "step": 8374
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6322476832274034e-05,
+      "loss": 18.0655,
+      "step": 8375
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6311112262514913e-05,
+      "loss": 21.5314,
+      "step": 8376
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6299749270545714e-05,
+      "loss": 0.9328,
+      "step": 8377
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6288387857123242e-05,
+      "loss": 20.2285,
+      "step": 8378
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6277028023004167e-05,
+      "loss": 0.9323,
+      "step": 8379
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6265669768945177e-05,
+      "loss": 0.9018,
+      "step": 8380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6254313095702782e-05,
+      "loss": 18.3721,
+      "step": 8381
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6242958004033335e-05,
+      "loss": 17.739,
+      "step": 8382
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6231604494693153e-05,
+      "loss": 18.5924,
+      "step": 8383
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6220252568438487e-05,
+      "loss": 0.9063,
+      "step": 8384
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6208902226025366e-05,
+      "loss": 0.229,
+      "step": 8385
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.619755346820978e-05,
+      "loss": 20.7906,
+      "step": 8386
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6186206295747662e-05,
+      "loss": 18.9165,
+      "step": 8387
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.617486070939473e-05,
+      "loss": 18.6093,
+      "step": 8388
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6163516709906677e-05,
+      "loss": 18.949,
+      "step": 8389
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6152174298039045e-05,
+      "loss": 0.9153,
+      "step": 8390
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6140833474547345e-05,
+      "loss": 19.0979,
+      "step": 8391
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.612949424018687e-05,
+      "loss": 18.1808,
+      "step": 8392
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6118156595712874e-05,
+      "loss": 0.9694,
+      "step": 8393
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.610682054188054e-05,
+      "loss": 0.188,
+      "step": 8394
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6095486079444863e-05,
+      "loss": 0.9722,
+      "step": 8395
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6084153209160754e-05,
+      "loss": 19.3036,
+      "step": 8396
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6072821931783103e-05,
+      "loss": 0.1859,
+      "step": 8397
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6061492248066564e-05,
+      "loss": 0.178,
+      "step": 8398
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6050164158765767e-05,
+      "loss": 19.4447,
+      "step": 8399
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6038837664635207e-05,
+      "loss": 0.9861,
+      "step": 8400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6027512766429296e-05,
+      "loss": 19.6722,
+      "step": 8401
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6016189464902317e-05,
+      "loss": 19.1723,
+      "step": 8402
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.600486776080846e-05,
+      "loss": 17.9638,
+      "step": 8403
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5993547654901785e-05,
+      "loss": 20.3763,
+      "step": 8404
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.598222914793629e-05,
+      "loss": 20.1968,
+      "step": 8405
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5970912240665813e-05,
+      "loss": 0.927,
+      "step": 8406
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.595959693384413e-05,
+      "loss": 21.4487,
+      "step": 8407
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.59482832282249e-05,
+      "loss": 0.2126,
+      "step": 8408
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.593697112456165e-05,
+      "loss": 19.4868,
+      "step": 8409
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5925660623607823e-05,
+      "loss": 19.1999,
+      "step": 8410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5914351726116758e-05,
+      "loss": 20.8104,
+      "step": 8411
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5903044432841672e-05,
+      "loss": 18.2461,
+      "step": 8412
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5891738744535688e-05,
+      "loss": 18.6029,
+      "step": 8413
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5880434661951823e-05,
+      "loss": 18.3762,
+      "step": 8414
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.586913218584297e-05,
+      "loss": 0.2161,
+      "step": 8415
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5857831316961934e-05,
+      "loss": 19.8379,
+      "step": 8416
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5846532056061396e-05,
+      "loss": 0.9055,
+      "step": 8417
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.583523440389395e-05,
+      "loss": 21.2357,
+      "step": 8418
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5823938361212063e-05,
+      "loss": 0.9617,
+      "step": 8419
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5812643928768115e-05,
+      "loss": 0.1875,
+      "step": 8420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5801351107314364e-05,
+      "loss": 18.1089,
+      "step": 8421
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5790059897602976e-05,
+      "loss": 0.9136,
+      "step": 8422
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5778770300385946e-05,
+      "loss": 19.0661,
+      "step": 8423
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.576748231641527e-05,
+      "loss": 0.9217,
+      "step": 8424
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.575619594644278e-05,
+      "loss": 0.9288,
+      "step": 8425
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5744911191220146e-05,
+      "loss": 18.2938,
+      "step": 8426
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5733628051499038e-05,
+      "loss": 18.0309,
+      "step": 8427
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5722346528030973e-05,
+      "loss": 19.7402,
+      "step": 8428
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5711066621567304e-05,
+      "loss": 20.0272,
+      "step": 8429
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.569978833285933e-05,
+      "loss": 18.2868,
+      "step": 8430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.56885116626583e-05,
+      "loss": 0.9282,
+      "step": 8431
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.567723661171522e-05,
+      "loss": 17.8809,
+      "step": 8432
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5665963180781083e-05,
+      "loss": 0.9448,
+      "step": 8433
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.565469137060677e-05,
+      "loss": 19.379,
+      "step": 8434
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.564342118194305e-05,
+      "loss": 0.8764,
+      "step": 8435
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5632152615540507e-05,
+      "loss": 19.8414,
+      "step": 8436
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.562088567214973e-05,
+      "loss": 0.9754,
+      "step": 8437
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5609620352521165e-05,
+      "loss": 0.8952,
+      "step": 8438
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.559835665740508e-05,
+      "loss": 0.9622,
+      "step": 8439
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5587094587551703e-05,
+      "loss": 20.5788,
+      "step": 8440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5575834143711187e-05,
+      "loss": 17.9014,
+      "step": 8441
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5564575326633483e-05,
+      "loss": 19.3532,
+      "step": 8442
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.555331813706847e-05,
+      "loss": 19.4974,
+      "step": 8443
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.554206257576599e-05,
+      "loss": 20.1605,
+      "step": 8444
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5530808643475668e-05,
+      "loss": 17.6671,
+      "step": 8445
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.551955634094707e-05,
+      "loss": 19.9113,
+      "step": 8446
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5508305668929665e-05,
+      "loss": 18.8347,
+      "step": 8447
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5497056628172793e-05,
+      "loss": 0.9441,
+      "step": 8448
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5485809219425695e-05,
+      "loss": 0.9376,
+      "step": 8449
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5474563443437503e-05,
+      "loss": 0.9585,
+      "step": 8450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5463319300957233e-05,
+      "loss": 20.7857,
+      "step": 8451
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.54520767927338e-05,
+      "loss": 0.9506,
+      "step": 8452
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5440835919516004e-05,
+      "loss": 18.9893,
+      "step": 8453
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5429596682052542e-05,
+      "loss": 19.7219,
+      "step": 8454
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5418359081092002e-05,
+      "loss": 19.6199,
+      "step": 8455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5407123117382858e-05,
+      "loss": 0.2061,
+      "step": 8456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.539588879167347e-05,
+      "loss": 0.9559,
+      "step": 8457
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.538465610471211e-05,
+      "loss": 21.5926,
+      "step": 8458
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.537342505724692e-05,
+      "loss": 0.9715,
+      "step": 8459
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5362195650025934e-05,
+      "loss": 0.1932,
+      "step": 8460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5350967883797096e-05,
+      "loss": 20.3607,
+      "step": 8461
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5339741759308218e-05,
+      "loss": 0.963,
+      "step": 8462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5328517277307013e-05,
+      "loss": 0.9331,
+      "step": 8463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5317294438541082e-05,
+      "loss": 19.7091,
+      "step": 8464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5306073243757926e-05,
+      "loss": 18.4904,
+      "step": 8465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5294853693704922e-05,
+      "loss": 0.9559,
+      "step": 8466
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5283635789129347e-05,
+      "loss": 0.228,
+      "step": 8467
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5272419530778357e-05,
+      "loss": 0.1971,
+      "step": 8468
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.526120491939904e-05,
+      "loss": 0.1873,
+      "step": 8469
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.524999195573828e-05,
+      "loss": 19.9554,
+      "step": 8470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.523878064054296e-05,
+      "loss": 18.5587,
+      "step": 8471
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5227570974559816e-05,
+      "loss": 19.4811,
+      "step": 8472
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.52163629585354e-05,
+      "loss": 0.9837,
+      "step": 8473
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5205156593216282e-05,
+      "loss": 20.1286,
+      "step": 8474
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5193951879348832e-05,
+      "loss": 0.9255,
+      "step": 8475
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5182748817679342e-05,
+      "loss": 0.9093,
+      "step": 8476
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.517154740895398e-05,
+      "loss": 20.4584,
+      "step": 8477
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5160347653918826e-05,
+      "loss": 0.9654,
+      "step": 8478
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5149149553319835e-05,
+      "loss": 20.6559,
+      "step": 8479
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5137953107902813e-05,
+      "loss": 0.9483,
+      "step": 8480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5126758318413535e-05,
+      "loss": 19.2825,
+      "step": 8481
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5115565185597646e-05,
+      "loss": 0.9646,
+      "step": 8482
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.510437371020059e-05,
+      "loss": 18.6128,
+      "step": 8483
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.509318389296783e-05,
+      "loss": 0.9799,
+      "step": 8484
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5081995734644663e-05,
+      "loss": 0.2721,
+      "step": 8485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5070809235976232e-05,
+      "loss": 0.9401,
+      "step": 8486
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5059624397707616e-05,
+      "loss": 18.7399,
+      "step": 8487
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5048441220583823e-05,
+      "loss": 17.9732,
+      "step": 8488
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5037259705349658e-05,
+      "loss": 19.6895,
+      "step": 8489
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.502607985274985e-05,
+      "loss": 0.2059,
+      "step": 8490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.50149016635291e-05,
+      "loss": 18.6376,
+      "step": 8491
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.500372513843186e-05,
+      "loss": 20.6528,
+      "step": 8492
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4992550278202566e-05,
+      "loss": 0.2044,
+      "step": 8493
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4981377083585504e-05,
+      "loss": 0.217,
+      "step": 8494
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.497020555532487e-05,
+      "loss": 18.4728,
+      "step": 8495
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4959035694164736e-05,
+      "loss": 0.9556,
+      "step": 8496
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.494786750084905e-05,
+      "loss": 0.9587,
+      "step": 8497
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.493670097612172e-05,
+      "loss": 0.9125,
+      "step": 8498
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.492553612072643e-05,
+      "loss": 18.2858,
+      "step": 8499
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4914372935406815e-05,
+      "loss": 18.2668,
+      "step": 8500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4903211420906446e-05,
+      "loss": 0.1715,
+      "step": 8501
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.489205157796869e-05,
+      "loss": 19.5051,
+      "step": 8502
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4880893407336842e-05,
+      "loss": 18.3847,
+      "step": 8503
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4869736909754106e-05,
+      "loss": 18.611,
+      "step": 8504
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4858582085963544e-05,
+      "loss": 0.1718,
+      "step": 8505
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4847428936708124e-05,
+      "loss": 18.8884,
+      "step": 8506
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.48362774627307e-05,
+      "loss": 0.9509,
+      "step": 8507
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4825127664774006e-05,
+      "loss": 0.9108,
+      "step": 8508
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.481397954358068e-05,
+      "loss": 18.8133,
+      "step": 8509
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4802833099893224e-05,
+      "loss": 17.4745,
+      "step": 8510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4791688334454055e-05,
+      "loss": 19.1275,
+      "step": 8511
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4780545248005464e-05,
+      "loss": 0.9611,
+      "step": 8512
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.476940384128963e-05,
+      "loss": 18.4149,
+      "step": 8513
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4758264115048624e-05,
+      "loss": 19.236,
+      "step": 8514
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.47471260700244e-05,
+      "loss": 0.9286,
+      "step": 8515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.473598970695883e-05,
+      "loss": 0.9677,
+      "step": 8516
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.472485502659358e-05,
+      "loss": 0.9934,
+      "step": 8517
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4713722029670335e-05,
+      "loss": 0.237,
+      "step": 8518
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4702590716930586e-05,
+      "loss": 0.8988,
+      "step": 8519
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4691461089115725e-05,
+      "loss": 0.2126,
+      "step": 8520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4680333146967045e-05,
+      "loss": 0.9191,
+      "step": 8521
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.466920689122571e-05,
+      "loss": 19.6791,
+      "step": 8522
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4658082322632787e-05,
+      "loss": 19.3366,
+      "step": 8523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.464695944192922e-05,
+      "loss": 0.954,
+      "step": 8524
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.463583824985584e-05,
+      "loss": 0.1916,
+      "step": 8525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4624718747153398e-05,
+      "loss": 0.196,
+      "step": 8526
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4613600934562446e-05,
+      "loss": 17.9361,
+      "step": 8527
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4602484812823533e-05,
+      "loss": 0.1921,
+      "step": 8528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4591370382677052e-05,
+      "loss": 18.5638,
+      "step": 8529
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4580257644863213e-05,
+      "loss": 0.174,
+      "step": 8530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.456914660012224e-05,
+      "loss": 0.2153,
+      "step": 8531
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.455803724919417e-05,
+      "loss": 0.1676,
+      "step": 8532
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.45469295928189e-05,
+      "loss": 0.9289,
+      "step": 8533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4535823631736267e-05,
+      "loss": 17.5338,
+      "step": 8534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4524719366686016e-05,
+      "loss": 0.2165,
+      "step": 8535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4513616798407695e-05,
+      "loss": 19.8463,
+      "step": 8536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.450251592764079e-05,
+      "loss": 0.9497,
+      "step": 8537
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4491416755124724e-05,
+      "loss": 0.201,
+      "step": 8538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4480319281598697e-05,
+      "loss": 0.9254,
+      "step": 8539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4469223507801848e-05,
+      "loss": 19.0959,
+      "step": 8540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4458129434473253e-05,
+      "loss": 18.4166,
+      "step": 8541
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4447037062351823e-05,
+      "loss": 18.6809,
+      "step": 8542
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4435946392176322e-05,
+      "loss": 0.8894,
+      "step": 8543
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.442485742468545e-05,
+      "loss": 0.978,
+      "step": 8544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4413770160617828e-05,
+      "loss": 0.1945,
+      "step": 8545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.440268460071187e-05,
+      "loss": 21.5313,
+      "step": 8546
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.439160074570592e-05,
+      "loss": 21.0244,
+      "step": 8547
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.438051859633828e-05,
+      "loss": 0.9863,
+      "step": 8548
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.436943815334701e-05,
+      "loss": 19.4547,
+      "step": 8549
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.435835941747014e-05,
+      "loss": 0.934,
+      "step": 8550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.434728238944556e-05,
+      "loss": 0.9399,
+      "step": 8551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4336207070011053e-05,
+      "loss": 0.1991,
+      "step": 8552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4325133459904293e-05,
+      "loss": 18.6194,
+      "step": 8553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4314061559862833e-05,
+      "loss": 21.7178,
+      "step": 8554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4302991370624107e-05,
+      "loss": 0.9467,
+      "step": 8555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.429192289292544e-05,
+      "loss": 0.9334,
+      "step": 8556
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.428085612750405e-05,
+      "loss": 0.9427,
+      "step": 8557
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.426979107509703e-05,
+      "loss": 17.8212,
+      "step": 8558
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4258727736441374e-05,
+      "loss": 0.1869,
+      "step": 8559
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4247666112273938e-05,
+      "loss": 18.9757,
+      "step": 8560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4236606203331486e-05,
+      "loss": 0.9309,
+      "step": 8561
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4225548010350656e-05,
+      "loss": 0.8812,
+      "step": 8562
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.421449153406798e-05,
+      "loss": 19.3619,
+      "step": 8563
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4203436775219862e-05,
+      "loss": 0.9667,
+      "step": 8564
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4192383734542607e-05,
+      "loss": 18.257,
+      "step": 8565
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4181332412772395e-05,
+      "loss": 19.9673,
+      "step": 8566
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.41702828106453e-05,
+      "loss": 0.9689,
+      "step": 8567
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4159234928897273e-05,
+      "loss": 21.8919,
+      "step": 8568
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4148188768264153e-05,
+      "loss": 19.3982,
+      "step": 8569
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4137144329481664e-05,
+      "loss": 0.9288,
+      "step": 8570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.412610161328543e-05,
+      "loss": 17.3265,
+      "step": 8571
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4115060620410935e-05,
+      "loss": 19.3407,
+      "step": 8572
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4104021351593577e-05,
+      "loss": 0.1549,
+      "step": 8573
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4092983807568576e-05,
+      "loss": 18.675,
+      "step": 8574
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.408194798907113e-05,
+      "loss": 0.9204,
+      "step": 8575
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4070913896836285e-05,
+      "loss": 0.1954,
+      "step": 8576
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.40598815315989e-05,
+      "loss": 0.1805,
+      "step": 8577
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4048850894093843e-05,
+      "loss": 0.2112,
+      "step": 8578
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4037821985055803e-05,
+      "loss": 0.215,
+      "step": 8579
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4026794805219293e-05,
+      "loss": 0.9758,
+      "step": 8580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4015769355318846e-05,
+      "loss": 0.1854,
+      "step": 8581
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4004745636088777e-05,
+      "loss": 19.4597,
+      "step": 8582
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3993723648263344e-05,
+      "loss": 18.557,
+      "step": 8583
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.39827033925766e-05,
+      "loss": 0.9147,
+      "step": 8584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.39716848697626e-05,
+      "loss": 22.7138,
+      "step": 8585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3960668080555237e-05,
+      "loss": 18.8962,
+      "step": 8586
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.394965302568822e-05,
+      "loss": 0.9361,
+      "step": 8587
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3938639705895254e-05,
+      "loss": 0.9409,
+      "step": 8588
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.392762812190989e-05,
+      "loss": 0.1841,
+      "step": 8589
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3916618274465495e-05,
+      "loss": 18.3309,
+      "step": 8590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3905610164295394e-05,
+      "loss": 0.9771,
+      "step": 8591
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.389460379213282e-05,
+      "loss": 19.6495,
+      "step": 8592
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.38835991587108e-05,
+      "loss": 0.1859,
+      "step": 8593
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3872596264762294e-05,
+      "loss": 0.9353,
+      "step": 8594
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.38615951110202e-05,
+      "loss": 19.7447,
+      "step": 8595
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3850595698217188e-05,
+      "loss": 0.9562,
+      "step": 8596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.383959802708589e-05,
+      "loss": 19.7807,
+      "step": 8597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3828602098358804e-05,
+      "loss": 0.9662,
+      "step": 8598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3817607912768308e-05,
+      "loss": 17.5947,
+      "step": 8599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3806615471046668e-05,
+      "loss": 0.9518,
+      "step": 8600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3795624773926027e-05,
+      "loss": 19.946,
+      "step": 8601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3784635822138424e-05,
+      "loss": 0.9616,
+      "step": 8602
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3773648616415766e-05,
+      "loss": 19.9072,
+      "step": 8603
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.376266315748984e-05,
+      "loss": 19.8772,
+      "step": 8604
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.375167944609238e-05,
+      "loss": 18.7638,
+      "step": 8605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3740697482954893e-05,
+      "loss": 0.167,
+      "step": 8606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3729717268808854e-05,
+      "loss": 0.9101,
+      "step": 8607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.371873880438559e-05,
+      "loss": 0.2148,
+      "step": 8608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.370776209041633e-05,
+      "loss": 20.4419,
+      "step": 8609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3696787127632166e-05,
+      "loss": 21.2533,
+      "step": 8610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3685813916764076e-05,
+      "loss": 18.1818,
+      "step": 8611
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3674842458542933e-05,
+      "loss": 20.8775,
+      "step": 8612
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.366387275369949e-05,
+      "loss": 18.5467,
+      "step": 8613
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.365290480296437e-05,
+      "loss": 18.5648,
+      "step": 8614
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3641938607068102e-05,
+      "loss": 18.4985,
+      "step": 8615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3630974166741075e-05,
+      "loss": 0.9399,
+      "step": 8616
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3620011482713585e-05,
+      "loss": 18.9224,
+      "step": 8617
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.360905055571578e-05,
+      "loss": 0.926,
+      "step": 8618
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3598091386477717e-05,
+      "loss": 0.2114,
+      "step": 8619
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3587133975729348e-05,
+      "loss": 0.9387,
+      "step": 8620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.357617832420043e-05,
+      "loss": 0.9562,
+      "step": 8621
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.356522443262072e-05,
+      "loss": 18.0122,
+      "step": 8622
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3554272301719786e-05,
+      "loss": 0.9542,
+      "step": 8623
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3543321932227053e-05,
+      "loss": 0.9334,
+      "step": 8624
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.353237332487191e-05,
+      "loss": 17.8715,
+      "step": 8625
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.352142648038356e-05,
+      "loss": 21.2176,
+      "step": 8626
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3510481399491136e-05,
+      "loss": 19.1442,
+      "step": 8627
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3499538082923606e-05,
+      "loss": 19.0305,
+      "step": 8628
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3488596531409863e-05,
+      "loss": 18.3125,
+      "step": 8629
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.347765674567868e-05,
+      "loss": 0.2267,
+      "step": 8630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3466718726458637e-05,
+      "loss": 19.662,
+      "step": 8631
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3455782474478317e-05,
+      "loss": 0.217,
+      "step": 8632
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3444847990466117e-05,
+      "loss": 19.8699,
+      "step": 8633
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.343391527515028e-05,
+      "loss": 18.5399,
+      "step": 8634
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.342298432925903e-05,
+      "loss": 0.1857,
+      "step": 8635
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.341205515352041e-05,
+      "loss": 0.2162,
+      "step": 8636
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3401127748662328e-05,
+      "loss": 19.5536,
+      "step": 8637
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.339020211541259e-05,
+      "loss": 0.9479,
+      "step": 8638
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3379278254498955e-05,
+      "loss": 19.1454,
+      "step": 8639
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3368356166648943e-05,
+      "loss": 0.9485,
+      "step": 8640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3357435852590022e-05,
+      "loss": 17.1505,
+      "step": 8641
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.334651731304958e-05,
+      "loss": 17.8956,
+      "step": 8642
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3335600548754805e-05,
+      "loss": 0.2164,
+      "step": 8643
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.332468556043279e-05,
+      "loss": 0.9637,
+      "step": 8644
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3313772348810563e-05,
+      "loss": 19.0875,
+      "step": 8645
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3302860914614993e-05,
+      "loss": 0.2157,
+      "step": 8646
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3291951258572802e-05,
+      "loss": 18.0826,
+      "step": 8647
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3281043381410622e-05,
+      "loss": 0.9482,
+      "step": 8648
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.327013728385502e-05,
+      "loss": 18.6592,
+      "step": 8649
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.325923296663234e-05,
+      "loss": 0.904,
+      "step": 8650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3248330430468857e-05,
+      "loss": 18.9056,
+      "step": 8651
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3237429676090787e-05,
+      "loss": 0.8871,
+      "step": 8652
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.322653070422412e-05,
+      "loss": 19.3538,
+      "step": 8653
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.321563351559479e-05,
+      "loss": 0.9087,
+      "step": 8654
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.320473811092861e-05,
+      "loss": 0.9218,
+      "step": 8655
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.319384449095125e-05,
+      "loss": 18.8977,
+      "step": 8656
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.318295265638829e-05,
+      "loss": 0.1728,
+      "step": 8657
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3172062607965166e-05,
+      "loss": 18.0961,
+      "step": 8658
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3161174346407212e-05,
+      "loss": 0.2348,
+      "step": 8659
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3150287872439635e-05,
+      "loss": 0.2142,
+      "step": 8660
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.313940318678752e-05,
+      "loss": 18.4897,
+      "step": 8661
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3128520290175848e-05,
+      "loss": 20.2885,
+      "step": 8662
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3117639183329464e-05,
+      "loss": 18.6306,
+      "step": 8663
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.31067598669731e-05,
+      "loss": 0.9255,
+      "step": 8664
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3095882341831372e-05,
+      "loss": 0.1789,
+      "step": 8665
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3085006608628768e-05,
+      "loss": 19.0175,
+      "step": 8666
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3074132668089665e-05,
+      "loss": 0.9775,
+      "step": 8667
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3063260520938324e-05,
+      "loss": 19.5256,
+      "step": 8668
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3052390167898873e-05,
+      "loss": 0.9163,
+      "step": 8669
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3041521609695325e-05,
+      "loss": 0.9258,
+      "step": 8670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3030654847051593e-05,
+      "loss": 20.6986,
+      "step": 8671
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3019789880691434e-05,
+      "loss": 0.1954,
+      "step": 8672
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3008926711338513e-05,
+      "loss": 19.6682,
+      "step": 8673
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2998065339716374e-05,
+      "loss": 0.9218,
+      "step": 8674
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2987205766548425e-05,
+      "loss": 19.9319,
+      "step": 8675
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.297634799255797e-05,
+      "loss": 20.9755,
+      "step": 8676
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.29654920184682e-05,
+      "loss": 18.8806,
+      "step": 8677
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2954637845002125e-05,
+      "loss": 0.9608,
+      "step": 8678
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2943785472882734e-05,
+      "loss": 17.7038,
+      "step": 8679
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2932934902832847e-05,
+      "loss": 0.2129,
+      "step": 8680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2922086135575105e-05,
+      "loss": 19.991,
+      "step": 8681
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2911239171832145e-05,
+      "loss": 18.4139,
+      "step": 8682
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2900394012326422e-05,
+      "loss": 0.1856,
+      "step": 8683
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.288955065778024e-05,
+      "loss": 21.0895,
+      "step": 8684
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2878709108915812e-05,
+      "loss": 18.9179,
+      "step": 8685
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2867869366455296e-05,
+      "loss": 0.1915,
+      "step": 8686
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2857031431120613e-05,
+      "loss": 19.116,
+      "step": 8687
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2846195303633615e-05,
+      "loss": 17.1758,
+      "step": 8688
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.283536098471608e-05,
+      "loss": 19.1924,
+      "step": 8689
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2824528475089624e-05,
+      "loss": 20.3202,
+      "step": 8690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2813697775475685e-05,
+      "loss": 18.5684,
+      "step": 8691
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2802868886595695e-05,
+      "loss": 0.9014,
+      "step": 8692
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2792041809170907e-05,
+      "loss": 0.19,
+      "step": 8693
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.278121654392242e-05,
+      "loss": 0.9546,
+      "step": 8694
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.277039309157124e-05,
+      "loss": 19.1182,
+      "step": 8695
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2759571452838324e-05,
+      "loss": 0.9181,
+      "step": 8696
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.274875162844438e-05,
+      "loss": 0.2108,
+      "step": 8697
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2737933619110064e-05,
+      "loss": 19.8935,
+      "step": 8698
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.272711742555596e-05,
+      "loss": 0.215,
+      "step": 8699
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2716303048502417e-05,
+      "loss": 0.203,
+      "step": 8700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.270549048866975e-05,
+      "loss": 19.5206,
+      "step": 8701
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2694679746778115e-05,
+      "loss": 20.6595,
+      "step": 8702
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2683870823547566e-05,
+      "loss": 18.3225,
+      "step": 8703
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2673063719698024e-05,
+      "loss": 0.9667,
+      "step": 8704
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.266225843594929e-05,
+      "loss": 19.1967,
+      "step": 8705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2651454973021054e-05,
+      "loss": 0.9536,
+      "step": 8706
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2640653331632867e-05,
+      "loss": 0.9334,
+      "step": 8707
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.262985351250417e-05,
+      "loss": 20.5391,
+      "step": 8708
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.261905551635429e-05,
+      "loss": 0.9808,
+      "step": 8709
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.260825934390241e-05,
+      "loss": 18.8231,
+      "step": 8710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2597464995867624e-05,
+      "loss": 18.4943,
+      "step": 8711
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2586672472968868e-05,
+      "loss": 19.1982,
+      "step": 8712
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.257588177592498e-05,
+      "loss": 20.543,
+      "step": 8713
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.256509290545467e-05,
+      "loss": 0.2209,
+      "step": 8714
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.255430586227653e-05,
+      "loss": 19.0841,
+      "step": 8715
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2543520647109024e-05,
+      "loss": 0.2162,
+      "step": 8716
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.25327372606705e-05,
+      "loss": 20.707,
+      "step": 8717
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2521955703679183e-05,
+      "loss": 19.2597,
+      "step": 8718
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2511175976853164e-05,
+      "loss": 0.2082,
+      "step": 8719
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.250039808091044e-05,
+      "loss": 20.5237,
+      "step": 8720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2489622016568858e-05,
+      "loss": 0.2113,
+      "step": 8721
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.247884778454615e-05,
+      "loss": 18.4829,
+      "step": 8722
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.246807538555994e-05,
+      "loss": 0.1707,
+      "step": 8723
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.245730482032773e-05,
+      "loss": 0.9182,
+      "step": 8724
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2446536089566837e-05,
+      "loss": 17.8933,
+      "step": 8725
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.243576919399457e-05,
+      "loss": 0.9002,
+      "step": 8726
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.242500413432805e-05,
+      "loss": 0.9377,
+      "step": 8727
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2414240911284217e-05,
+      "loss": 0.9334,
+      "step": 8728
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.240347952558002e-05,
+      "loss": 18.9565,
+      "step": 8729
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2392719977932188e-05,
+      "loss": 19.1313,
+      "step": 8730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.238196226905736e-05,
+      "loss": 17.7796,
+      "step": 8731
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2371206399672052e-05,
+      "loss": 20.1584,
+      "step": 8732
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.236045237049265e-05,
+      "loss": 0.1922,
+      "step": 8733
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.234970018223545e-05,
+      "loss": 19.2553,
+      "step": 8734
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2338949835616533e-05,
+      "loss": 0.1785,
+      "step": 8735
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2328201331351984e-05,
+      "loss": 19.3163,
+      "step": 8736
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.23174546701577e-05,
+      "loss": 20.0412,
+      "step": 8737
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2306709852749402e-05,
+      "loss": 0.963,
+      "step": 8738
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.22959668798428e-05,
+      "loss": 19.2801,
+      "step": 8739
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2285225752153426e-05,
+      "loss": 0.9348,
+      "step": 8740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2274486470396654e-05,
+      "loss": 17.4362,
+      "step": 8741
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2263749035287768e-05,
+      "loss": 19.3109,
+      "step": 8742
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.225301344754199e-05,
+      "loss": 0.9504,
+      "step": 8743
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2242279707874302e-05,
+      "loss": 19.0867,
+      "step": 8744
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.223154781699962e-05,
+      "loss": 21.0134,
+      "step": 8745
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2220817775632795e-05,
+      "loss": 0.9125,
+      "step": 8746
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.221008958448844e-05,
+      "loss": 18.8119,
+      "step": 8747
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.219936324428111e-05,
+      "loss": 16.8896,
+      "step": 8748
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2188638755725266e-05,
+      "loss": 18.6175,
+      "step": 8749
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2177916119535174e-05,
+      "loss": 0.9297,
+      "step": 8750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2167195336425017e-05,
+      "loss": 19.5185,
+      "step": 8751
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.215647640710884e-05,
+      "loss": 0.2101,
+      "step": 8752
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.214575933230062e-05,
+      "loss": 0.1971,
+      "step": 8753
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.213504411271411e-05,
+      "loss": 0.9655,
+      "step": 8754
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2124330749063e-05,
+      "loss": 19.3943,
+      "step": 8755
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.21136192420609e-05,
+      "loss": 20.6294,
+      "step": 8756
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.210290959242119e-05,
+      "loss": 18.4465,
+      "step": 8757
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2092201800857205e-05,
+      "loss": 0.1896,
+      "step": 8758
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2081495868082124e-05,
+      "loss": 19.0318,
+      "step": 8759
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.207079179480902e-05,
+      "loss": 19.3635,
+      "step": 8760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2060089581750836e-05,
+      "loss": 0.1707,
+      "step": 8761
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.204938922962039e-05,
+      "loss": 19.6965,
+      "step": 8762
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2038690739130364e-05,
+      "loss": 0.8901,
+      "step": 8763
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2027994110993334e-05,
+      "loss": 20.2744,
+      "step": 8764
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2017299345921744e-05,
+      "loss": 18.5673,
+      "step": 8765
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.200660644462792e-05,
+      "loss": 18.2013,
+      "step": 8766
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.199591540782405e-05,
+      "loss": 20.5481,
+      "step": 8767
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1985226236222217e-05,
+      "loss": 0.9193,
+      "step": 8768
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.197453893053436e-05,
+      "loss": 21.5363,
+      "step": 8769
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1963853491472314e-05,
+      "loss": 0.9868,
+      "step": 8770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.195316991974776e-05,
+      "loss": 20.8047,
+      "step": 8771
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.19424882160723e-05,
+      "loss": 0.9451,
+      "step": 8772
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.193180838115736e-05,
+      "loss": 0.2068,
+      "step": 8773
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.192113041571428e-05,
+      "loss": 0.1851,
+      "step": 8774
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.191045432045426e-05,
+      "loss": 0.1569,
+      "step": 8775
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1899780096088375e-05,
+      "loss": 0.9648,
+      "step": 8776
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.188910774332758e-05,
+      "loss": 21.1531,
+      "step": 8777
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.18784372628827e-05,
+      "loss": 0.9157,
+      "step": 8778
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.186776865546445e-05,
+      "loss": 0.2716,
+      "step": 8779
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1857101921783392e-05,
+      "loss": 0.9128,
+      "step": 8780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1846437062550008e-05,
+      "loss": 18.4753,
+      "step": 8781
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1835774078474568e-05,
+      "loss": 0.9141,
+      "step": 8782
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1825112970267342e-05,
+      "loss": 0.9585,
+      "step": 8783
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1814453738638396e-05,
+      "loss": 0.1755,
+      "step": 8784
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.180379638429763e-05,
+      "loss": 18.9131,
+      "step": 8785
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.179314090795494e-05,
+      "loss": 0.986,
+      "step": 8786
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1782487310320022e-05,
+      "loss": 21.5199,
+      "step": 8787
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1771835592102413e-05,
+      "loss": 19.3095,
+      "step": 8788
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1761185754011582e-05,
+      "loss": 0.9303,
+      "step": 8789
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.17505377967569e-05,
+      "loss": 19.2347,
+      "step": 8790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1739891721047517e-05,
+      "loss": 0.9405,
+      "step": 8791
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.172924752759251e-05,
+      "loss": 17.8115,
+      "step": 8792
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1718605217100897e-05,
+      "loss": 18.9649,
+      "step": 8793
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1707964790281443e-05,
+      "loss": 0.9373,
+      "step": 8794
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1697326247842848e-05,
+      "loss": 18.8825,
+      "step": 8795
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1686689590493725e-05,
+      "loss": 0.8872,
+      "step": 8796
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1676054818942528e-05,
+      "loss": 0.938,
+      "step": 8797
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1665421933897547e-05,
+      "loss": 0.1749,
+      "step": 8798
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1654790936066976e-05,
+      "loss": 18.8297,
+      "step": 8799
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1644161826158944e-05,
+      "loss": 19.3662,
+      "step": 8800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.163353460488135e-05,
+      "loss": 0.2028,
+      "step": 8801
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1622909272942016e-05,
+      "loss": 19.2507,
+      "step": 8802
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1612285831048685e-05,
+      "loss": 0.1861,
+      "step": 8803
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.160166427990888e-05,
+      "loss": 0.8999,
+      "step": 8804
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.159104462023006e-05,
+      "loss": 18.4852,
+      "step": 8805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.158042685271955e-05,
+      "loss": 0.1909,
+      "step": 8806
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1569810978084532e-05,
+      "loss": 20.3357,
+      "step": 8807
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.155919699703208e-05,
+      "loss": 18.6891,
+      "step": 8808
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1548584910269136e-05,
+      "loss": 18.3178,
+      "step": 8809
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1537974718502506e-05,
+      "loss": 18.9823,
+      "step": 8810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.152736642243889e-05,
+      "loss": 0.8453,
+      "step": 8811
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1516760022784838e-05,
+      "loss": 19.2393,
+      "step": 8812
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1506155520246797e-05,
+      "loss": 19.1662,
+      "step": 8813
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.149555291553106e-05,
+      "loss": 19.2217,
+      "step": 8814
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1484952209343824e-05,
+      "loss": 0.9462,
+      "step": 8815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.147435340239114e-05,
+      "loss": 0.9612,
+      "step": 8816
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.146375649537894e-05,
+      "loss": 19.5012,
+      "step": 8817
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1453161489013025e-05,
+      "loss": 19.5373,
+      "step": 8818
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.144256838399907e-05,
+      "loss": 20.5045,
+      "step": 8819
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1431977181042628e-05,
+      "loss": 21.1731,
+      "step": 8820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.142138788084912e-05,
+      "loss": 19.7649,
+      "step": 8821
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.141080048412385e-05,
+      "loss": 0.2039,
+      "step": 8822
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1400214991571975e-05,
+      "loss": 0.1968,
+      "step": 8823
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1389631403898548e-05,
+      "loss": 0.9476,
+      "step": 8824
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1379049721808474e-05,
+      "loss": 17.7842,
+      "step": 8825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1368469946006557e-05,
+      "loss": 17.8032,
+      "step": 8826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.135789207719745e-05,
+      "loss": 19.8217,
+      "step": 8827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1347316116085715e-05,
+      "loss": 0.2082,
+      "step": 8828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1336742063375693e-05,
+      "loss": 18.169,
+      "step": 8829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.132616991977172e-05,
+      "loss": 18.784,
+      "step": 8830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.131559968597796e-05,
+      "loss": 19.8166,
+      "step": 8831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1305031362698385e-05,
+      "loss": 0.1859,
+      "step": 8832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1294464950636938e-05,
+      "loss": 0.938,
+      "step": 8833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.12839004504974e-05,
+      "loss": 0.9039,
+      "step": 8834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1273337862983357e-05,
+      "loss": 19.4914,
+      "step": 8835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1262777188798382e-05,
+      "loss": 0.9318,
+      "step": 8836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1252218428645846e-05,
+      "loss": 0.1806,
+      "step": 8837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.124166158322904e-05,
+      "loss": 19.1455,
+      "step": 8838
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1231106653251027e-05,
+      "loss": 0.9183,
+      "step": 8839
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1220553639414875e-05,
+      "loss": 19.4478,
+      "step": 8840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1210002542423474e-05,
+      "loss": 0.9477,
+      "step": 8841
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1199453362979504e-05,
+      "loss": 0.1681,
+      "step": 8842
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1188906101785662e-05,
+      "loss": 18.7271,
+      "step": 8843
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.117836075954443e-05,
+      "loss": 0.9637,
+      "step": 8844
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1167817336958147e-05,
+      "loss": 19.1605,
+      "step": 8845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1157275834729058e-05,
+      "loss": 0.2115,
+      "step": 8846
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1146736253559317e-05,
+      "loss": 18.1194,
+      "step": 8847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.113619859415087e-05,
+      "loss": 0.9419,
+      "step": 8848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1125662857205565e-05,
+      "loss": 0.1965,
+      "step": 8849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1115129043425187e-05,
+      "loss": 0.9292,
+      "step": 8850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.110459715351128e-05,
+      "loss": 0.9638,
+      "step": 8851
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.109406718816532e-05,
+      "loss": 17.1593,
+      "step": 8852
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.108353914808871e-05,
+      "loss": 0.1983,
+      "step": 8853
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1073013033982608e-05,
+      "loss": 18.4056,
+      "step": 8854
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1062488846548117e-05,
+      "loss": 20.2401,
+      "step": 8855
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.10519665864862e-05,
+      "loss": 1.0097,
+      "step": 8856
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1041446254497687e-05,
+      "loss": 18.0977,
+      "step": 8857
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1030927851283283e-05,
+      "loss": 19.3196,
+      "step": 8858
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1020411377543543e-05,
+      "loss": 18.6193,
+      "step": 8859
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1009896833978966e-05,
+      "loss": 19.4706,
+      "step": 8860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.099938422128982e-05,
+      "loss": 18.4007,
+      "step": 8861
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0988873540176307e-05,
+      "loss": 18.0648,
+      "step": 8862
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0978364791338485e-05,
+      "loss": 0.9043,
+      "step": 8863
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0967857975476297e-05,
+      "loss": 0.2237,
+      "step": 8864
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0957353093289534e-05,
+      "loss": 19.3983,
+      "step": 8865
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.094685014547787e-05,
+      "loss": 0.9228,
+      "step": 8866
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0936349132740856e-05,
+      "loss": 0.9101,
+      "step": 8867
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0925850055777908e-05,
+      "loss": 18.6239,
+      "step": 8868
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.091535291528831e-05,
+      "loss": 18.7791,
+      "step": 8869
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0904857711971222e-05,
+      "loss": 0.9758,
+      "step": 8870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0894364446525672e-05,
+      "loss": 18.3884,
+      "step": 8871
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0883873119650564e-05,
+      "loss": 20.2175,
+      "step": 8872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0873383732044666e-05,
+      "loss": 17.4133,
+      "step": 8873
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0862896284406615e-05,
+      "loss": 19.7347,
+      "step": 8874
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0852410777434935e-05,
+      "loss": 20.4744,
+      "step": 8875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0841927211828e-05,
+      "loss": 0.2423,
+      "step": 8876
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0831445588284065e-05,
+      "loss": 0.9599,
+      "step": 8877
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0820965907501278e-05,
+      "loss": 19.2103,
+      "step": 8878
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0810488170177576e-05,
+      "loss": 19.356,
+      "step": 8879
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.080001237701088e-05,
+      "loss": 0.9236,
+      "step": 8880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.078953852869891e-05,
+      "loss": 19.5519,
+      "step": 8881
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0779066625939275e-05,
+      "loss": 19.2722,
+      "step": 8882
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0768596669429446e-05,
+      "loss": 0.9782,
+      "step": 8883
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.075812865986677e-05,
+      "loss": 0.1921,
+      "step": 8884
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0747662597948493e-05,
+      "loss": 0.9668,
+      "step": 8885
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.073719848437164e-05,
+      "loss": 0.2263,
+      "step": 8886
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0726736319833228e-05,
+      "loss": 18.9591,
+      "step": 8887
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0716276105030084e-05,
+      "loss": 20.6819,
+      "step": 8888
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0705817840658852e-05,
+      "loss": 0.9205,
+      "step": 8889
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0695361527416156e-05,
+      "loss": 19.1056,
+      "step": 8890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0684907165998442e-05,
+      "loss": 0.9403,
+      "step": 8891
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.067445475710197e-05,
+      "loss": 19.5645,
+      "step": 8892
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0664004301422924e-05,
+      "loss": 18.7844,
+      "step": 8893
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.065355579965741e-05,
+      "loss": 18.8223,
+      "step": 8894
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.064310925250129e-05,
+      "loss": 0.9371,
+      "step": 8895
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.063266466065036e-05,
+      "loss": 0.9684,
+      "step": 8896
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.062222202480032e-05,
+      "loss": 0.1941,
+      "step": 8897
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0611781345646647e-05,
+      "loss": 17.5887,
+      "step": 8898
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0601342623884743e-05,
+      "loss": 0.2332,
+      "step": 8899
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.059090586020993e-05,
+      "loss": 19.659,
+      "step": 8900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0580471055317284e-05,
+      "loss": 18.9427,
+      "step": 8901
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0570038209901844e-05,
+      "loss": 0.9198,
+      "step": 8902
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0559607324658447e-05,
+      "loss": 0.8904,
+      "step": 8903
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.054917840028191e-05,
+      "loss": 20.5219,
+      "step": 8904
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0538751437466784e-05,
+      "loss": 0.9412,
+      "step": 8905
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.052832643690756e-05,
+      "loss": 0.9286,
+      "step": 8906
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0517903399298633e-05,
+      "loss": 19.6002,
+      "step": 8907
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0507482325334188e-05,
+      "loss": 19.7808,
+      "step": 8908
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0497063215708324e-05,
+      "loss": 18.7762,
+      "step": 8909
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0486646071115006e-05,
+      "loss": 0.9697,
+      "step": 8910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0476230892248065e-05,
+      "loss": 0.9192,
+      "step": 8911
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0465817679801196e-05,
+      "loss": 19.5368,
+      "step": 8912
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0455406434467976e-05,
+      "loss": 19.859,
+      "step": 8913
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0444997156941832e-05,
+      "loss": 0.914,
+      "step": 8914
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0434589847916073e-05,
+      "loss": 0.9538,
+      "step": 8915
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0424184508083877e-05,
+      "loss": 0.9346,
+      "step": 8916
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.041378113813829e-05,
+      "loss": 18.5042,
+      "step": 8917
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0403379738772217e-05,
+      "loss": 0.1946,
+      "step": 8918
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0392980310678446e-05,
+      "loss": 18.1148,
+      "step": 8919
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0382582854549627e-05,
+      "loss": 19.2033,
+      "step": 8920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0372187371078272e-05,
+      "loss": 18.8933,
+      "step": 8921
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0361793860956773e-05,
+      "loss": 18.717,
+      "step": 8922
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0351402324877388e-05,
+      "loss": 0.1661,
+      "step": 8923
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0341012763532243e-05,
+      "loss": 19.2673,
+      "step": 8924
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0330625177613322e-05,
+      "loss": 0.8799,
+      "step": 8925
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.03202395678125e-05,
+      "loss": 0.2104,
+      "step": 8926
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0309855934821492e-05,
+      "loss": 0.1779,
+      "step": 8927
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0299474279331903e-05,
+      "loss": 18.4113,
+      "step": 8928
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.028909460203521e-05,
+      "loss": 20.3472,
+      "step": 8929
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0278716903622724e-05,
+      "loss": 19.1166,
+      "step": 8930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.026834118478567e-05,
+      "loss": 0.8945,
+      "step": 8931
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0257967446215125e-05,
+      "loss": 0.8945,
+      "step": 8932
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.024759568860198e-05,
+      "loss": 18.4414,
+      "step": 8933
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.023722591263709e-05,
+      "loss": 18.8251,
+      "step": 8934
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0226858119011137e-05,
+      "loss": 0.9595,
+      "step": 8935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0216492308414603e-05,
+      "loss": 19.7047,
+      "step": 8936
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0206128481537962e-05,
+      "loss": 0.1945,
+      "step": 8937
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0195766639071482e-05,
+      "loss": 0.1934,
+      "step": 8938
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0185406781705268e-05,
+      "loss": 0.9499,
+      "step": 8939
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0175048910129384e-05,
+      "loss": 17.8702,
+      "step": 8940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0164693025033704e-05,
+      "loss": 0.9425,
+      "step": 8941
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0154339127107957e-05,
+      "loss": 0.1823,
+      "step": 8942
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0143987217041747e-05,
+      "loss": 20.0402,
+      "step": 8943
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.01336372955246e-05,
+      "loss": 0.9912,
+      "step": 8944
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0123289363245872e-05,
+      "loss": 0.9224,
+      "step": 8945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0112943420894727e-05,
+      "loss": 19.0267,
+      "step": 8946
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.010259946916031e-05,
+      "loss": 20.8277,
+      "step": 8947
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0092257508731566e-05,
+      "loss": 18.3375,
+      "step": 8948
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.008191754029729e-05,
+      "loss": 0.1901,
+      "step": 8949
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0071579564546177e-05,
+      "loss": 18.2581,
+      "step": 8950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0061243582166827e-05,
+      "loss": 17.0362,
+      "step": 8951
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0050909593847616e-05,
+      "loss": 0.9187,
+      "step": 8952
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0040577600276833e-05,
+      "loss": 19.011,
+      "step": 8953
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0030247602142693e-05,
+      "loss": 19.1559,
+      "step": 8954
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.001991960013317e-05,
+      "loss": 0.9749,
+      "step": 8955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0009593594936165e-05,
+      "loss": 20.2332,
+      "step": 8956
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999926958723945e-05,
+      "loss": 18.472,
+      "step": 8957
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.998894757773065e-05,
+      "loss": 19.4424,
+      "step": 8958
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9978627567097248e-05,
+      "loss": 18.8025,
+      "step": 8959
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9968309556026614e-05,
+      "loss": 0.1877,
+      "step": 8960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.995799354520598e-05,
+      "loss": 0.2236,
+      "step": 8961
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.994767953532243e-05,
+      "loss": 19.5808,
+      "step": 8962
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9937367527062927e-05,
+      "loss": 19.769,
+      "step": 8963
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.99270575211143e-05,
+      "loss": 17.8291,
+      "step": 8964
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9916749518163247e-05,
+      "loss": 0.1643,
+      "step": 8965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.990644351889632e-05,
+      "loss": 19.087,
+      "step": 8966
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9896139523999956e-05,
+      "loss": 18.7971,
+      "step": 8967
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9885837534160447e-05,
+      "loss": 18.9932,
+      "step": 8968
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9875537550063946e-05,
+      "loss": 18.8259,
+      "step": 8969
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9865239572396488e-05,
+      "loss": 0.2033,
+      "step": 8970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9854943601843963e-05,
+      "loss": 19.4759,
+      "step": 8971
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.984464963909213e-05,
+      "loss": 18.669,
+      "step": 8972
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.983435768482661e-05,
+      "loss": 0.9911,
+      "step": 8973
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9824067739732906e-05,
+      "loss": 0.9637,
+      "step": 8974
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9813779804496368e-05,
+      "loss": 18.7092,
+      "step": 8975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9803493879802225e-05,
+      "loss": 0.1799,
+      "step": 8976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9793209966335563e-05,
+      "loss": 0.9731,
+      "step": 8977
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.978292806478134e-05,
+      "loss": 17.7408,
+      "step": 8978
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9772648175824378e-05,
+      "loss": 18.5463,
+      "step": 8979
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9762370300149368e-05,
+      "loss": 0.2094,
+      "step": 8980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.975209443844086e-05,
+      "loss": 20.9629,
+      "step": 8981
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9741820591383292e-05,
+      "loss": 17.3842,
+      "step": 8982
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.97315487596609e-05,
+      "loss": 0.1953,
+      "step": 8983
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9721278943957887e-05,
+      "loss": 18.4284,
+      "step": 8984
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9711011144958276e-05,
+      "loss": 0.9456,
+      "step": 8985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9700745363345885e-05,
+      "loss": 19.7746,
+      "step": 8986
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9690481599804527e-05,
+      "loss": 18.2265,
+      "step": 8987
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9680219855017788e-05,
+      "loss": 17.5193,
+      "step": 8988
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.966996012966918e-05,
+      "loss": 19.0274,
+      "step": 8989
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9659702424441982e-05,
+      "loss": 19.6571,
+      "step": 8990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9649446740019467e-05,
+      "loss": 18.4494,
+      "step": 8991
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.96391930770847e-05,
+      "loss": 19.6419,
+      "step": 8992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.962894143632058e-05,
+      "loss": 0.1844,
+      "step": 8993
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9618691818409967e-05,
+      "loss": 0.9812,
+      "step": 8994
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9608444224035526e-05,
+      "loss": 0.2092,
+      "step": 8995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.959819865387976e-05,
+      "loss": 0.9131,
+      "step": 8996
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9587955108625077e-05,
+      "loss": 19.5408,
+      "step": 8997
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9577713588953795e-05,
+      "loss": 20.9685,
+      "step": 8998
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9567474095547994e-05,
+      "loss": 19.4345,
+      "step": 8999
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.955723662908967e-05,
+      "loss": 0.9182,
+      "step": 9000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9547001190260744e-05,
+      "loss": 0.1805,
+      "step": 9001
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.953676777974288e-05,
+      "loss": 0.2259,
+      "step": 9002
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9526536398217682e-05,
+      "loss": 20.0906,
+      "step": 9003
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9516307046366655e-05,
+      "loss": 0.1893,
+      "step": 9004
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.950607972487107e-05,
+      "loss": 0.9817,
+      "step": 9005
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.949585443441213e-05,
+      "loss": 17.4014,
+      "step": 9006
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9485631175670872e-05,
+      "loss": 16.8623,
+      "step": 9007
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.947540994932826e-05,
+      "loss": 0.8911,
+      "step": 9008
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.946519075606503e-05,
+      "loss": 20.9919,
+      "step": 9009
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.945497359656182e-05,
+      "loss": 0.2315,
+      "step": 9010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9444758471499202e-05,
+      "loss": 20.1912,
+      "step": 9011
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.943454538155749e-05,
+      "loss": 0.9067,
+      "step": 9012
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9424334327416936e-05,
+      "loss": 18.5686,
+      "step": 9013
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9414125309757665e-05,
+      "loss": 19.5912,
+      "step": 9014
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9403918329259618e-05,
+      "loss": 19.2664,
+      "step": 9015
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9393713386602648e-05,
+      "loss": 0.9725,
+      "step": 9016
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9383510482466442e-05,
+      "loss": 0.199,
+      "step": 9017
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.937330961753056e-05,
+      "loss": 18.3883,
+      "step": 9018
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9363110792474432e-05,
+      "loss": 19.3397,
+      "step": 9019
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9352914007977347e-05,
+      "loss": 17.4288,
+      "step": 9020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9342719264718452e-05,
+      "loss": 0.9585,
+      "step": 9021
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9332526563376768e-05,
+      "loss": 0.9598,
+      "step": 9022
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932233590463118e-05,
+      "loss": 20.7985,
+      "step": 9023
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9312147289160426e-05,
+      "loss": 18.4569,
+      "step": 9024
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9301960717643118e-05,
+      "loss": 0.9649,
+      "step": 9025
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9291776190757734e-05,
+      "loss": 0.9333,
+      "step": 9026
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.92815937091826e-05,
+      "loss": 17.4427,
+      "step": 9027
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9271413273595923e-05,
+      "loss": 19.2704,
+      "step": 9028
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9261234884675773e-05,
+      "loss": 0.9384,
+      "step": 9029
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.925105854310007e-05,
+      "loss": 19.9191,
+      "step": 9030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9240884249546603e-05,
+      "loss": 0.9249,
+      "step": 9031
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9230712004693035e-05,
+      "loss": 0.1868,
+      "step": 9032
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9220541809216886e-05,
+      "loss": 18.4852,
+      "step": 9033
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9210373663795527e-05,
+      "loss": 20.4825,
+      "step": 9034
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9200207569106216e-05,
+      "loss": 18.6442,
+      "step": 9035
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.919004352582607e-05,
+      "loss": 18.1481,
+      "step": 9036
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.917988153463201e-05,
+      "loss": 19.8835,
+      "step": 9037
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9169721596200936e-05,
+      "loss": 17.488,
+      "step": 9038
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9159563711209534e-05,
+      "loss": 0.8865,
+      "step": 9039
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9149407880334314e-05,
+      "loss": 18.5998,
+      "step": 9040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9139254104251764e-05,
+      "loss": 17.4639,
+      "step": 9041
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9129102383638164e-05,
+      "loss": 0.9462,
+      "step": 9042
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.911895271916962e-05,
+      "loss": 0.9345,
+      "step": 9043
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9108805111522187e-05,
+      "loss": 0.8792,
+      "step": 9044
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9098659561371755e-05,
+      "loss": 20.7697,
+      "step": 9045
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.908851606939403e-05,
+      "loss": 19.4502,
+      "step": 9046
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9078374636264606e-05,
+      "loss": 18.4744,
+      "step": 9047
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9068235262659012e-05,
+      "loss": 19.0779,
+      "step": 9048
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.905809794925252e-05,
+      "loss": 0.9521,
+      "step": 9049
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9047962696720317e-05,
+      "loss": 0.9294,
+      "step": 9050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9037829505737496e-05,
+      "loss": 0.9226,
+      "step": 9051
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9027698376978988e-05,
+      "loss": 0.1829,
+      "step": 9052
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9017569311119522e-05,
+      "loss": 0.2191,
+      "step": 9053
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9007442308833744e-05,
+      "loss": 0.9144,
+      "step": 9054
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8997317370796213e-05,
+      "loss": 0.1887,
+      "step": 9055
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.898719449768125e-05,
+      "loss": 0.1687,
+      "step": 9056
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8977073690163073e-05,
+      "loss": 17.5919,
+      "step": 9057
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.896695494891584e-05,
+      "loss": 0.189,
+      "step": 9058
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8956838274613447e-05,
+      "loss": 0.9102,
+      "step": 9059
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8946723667929727e-05,
+      "loss": 0.1929,
+      "step": 9060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8936611129538366e-05,
+      "loss": 18.286,
+      "step": 9061
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8926500660112907e-05,
+      "loss": 0.9174,
+      "step": 9062
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.891639226032675e-05,
+      "loss": 0.8948,
+      "step": 9063
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8906285930853157e-05,
+      "loss": 0.9622,
+      "step": 9064
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.889618167236526e-05,
+      "loss": 17.993,
+      "step": 9065
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.888607948553605e-05,
+      "loss": 0.1817,
+      "step": 9066
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8875979371038388e-05,
+      "loss": 19.0239,
+      "step": 9067
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8865881329544978e-05,
+      "loss": 0.1945,
+      "step": 9068
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8855785361728394e-05,
+      "loss": 18.9011,
+      "step": 9069
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8845691468261077e-05,
+      "loss": 18.2872,
+      "step": 9070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.883559964981534e-05,
+      "loss": 0.9679,
+      "step": 9071
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8825509907063327e-05,
+      "loss": 19.8133,
+      "step": 9072
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8815422240677073e-05,
+      "loss": 0.2201,
+      "step": 9073
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8805336651328458e-05,
+      "loss": 18.6864,
+      "step": 9074
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.879525313968923e-05,
+      "loss": 0.196,
+      "step": 9075
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8785171706431e-05,
+      "loss": 0.9469,
+      "step": 9076
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.877509235222523e-05,
+      "loss": 0.199,
+      "step": 9077
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.876501507774327e-05,
+      "loss": 20.3397,
+      "step": 9078
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8754939883656293e-05,
+      "loss": 18.0406,
+      "step": 9079
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8744866770635367e-05,
+      "loss": 0.9084,
+      "step": 9080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8734795739351402e-05,
+      "loss": 19.0298,
+      "step": 9081
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8724726790475173e-05,
+      "loss": 0.951,
+      "step": 9082
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.871465992467733e-05,
+      "loss": 0.9625,
+      "step": 9083
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.870459514262836e-05,
+      "loss": 0.1879,
+      "step": 9084
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.869453244499863e-05,
+      "loss": 0.9577,
+      "step": 9085
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8684471832458384e-05,
+      "loss": 0.2546,
+      "step": 9086
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8674413305677647e-05,
+      "loss": 18.6743,
+      "step": 9087
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.866435686532641e-05,
+      "loss": 0.2099,
+      "step": 9088
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8654302512074494e-05,
+      "loss": 0.1489,
+      "step": 9089
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8644250246591505e-05,
+      "loss": 18.8755,
+      "step": 9090
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8634200069547024e-05,
+      "loss": 18.5637,
+      "step": 9091
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8624151981610422e-05,
+      "loss": 18.5855,
+      "step": 9092
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8614105983450963e-05,
+      "loss": 0.1906,
+      "step": 9093
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8604062075737705e-05,
+      "loss": 18.695,
+      "step": 9094
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8594020259139678e-05,
+      "loss": 0.1578,
+      "step": 9095
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8583980534325706e-05,
+      "loss": 18.5582,
+      "step": 9096
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.857394290196443e-05,
+      "loss": 20.7271,
+      "step": 9097
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8563907362724458e-05,
+      "loss": 0.2145,
+      "step": 9098
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8553873917274206e-05,
+      "loss": 19.6599,
+      "step": 9099
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8543842566281903e-05,
+      "loss": 21.3326,
+      "step": 9100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.85338133104157e-05,
+      "loss": 0.9207,
+      "step": 9101
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8523786150343624e-05,
+      "loss": 0.9409,
+      "step": 9102
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.85137610867335e-05,
+      "loss": 0.96,
+      "step": 9103
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.850373812025302e-05,
+      "loss": 19.3309,
+      "step": 9104
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8493717251569838e-05,
+      "loss": 19.6812,
+      "step": 9105
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8483698481351312e-05,
+      "loss": 0.9076,
+      "step": 9106
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.847368181026476e-05,
+      "loss": 0.9203,
+      "step": 9107
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8463667238977377e-05,
+      "loss": 17.5086,
+      "step": 9108
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8453654768156138e-05,
+      "loss": 0.9344,
+      "step": 9109
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.844364439846793e-05,
+      "loss": 0.9035,
+      "step": 9110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.843363613057949e-05,
+      "loss": 19.5861,
+      "step": 9111
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8423629965157425e-05,
+      "loss": 0.9592,
+      "step": 9112
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.841362590286818e-05,
+      "loss": 0.1908,
+      "step": 9113
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8403623944378057e-05,
+      "loss": 0.2143,
+      "step": 9114
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8393624090353294e-05,
+      "loss": 0.2014,
+      "step": 9115
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8383626341459865e-05,
+      "loss": 17.6321,
+      "step": 9116
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8373630698363688e-05,
+      "loss": 0.9565,
+      "step": 9117
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.836363716173052e-05,
+      "loss": 20.0357,
+      "step": 9118
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8353645732225976e-05,
+      "loss": 20.1462,
+      "step": 9119
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.834365641051553e-05,
+      "loss": 17.3985,
+      "step": 9120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8333669197264526e-05,
+      "loss": 0.9264,
+      "step": 9121
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8323684093138145e-05,
+      "loss": 0.2167,
+      "step": 9122
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.831370109880145e-05,
+      "loss": 19.686,
+      "step": 9123
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.830372021491935e-05,
+      "loss": 0.1971,
+      "step": 9124
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.829374144215662e-05,
+      "loss": 0.9494,
+      "step": 9125
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.82837647811779e-05,
+      "loss": 0.9294,
+      "step": 9126
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.827379023264767e-05,
+      "loss": 0.968,
+      "step": 9127
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8263817797230286e-05,
+      "loss": 18.5186,
+      "step": 9128
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8253847475589965e-05,
+      "loss": 0.2259,
+      "step": 9129
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8243879268390767e-05,
+      "loss": 17.8564,
+      "step": 9130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8233913176296624e-05,
+      "loss": 0.93,
+      "step": 9131
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8223949199971324e-05,
+      "loss": 18.8893,
+      "step": 9132
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8213987340078538e-05,
+      "loss": 0.1735,
+      "step": 9133
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.820402759728171e-05,
+      "loss": 0.1845,
+      "step": 9134
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.819406997224426e-05,
+      "loss": 0.9546,
+      "step": 9135
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.81841144656294e-05,
+      "loss": 17.3026,
+      "step": 9136
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8174161078100205e-05,
+      "loss": 18.1951,
+      "step": 9137
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.816420981031962e-05,
+      "loss": 0.1947,
+      "step": 9138
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8154260662950452e-05,
+      "loss": 19.8184,
+      "step": 9139
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8144313636655368e-05,
+      "loss": 0.9,
+      "step": 9140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.813436873209684e-05,
+      "loss": 0.9971,
+      "step": 9141
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8124425949937297e-05,
+      "loss": 0.9258,
+      "step": 9142
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8114485290838967e-05,
+      "loss": 0.1921,
+      "step": 9143
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8104546755463903e-05,
+      "loss": 0.9869,
+      "step": 9144
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.80946103444741e-05,
+      "loss": 0.9618,
+      "step": 9145
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8084676058531373e-05,
+      "loss": 18.6504,
+      "step": 9146
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8074743898297342e-05,
+      "loss": 17.5397,
+      "step": 9147
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8064813864433588e-05,
+      "loss": 17.1525,
+      "step": 9148
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8054885957601492e-05,
+      "loss": 0.9251,
+      "step": 9149
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.804496017846226e-05,
+      "loss": 0.9566,
+      "step": 9150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8035036527677014e-05,
+      "loss": 20.2345,
+      "step": 9151
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8025115005906746e-05,
+      "loss": 0.218,
+      "step": 9152
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.801519561381223e-05,
+      "loss": 0.922,
+      "step": 9153
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.800527835205415e-05,
+      "loss": 22.5683,
+      "step": 9154
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.799536322129309e-05,
+      "loss": 18.9278,
+      "step": 9155
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7985450222189382e-05,
+      "loss": 17.5168,
+      "step": 9156
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.797553935540331e-05,
+      "loss": 19.3988,
+      "step": 9157
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7965630621594957e-05,
+      "loss": 0.202,
+      "step": 9158
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7955724021424354e-05,
+      "loss": 19.5733,
+      "step": 9159
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7945819555551264e-05,
+      "loss": 19.3215,
+      "step": 9160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.793591722463537e-05,
+      "loss": 16.8573,
+      "step": 9161
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.792601702933627e-05,
+      "loss": 0.1915,
+      "step": 9162
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.79161189703133e-05,
+      "loss": 18.9394,
+      "step": 9163
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.790622304822575e-05,
+      "loss": 0.9701,
+      "step": 9164
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.789632926373273e-05,
+      "loss": 18.8528,
+      "step": 9165
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7886437617493205e-05,
+      "loss": 18.8363,
+      "step": 9166
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7876548110166015e-05,
+      "loss": 0.9648,
+      "step": 9167
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7866660742409837e-05,
+      "loss": 19.4881,
+      "step": 9168
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7856775514883224e-05,
+      "loss": 19.7878,
+      "step": 9169
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7846892428244566e-05,
+      "loss": 0.9517,
+      "step": 9170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7837011483152132e-05,
+      "loss": 19.6857,
+      "step": 9171
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.782713268026404e-05,
+      "loss": 18.5718,
+      "step": 9172
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.781725602023826e-05,
+      "loss": 17.2627,
+      "step": 9173
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7807381503732625e-05,
+      "loss": 0.9482,
+      "step": 9174
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.779750913140482e-05,
+      "loss": 19.8944,
+      "step": 9175
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7787638903912395e-05,
+      "loss": 20.3861,
+      "step": 9176
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7777770821912758e-05,
+      "loss": 0.1897,
+      "step": 9177
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.776790488606316e-05,
+      "loss": 19.9223,
+      "step": 9178
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7758041097020716e-05,
+      "loss": 19.6502,
+      "step": 9179
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.774817945544241e-05,
+      "loss": 18.8366,
+      "step": 9180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7738319961985067e-05,
+      "loss": 0.1752,
+      "step": 9181
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7728462617305375e-05,
+      "loss": 19.9425,
+      "step": 9182
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771860742205988e-05,
+      "loss": 18.0991,
+      "step": 9183
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7708754376904986e-05,
+      "loss": 18.7278,
+      "step": 9184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769890348249695e-05,
+      "loss": 18.2237,
+      "step": 9185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768905473949189e-05,
+      "loss": 0.9432,
+      "step": 9186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7679208148545784e-05,
+      "loss": 18.1632,
+      "step": 9187
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7669363710314422e-05,
+      "loss": 0.1887,
+      "step": 9188
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7659521425453533e-05,
+      "loss": 18.3001,
+      "step": 9189
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7649681294618665e-05,
+      "loss": 0.9684,
+      "step": 9190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7639843318465165e-05,
+      "loss": 19.3323,
+      "step": 9191
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.763000749764833e-05,
+      "loss": 17.5129,
+      "step": 9192
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.762017383282328e-05,
+      "loss": 19.5036,
+      "step": 9193
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7610342324644928e-05,
+      "loss": 20.8545,
+      "step": 9194
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7600512973768153e-05,
+      "loss": 0.9239,
+      "step": 9195
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7590685780847626e-05,
+      "loss": 0.1816,
+      "step": 9196
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7580860746537865e-05,
+      "loss": 21.3986,
+      "step": 9197
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7571037871493246e-05,
+      "loss": 0.9044,
+      "step": 9198
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7561217156368066e-05,
+      "loss": 0.2463,
+      "step": 9199
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7551398601816422e-05,
+      "loss": 18.5511,
+      "step": 9200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7541582208492226e-05,
+      "loss": 20.3545,
+      "step": 9201
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7531767977049347e-05,
+      "loss": 18.6476,
+      "step": 9202
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7521955908141458e-05,
+      "loss": 17.9342,
+      "step": 9203
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7512146002422057e-05,
+      "loss": 0.2015,
+      "step": 9204
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7502338260544527e-05,
+      "loss": 0.8944,
+      "step": 9205
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7492532683162162e-05,
+      "loss": 0.9232,
+      "step": 9206
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7482729270927998e-05,
+      "loss": 0.2277,
+      "step": 9207
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7472928024495e-05,
+      "loss": 17.9534,
+      "step": 9208
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7463128944516027e-05,
+      "loss": 0.1967,
+      "step": 9209
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.745333203164369e-05,
+      "loss": 0.9568,
+      "step": 9210
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.74435372865305e-05,
+      "loss": 0.9271,
+      "step": 9211
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7433744709828893e-05,
+      "loss": 0.1633,
+      "step": 9212
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7423954302191047e-05,
+      "loss": 18.7633,
+      "step": 9213
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7414166064269073e-05,
+      "loss": 22.0784,
+      "step": 9214
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.74043799967149e-05,
+      "loss": 0.9048,
+      "step": 9215
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.739459610018033e-05,
+      "loss": 18.5317,
+      "step": 9216
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7384814375317014e-05,
+      "loss": 20.5774,
+      "step": 9217
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.737503482277647e-05,
+      "loss": 18.0661,
+      "step": 9218
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7365257443210054e-05,
+      "loss": 0.1995,
+      "step": 9219
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7355482237268983e-05,
+      "loss": 20.0802,
+      "step": 9220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.734570920560434e-05,
+      "loss": 20.5211,
+      "step": 9221
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.733593834886704e-05,
+      "loss": 0.9866,
+      "step": 9222
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7326169667707882e-05,
+      "loss": 19.794,
+      "step": 9223
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7316403162777493e-05,
+      "loss": 18.5229,
+      "step": 9224
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.730663883472638e-05,
+      "loss": 0.2149,
+      "step": 9225
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.729687668420488e-05,
+      "loss": 0.2035,
+      "step": 9226
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.728711671186321e-05,
+      "loss": 0.952,
+      "step": 9227
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7277358918351427e-05,
+      "loss": 21.8369,
+      "step": 9228
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7267603304319435e-05,
+      "loss": 18.2806,
+      "step": 9229
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.725784987041701e-05,
+      "loss": 19.8615,
+      "step": 9230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.724809861729378e-05,
+      "loss": 18.8258,
+      "step": 9231
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7238349545599215e-05,
+      "loss": 0.9048,
+      "step": 9232
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7228602655982652e-05,
+      "loss": 20.5813,
+      "step": 9233
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7218857949093287e-05,
+      "loss": 17.8953,
+      "step": 9234
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7209115425580146e-05,
+      "loss": 0.2028,
+      "step": 9235
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7199375086092146e-05,
+      "loss": 18.6164,
+      "step": 9236
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7189636931278042e-05,
+      "loss": 18.6521,
+      "step": 9237
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7179900961786387e-05,
+      "loss": 20.0255,
+      "step": 9238
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.717016717826571e-05,
+      "loss": 18.0271,
+      "step": 9239
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.716043558136431e-05,
+      "loss": 0.1928,
+      "step": 9240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7150706171730307e-05,
+      "loss": 22.3005,
+      "step": 9241
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7140978950011788e-05,
+      "loss": 0.2025,
+      "step": 9242
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.71312539168566e-05,
+      "loss": 0.1973,
+      "step": 9243
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.71215310729125e-05,
+      "loss": 17.2114,
+      "step": 9244
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7111810418827016e-05,
+      "loss": 19.6128,
+      "step": 9245
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.710209195524765e-05,
+      "loss": 21.427,
+      "step": 9246
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7092375682821693e-05,
+      "loss": 0.2008,
+      "step": 9247
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7082661602196242e-05,
+      "loss": 16.3367,
+      "step": 9248
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.707294971401835e-05,
+      "loss": 17.6323,
+      "step": 9249
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.706324001893488e-05,
+      "loss": 0.2246,
+      "step": 9250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7053532517592475e-05,
+      "loss": 18.4474,
+      "step": 9251
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.704382721063777e-05,
+      "loss": 19.6731,
+      "step": 9252
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7034124098717176e-05,
+      "loss": 21.0136,
+      "step": 9253
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7024423182476923e-05,
+      "loss": 18.0718,
+      "step": 9254
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.701472446256315e-05,
+      "loss": 19.1259,
+      "step": 9255
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7005027939621886e-05,
+      "loss": 21.0132,
+      "step": 9256
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.699533361429891e-05,
+      "loss": 19.5382,
+      "step": 9257
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.69856414872399e-05,
+      "loss": 18.4641,
+      "step": 9258
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6975951559090464e-05,
+      "loss": 16.5978,
+      "step": 9259
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6966263830495936e-05,
+      "loss": 18.5132,
+      "step": 9260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6956578302101585e-05,
+      "loss": 0.9604,
+      "step": 9261
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6946894974552506e-05,
+      "loss": 18.0501,
+      "step": 9262
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.693721384849366e-05,
+      "loss": 21.3802,
+      "step": 9263
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6927534924569855e-05,
+      "loss": 0.194,
+      "step": 9264
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.691785820342573e-05,
+      "loss": 18.1496,
+      "step": 9265
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.690818368570585e-05,
+      "loss": 18.284,
+      "step": 9266
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6898511372054537e-05,
+      "loss": 19.2205,
+      "step": 9267
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.688884126311603e-05,
+      "loss": 0.2153,
+      "step": 9268
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.68791733595344e-05,
+      "loss": 20.0383,
+      "step": 9269
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.686950766195357e-05,
+      "loss": 18.3409,
+      "step": 9270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6859844171017332e-05,
+      "loss": 0.1948,
+      "step": 9271
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6850182887369304e-05,
+      "loss": 19.3814,
+      "step": 9272
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.684052381165298e-05,
+      "loss": 19.1863,
+      "step": 9273
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6830866944511703e-05,
+      "loss": 19.3633,
+      "step": 9274
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.682121228658866e-05,
+      "loss": 20.5249,
+      "step": 9275
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.68115598385269e-05,
+      "loss": 0.9668,
+      "step": 9276
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6801909600969324e-05,
+      "loss": 0.1931,
+      "step": 9277
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6792261574558672e-05,
+      "loss": 18.6234,
+      "step": 9278
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.678261575993755e-05,
+      "loss": 19.0015,
+      "step": 9279
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6772972157748418e-05,
+      "loss": 19.2353,
+      "step": 9280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6763330768633585e-05,
+      "loss": 0.9384,
+      "step": 9281
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.675369159323521e-05,
+      "loss": 18.4446,
+      "step": 9282
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.674405463219531e-05,
+      "loss": 0.1799,
+      "step": 9283
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6734419886155745e-05,
+      "loss": 17.9931,
+      "step": 9284
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6724787355758236e-05,
+      "loss": 20.3976,
+      "step": 9285
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6715157041644363e-05,
+      "loss": 21.14,
+      "step": 9286
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.670552894445554e-05,
+      "loss": 0.2284,
+      "step": 9287
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.669590306483304e-05,
+      "loss": 19.804,
+      "step": 9288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6686279403418004e-05,
+      "loss": 0.1826,
+      "step": 9289
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.66766579608514e-05,
+      "loss": 0.2342,
+      "step": 9290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6667038737774088e-05,
+      "loss": 1.0024,
+      "step": 9291
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.665742173482669e-05,
+      "loss": 0.2074,
+      "step": 9292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6647806952649807e-05,
+      "loss": 18.5372,
+      "step": 9293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.663819439188382e-05,
+      "loss": 18.5546,
+      "step": 9294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.662858405316892e-05,
+      "loss": 0.8755,
+      "step": 9295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6618975937145255e-05,
+      "loss": 0.1893,
+      "step": 9296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6609370044452765e-05,
+      "loss": 19.8143,
+      "step": 9297
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.65997663757312e-05,
+      "loss": 0.1648,
+      "step": 9298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6590164931620257e-05,
+      "loss": 20.3004,
+      "step": 9299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6580565712759437e-05,
+      "loss": 19.7779,
+      "step": 9300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6570968719788067e-05,
+      "loss": 18.3496,
+      "step": 9301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.656137395334534e-05,
+      "loss": 0.2242,
+      "step": 9302
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6551781414070367e-05,
+      "loss": 17.872,
+      "step": 9303
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6542191102602e-05,
+      "loss": 0.9646,
+      "step": 9304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6532603019579007e-05,
+      "loss": 19.1916,
+      "step": 9305
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.652301716564003e-05,
+      "loss": 0.9217,
+      "step": 9306
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6513433541423528e-05,
+      "loss": 19.2517,
+      "step": 9307
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6503852147567785e-05,
+      "loss": 0.2006,
+      "step": 9308
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6494272984710962e-05,
+      "loss": 0.1782,
+      "step": 9309
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6484696053491134e-05,
+      "loss": 18.3773,
+      "step": 9310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.647512135454611e-05,
+      "loss": 16.6461,
+      "step": 9311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.646554888851361e-05,
+      "loss": 18.6422,
+      "step": 9312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6455978656031262e-05,
+      "loss": 0.9399,
+      "step": 9313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6446410657736428e-05,
+      "loss": 0.8894,
+      "step": 9314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.643684489426639e-05,
+      "loss": 18.1191,
+      "step": 9315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.642728136625832e-05,
+      "loss": 0.9476,
+      "step": 9316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6417720074349142e-05,
+      "loss": 0.1757,
+      "step": 9317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.64081610191757e-05,
+      "loss": 0.2093,
+      "step": 9318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6398604201374667e-05,
+      "loss": 20.3706,
+      "step": 9319
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6389049621582574e-05,
+      "loss": 0.9205,
+      "step": 9320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6379497280435807e-05,
+      "loss": 0.1764,
+      "step": 9321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6369947178570587e-05,
+      "loss": 18.536,
+      "step": 9322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6360399316623e-05,
+      "loss": 20.366,
+      "step": 9323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.635085369522898e-05,
+      "loss": 18.3706,
+      "step": 9324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.63413103150243e-05,
+      "loss": 18.3549,
+      "step": 9325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6331769176644608e-05,
+      "loss": 0.8814,
+      "step": 9326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.632223028072537e-05,
+      "loss": 20.4392,
+      "step": 9327
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6312693627901937e-05,
+      "loss": 0.9511,
+      "step": 9328
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.630315921880949e-05,
+      "loss": 0.9586,
+      "step": 9329
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.629362705408306e-05,
+      "loss": 18.0369,
+      "step": 9330
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6284097134357536e-05,
+      "loss": 0.9684,
+      "step": 9331
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6274569460267658e-05,
+      "loss": 20.4672,
+      "step": 9332
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6265044032448007e-05,
+      "loss": 19.5792,
+      "step": 9333
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.625552085153303e-05,
+      "loss": 0.1754,
+      "step": 9334
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6245999918157e-05,
+      "loss": 18.8123,
+      "step": 9335
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.623648123295407e-05,
+      "loss": 0.9436,
+      "step": 9336
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6226964796558224e-05,
+      "loss": 18.0263,
+      "step": 9337
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6217450609603302e-05,
+      "loss": 19.9959,
+      "step": 9338
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6207938672722987e-05,
+      "loss": 18.0555,
+      "step": 9339
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6198428986550828e-05,
+      "loss": 0.9245,
+      "step": 9340
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.618892155172022e-05,
+      "loss": 0.9598,
+      "step": 9341
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6179416368864363e-05,
+      "loss": 0.2012,
+      "step": 9342
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6169913438616386e-05,
+      "loss": 18.1798,
+      "step": 9343
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6160412761609232e-05,
+      "loss": 18.7937,
+      "step": 9344
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6150914338475646e-05,
+      "loss": 0.1981,
+      "step": 9345
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6141418169848317e-05,
+      "loss": 0.9418,
+      "step": 9346
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6131924256359722e-05,
+      "loss": 0.9541,
+      "step": 9347
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6122432598642173e-05,
+      "loss": 0.9278,
+      "step": 9348
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6112943197327862e-05,
+      "loss": 0.9292,
+      "step": 9349
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6103456053048856e-05,
+      "loss": 20.2108,
+      "step": 9350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6093971166437044e-05,
+      "loss": 17.572,
+      "step": 9351
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6084488538124116e-05,
+      "loss": 0.2201,
+      "step": 9352
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6075008168741705e-05,
+      "loss": 0.9884,
+      "step": 9353
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.606553005892125e-05,
+      "loss": 20.7942,
+      "step": 9354
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6056054209293992e-05,
+      "loss": 20.0276,
+      "step": 9355
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6046580620491107e-05,
+      "loss": 20.2209,
+      "step": 9356
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.603710929314359e-05,
+      "loss": 19.0687,
+      "step": 9357
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.602764022788223e-05,
+      "loss": 20.7151,
+      "step": 9358
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6018173425337718e-05,
+      "loss": 0.9585,
+      "step": 9359
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6008708886140635e-05,
+      "loss": 0.1978,
+      "step": 9360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5999246610921314e-05,
+      "loss": 0.2014,
+      "step": 9361
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5989786600309992e-05,
+      "loss": 19.0126,
+      "step": 9362
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5980328854936795e-05,
+      "loss": 0.1781,
+      "step": 9363
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5970873375431595e-05,
+      "loss": 0.2144,
+      "step": 9364
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.59614201624242e-05,
+      "loss": 19.7554,
+      "step": 9365
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.595196921654423e-05,
+      "loss": 0.9348,
+      "step": 9366
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5942520538421162e-05,
+      "loss": 20.6755,
+      "step": 9367
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5933074128684332e-05,
+      "loss": 20.0952,
+      "step": 9368
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5923629987962886e-05,
+      "loss": 0.8758,
+      "step": 9369
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5914188116885903e-05,
+      "loss": 18.4781,
+      "step": 9370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.59047485160822e-05,
+      "loss": 19.9674,
+      "step": 9371
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5895311186180522e-05,
+      "loss": 0.2146,
+      "step": 9372
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.588587612780944e-05,
+      "loss": 0.1996,
+      "step": 9373
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5876443341597363e-05,
+      "loss": 20.4405,
+      "step": 9374
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.586701282817257e-05,
+      "loss": 0.9352,
+      "step": 9375
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.585758458816317e-05,
+      "loss": 0.9316,
+      "step": 9376
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.584815862219713e-05,
+      "loss": 20.0873,
+      "step": 9377
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5838734930902254e-05,
+      "loss": 0.2067,
+      "step": 9378
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.582931351490622e-05,
+      "loss": 0.9543,
+      "step": 9379
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5819894374836526e-05,
+      "loss": 17.0908,
+      "step": 9380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5810477511320533e-05,
+      "loss": 0.1978,
+      "step": 9381
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.580106292498545e-05,
+      "loss": 0.9225,
+      "step": 9382
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.579165061645833e-05,
+      "loss": 1.0147,
+      "step": 9383
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.578224058636608e-05,
+      "loss": 0.2152,
+      "step": 9384
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.577283283533545e-05,
+      "loss": 20.7043,
+      "step": 9385
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5763427363993032e-05,
+      "loss": 0.9383,
+      "step": 9386
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5754024172965288e-05,
+      "loss": 19.2348,
+      "step": 9387
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5744623262878523e-05,
+      "loss": 17.9613,
+      "step": 9388
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.573522463435883e-05,
+      "loss": 0.1897,
+      "step": 9389
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5725828288032257e-05,
+      "loss": 0.9203,
+      "step": 9390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5716434224524616e-05,
+      "loss": 0.9563,
+      "step": 9391
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5707042444461612e-05,
+      "loss": 19.7026,
+      "step": 9392
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.569765294846877e-05,
+      "loss": 0.1937,
+      "step": 9393
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5688265737171477e-05,
+      "loss": 16.9619,
+      "step": 9394
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.567888081119498e-05,
+      "loss": 19.9122,
+      "step": 9395
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.566949817116431e-05,
+      "loss": 17.9005,
+      "step": 9396
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.566011781770444e-05,
+      "loss": 0.209,
+      "step": 9397
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5650739751440152e-05,
+      "loss": 0.9207,
+      "step": 9398
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5641363972996014e-05,
+      "loss": 0.155,
+      "step": 9399
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5631990482996555e-05,
+      "loss": 17.5294,
+      "step": 9400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5622619282066082e-05,
+      "loss": 17.7508,
+      "step": 9401
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5613250370828715e-05,
+      "loss": 20.7957,
+      "step": 9402
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.560388374990851e-05,
+      "loss": 0.9559,
+      "step": 9403
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5594519419929343e-05,
+      "loss": 19.6068,
+      "step": 9404
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5585157381514875e-05,
+      "loss": 19.6404,
+      "step": 9405
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.557579763528867e-05,
+      "loss": 0.8708,
+      "step": 9406
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.556644018187417e-05,
+      "loss": 0.8964,
+      "step": 9407
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5557085021894585e-05,
+      "loss": 0.205,
+      "step": 9408
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5547732155973004e-05,
+      "loss": 0.9,
+      "step": 9409
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5538381584732432e-05,
+      "loss": 17.9454,
+      "step": 9410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.55290333087956e-05,
+      "loss": 20.394,
+      "step": 9411
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.551968732878517e-05,
+      "loss": 0.1851,
+      "step": 9412
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5510343645323605e-05,
+      "loss": 20.9156,
+      "step": 9413
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5501002259033292e-05,
+      "loss": 17.094,
+      "step": 9414
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5491663170536364e-05,
+      "loss": 18.8261,
+      "step": 9415
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5482326380454837e-05,
+      "loss": 19.7769,
+      "step": 9416
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5472991889410644e-05,
+      "loss": 19.4596,
+      "step": 9417
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5463659698025452e-05,
+      "loss": 19.5035,
+      "step": 9418
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5454329806920826e-05,
+      "loss": 0.1768,
+      "step": 9419
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5445002216718236e-05,
+      "loss": 0.1815,
+      "step": 9420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5435676928038885e-05,
+      "loss": 0.9628,
+      "step": 9421
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5426353941503907e-05,
+      "loss": 18.7071,
+      "step": 9422
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.541703325773425e-05,
+      "loss": 17.4381,
+      "step": 9423
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.540771487735071e-05,
+      "loss": 17.8426,
+      "step": 9424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5398398800973944e-05,
+      "loss": 20.6553,
+      "step": 9425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5389085029224443e-05,
+      "loss": 0.224,
+      "step": 9426
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5379773562722545e-05,
+      "loss": 17.6575,
+      "step": 9427
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5370464402088437e-05,
+      "loss": 0.9489,
+      "step": 9428
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.536115754794215e-05,
+      "loss": 19.4554,
+      "step": 9429
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.535185300090357e-05,
+      "loss": 0.2398,
+      "step": 9430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5342550761592423e-05,
+      "loss": 0.9347,
+      "step": 9431
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5333250830628275e-05,
+      "loss": 19.68,
+      "step": 9432
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5323953208630553e-05,
+      "loss": 19.6706,
+      "step": 9433
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5314657896218514e-05,
+      "loss": 0.2578,
+      "step": 9434
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.530536489401127e-05,
+      "loss": 0.1846,
+      "step": 9435
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5296074202627787e-05,
+      "loss": 0.9393,
+      "step": 9436
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5286785822686855e-05,
+      "loss": 17.5133,
+      "step": 9437
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5277499754807127e-05,
+      "loss": 19.1771,
+      "step": 9438
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5268215999607107e-05,
+      "loss": 20.1617,
+      "step": 9439
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5258934557705124e-05,
+      "loss": 0.9732,
+      "step": 9440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5249655429719373e-05,
+      "loss": 0.2211,
+      "step": 9441
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5240378616267886e-05,
+      "loss": 0.195,
+      "step": 9442
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5231104117968542e-05,
+      "loss": 0.8866,
+      "step": 9443
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5221831935439062e-05,
+      "loss": 0.9139,
+      "step": 9444
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5212562069297031e-05,
+      "loss": 20.1547,
+      "step": 9445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5203294520159827e-05,
+      "loss": 0.1517,
+      "step": 9446
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.519402928864475e-05,
+      "loss": 0.9669,
+      "step": 9447
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5184766375368915e-05,
+      "loss": 18.7429,
+      "step": 9448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5175505780949223e-05,
+      "loss": 0.2096,
+      "step": 9449
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5166247506002518e-05,
+      "loss": 19.0357,
+      "step": 9450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5156991551145444e-05,
+      "loss": 20.2473,
+      "step": 9451
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5147737916994464e-05,
+      "loss": 0.8964,
+      "step": 9452
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5138486604165914e-05,
+      "loss": 19.1622,
+      "step": 9453
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5129237613275998e-05,
+      "loss": 20.3952,
+      "step": 9454
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5119990944940749e-05,
+      "loss": 0.9292,
+      "step": 9455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5110746599775988e-05,
+      "loss": 0.9395,
+      "step": 9456
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5101504578397473e-05,
+      "loss": 18.7208,
+      "step": 9457
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5092264881420775e-05,
+      "loss": 0.9369,
+      "step": 9458
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5083027509461268e-05,
+      "loss": 20.3226,
+      "step": 9459
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5073792463134196e-05,
+      "loss": 0.2288,
+      "step": 9460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5064559743054712e-05,
+      "loss": 0.9204,
+      "step": 9461
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5055329349837705e-05,
+      "loss": 0.986,
+      "step": 9462
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5046101284097968e-05,
+      "loss": 0.1928,
+      "step": 9463
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5036875546450173e-05,
+      "loss": 20.7491,
+      "step": 9464
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5027652137508752e-05,
+      "loss": 0.9362,
+      "step": 9465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5018431057888027e-05,
+      "loss": 0.9285,
+      "step": 9466
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.500921230820222e-05,
+      "loss": 18.1752,
+      "step": 9467
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4999995889065282e-05,
+      "loss": 0.196,
+      "step": 9468
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4990781801091092e-05,
+      "loss": 20.1167,
+      "step": 9469
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4981570044893351e-05,
+      "loss": 0.195,
+      "step": 9470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4972360621085602e-05,
+      "loss": 0.2051,
+      "step": 9471
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4963153530281244e-05,
+      "loss": 18.1977,
+      "step": 9472
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4953948773093495e-05,
+      "loss": 18.9315,
+      "step": 9473
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4944746350135452e-05,
+      "loss": 18.3304,
+      "step": 9474
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4935546262020029e-05,
+      "loss": 0.9606,
+      "step": 9475
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.492634850936e-05,
+      "loss": 19.307,
+      "step": 9476
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4917153092767977e-05,
+      "loss": 18.5235,
+      "step": 9477
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.490796001285641e-05,
+      "loss": 0.2154,
+      "step": 9478
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.489876927023761e-05,
+      "loss": 0.8993,
+      "step": 9479
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4889580865523723e-05,
+      "loss": 0.9312,
+      "step": 9480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.488039479932673e-05,
+      "loss": 18.5489,
+      "step": 9481
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.487121107225848e-05,
+      "loss": 18.72,
+      "step": 9482
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.486202968493064e-05,
+      "loss": 19.4549,
+      "step": 9483
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4852850637954734e-05,
+      "loss": 0.9701,
+      "step": 9484
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4843673931942131e-05,
+      "loss": 19.7561,
+      "step": 9485
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.483449956750405e-05,
+      "loss": 0.2183,
+      "step": 9486
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4825327545251533e-05,
+      "loss": 18.4932,
+      "step": 9487
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4816157865795487e-05,
+      "loss": 0.2151,
+      "step": 9488
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4806990529746662e-05,
+      "loss": 20.2042,
+      "step": 9489
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4797825537715627e-05,
+      "loss": 19.3053,
+      "step": 9490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4788662890312826e-05,
+      "loss": 19.5038,
+      "step": 9491
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4779502588148547e-05,
+      "loss": 19.5191,
+      "step": 9492
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.477034463183286e-05,
+      "loss": 19.0772,
+      "step": 9493
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.476118902197578e-05,
+      "loss": 0.1677,
+      "step": 9494
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4752035759187106e-05,
+      "loss": 18.656,
+      "step": 9495
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4742884844076443e-05,
+      "loss": 18.4478,
+      "step": 9496
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4733736277253334e-05,
+      "loss": 19.6368,
+      "step": 9497
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4724590059327098e-05,
+      "loss": 0.9215,
+      "step": 9498
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4715446190906929e-05,
+      "loss": 0.9399,
+      "step": 9499
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4706304672601812e-05,
+      "loss": 0.9629,
+      "step": 9500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4697165505020648e-05,
+      "loss": 0.2006,
+      "step": 9501
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.468802868877216e-05,
+      "loss": 19.5139,
+      "step": 9502
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4678894224464857e-05,
+      "loss": 19.3581,
+      "step": 9503
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4669762112707169e-05,
+      "loss": 0.9118,
+      "step": 9504
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.466063235410735e-05,
+      "loss": 21.7875,
+      "step": 9505
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4651504949273437e-05,
+      "loss": 18.4319,
+      "step": 9506
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4642379898813403e-05,
+      "loss": 20.0687,
+      "step": 9507
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4633257203335011e-05,
+      "loss": 20.9238,
+      "step": 9508
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4624136863445848e-05,
+      "loss": 0.9715,
+      "step": 9509
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4615018879753378e-05,
+      "loss": 19.5027,
+      "step": 9510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4605903252864939e-05,
+      "loss": 18.5975,
+      "step": 9511
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4596789983387632e-05,
+      "loss": 18.3835,
+      "step": 9512
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4587679071928439e-05,
+      "loss": 0.1808,
+      "step": 9513
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4578570519094243e-05,
+      "loss": 16.7862,
+      "step": 9514
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4569464325491661e-05,
+      "loss": 20.7219,
+      "step": 9515
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4560360491727231e-05,
+      "loss": 19.3358,
+      "step": 9516
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4551259018407304e-05,
+      "loss": 19.0608,
+      "step": 9517
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.454215990613808e-05,
+      "loss": 0.9032,
+      "step": 9518
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4533063155525612e-05,
+      "loss": 0.1855,
+      "step": 9519
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4523968767175754e-05,
+      "loss": 0.926,
+      "step": 9520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4514876741694295e-05,
+      "loss": 18.2626,
+      "step": 9521
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4505787079686756e-05,
+      "loss": 20.0101,
+      "step": 9522
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4496699781758549e-05,
+      "loss": 19.0501,
+      "step": 9523
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.448761484851498e-05,
+      "loss": 20.3977,
+      "step": 9524
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4478532280561091e-05,
+      "loss": 0.9795,
+      "step": 9525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4469452078501854e-05,
+      "loss": 19.8612,
+      "step": 9526
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4460374242942038e-05,
+      "loss": 17.7683,
+      "step": 9527
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4451298774486282e-05,
+      "loss": 19.9629,
+      "step": 9528
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4442225673739046e-05,
+      "loss": 19.7125,
+      "step": 9529
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4433154941304638e-05,
+      "loss": 0.935,
+      "step": 9530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4424086577787221e-05,
+      "loss": 19.5268,
+      "step": 9531
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4415020583790784e-05,
+      "loss": 19.9939,
+      "step": 9532
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4405956959919165e-05,
+      "loss": 19.7676,
+      "step": 9533
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4396895706776042e-05,
+      "loss": 17.5711,
+      "step": 9534
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.438783682496494e-05,
+      "loss": 0.9035,
+      "step": 9535
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4378780315089225e-05,
+      "loss": 0.9302,
+      "step": 9536
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.436972617775209e-05,
+      "loss": 0.1946,
+      "step": 9537
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4360674413556591e-05,
+      "loss": 0.9278,
+      "step": 9538
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4351625023105626e-05,
+      "loss": 0.1936,
+      "step": 9539
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4342578007001905e-05,
+      "loss": 21.0842,
+      "step": 9540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4333533365848024e-05,
+      "loss": 19.7785,
+      "step": 9541
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4324491100246385e-05,
+      "loss": 0.8978,
+      "step": 9542
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4315451210799246e-05,
+      "loss": 19.4585,
+      "step": 9543
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4306413698108712e-05,
+      "loss": 0.9425,
+      "step": 9544
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4297378562776714e-05,
+      "loss": 18.1155,
+      "step": 9545
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.428834580540504e-05,
+      "loss": 17.8125,
+      "step": 9546
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4279315426595313e-05,
+      "loss": 0.1789,
+      "step": 9547
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4270287426948997e-05,
+      "loss": 19.9741,
+      "step": 9548
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4261261807067417e-05,
+      "loss": 18.3803,
+      "step": 9549
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4252238567551668e-05,
+      "loss": 17.5196,
+      "step": 9550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4243217709002794e-05,
+      "loss": 20.0104,
+      "step": 9551
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4234199232021622e-05,
+      "loss": 0.1548,
+      "step": 9552
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4225183137208776e-05,
+      "loss": 0.9789,
+      "step": 9553
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4216169425164821e-05,
+      "loss": 19.8339,
+      "step": 9554
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4207158096490109e-05,
+      "loss": 21.2407,
+      "step": 9555
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4198149151784807e-05,
+      "loss": 18.7429,
+      "step": 9556
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4189142591648952e-05,
+      "loss": 20.0113,
+      "step": 9557
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.418013841668247e-05,
+      "loss": 18.7489,
+      "step": 9558
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4171136627485032e-05,
+      "loss": 0.9268,
+      "step": 9559
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4162137224656196e-05,
+      "loss": 18.5933,
+      "step": 9560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4153140208795401e-05,
+      "loss": 18.4917,
+      "step": 9561
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4144145580501895e-05,
+      "loss": 0.1936,
+      "step": 9562
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4135153340374718e-05,
+      "loss": 19.6161,
+      "step": 9563
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.41261634890128e-05,
+      "loss": 18.5179,
+      "step": 9564
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4117176027014956e-05,
+      "loss": 19.3603,
+      "step": 9565
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4108190954979743e-05,
+      "loss": 19.115,
+      "step": 9566
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4099208273505615e-05,
+      "loss": 19.6744,
+      "step": 9567
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4090227983190902e-05,
+      "loss": 0.9797,
+      "step": 9568
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4081250084633684e-05,
+      "loss": 18.293,
+      "step": 9569
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4072274578431932e-05,
+      "loss": 19.9685,
+      "step": 9570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4063301465183504e-05,
+      "loss": 0.2292,
+      "step": 9571
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4054330745486005e-05,
+      "loss": 0.1662,
+      "step": 9572
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4045362419936942e-05,
+      "loss": 18.9564,
+      "step": 9573
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4036396489133647e-05,
+      "loss": 19.131,
+      "step": 9574
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4027432953673292e-05,
+      "loss": 19.6586,
+      "step": 9575
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4018471814152896e-05,
+      "loss": 17.5029,
+      "step": 9576
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.40095130711693e-05,
+      "loss": 0.2165,
+      "step": 9577
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4000556725319208e-05,
+      "loss": 0.9394,
+      "step": 9578
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3991602777199147e-05,
+      "loss": 20.2511,
+      "step": 9579
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3982651227405497e-05,
+      "loss": 19.0913,
+      "step": 9580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3973702076534474e-05,
+      "loss": 0.9367,
+      "step": 9581
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3964755325182127e-05,
+      "loss": 0.2112,
+      "step": 9582
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3955810973944355e-05,
+      "loss": 0.9439,
+      "step": 9583
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3946869023416897e-05,
+      "loss": 0.1876,
+      "step": 9584
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.393792947419532e-05,
+      "loss": 18.6427,
+      "step": 9585
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3928992326875045e-05,
+      "loss": 0.2283,
+      "step": 9586
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3920057582051327e-05,
+      "loss": 19.5802,
+      "step": 9587
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3911125240319261e-05,
+      "loss": 0.1955,
+      "step": 9588
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3902195302273779e-05,
+      "loss": 0.9341,
+      "step": 9589
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.389326776850966e-05,
+      "loss": 19.3107,
+      "step": 9590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3884342639621523e-05,
+      "loss": 0.9106,
+      "step": 9591
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3875419916203814e-05,
+      "loss": 19.4249,
+      "step": 9592
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3866499598850836e-05,
+      "loss": 0.1826,
+      "step": 9593
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3857581688156713e-05,
+      "loss": 18.4384,
+      "step": 9594
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3848666184715431e-05,
+      "loss": 0.9124,
+      "step": 9595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3839753089120816e-05,
+      "loss": 0.1941,
+      "step": 9596
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3830842401966471e-05,
+      "loss": 19.7236,
+      "step": 9597
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3821934123845936e-05,
+      "loss": 19.2309,
+      "step": 9598
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.381302825535255e-05,
+      "loss": 17.958,
+      "step": 9599
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3804124797079438e-05,
+      "loss": 20.1773,
+      "step": 9600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3795223749619646e-05,
+      "loss": 20.7665,
+      "step": 9601
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.378632511356604e-05,
+      "loss": 20.4517,
+      "step": 9602
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3777428889511273e-05,
+      "loss": 0.2137,
+      "step": 9603
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3768535078047873e-05,
+      "loss": 0.1951,
+      "step": 9604
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3759643679768242e-05,
+      "loss": 18.3753,
+      "step": 9605
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3750754695264589e-05,
+      "loss": 20.0987,
+      "step": 9606
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.374186812512892e-05,
+      "loss": 0.9297,
+      "step": 9607
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3732983969953156e-05,
+      "loss": 19.3944,
+      "step": 9608
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3724102230329033e-05,
+      "loss": 0.9424,
+      "step": 9609
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3715222906848068e-05,
+      "loss": 19.4364,
+      "step": 9610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.370634600010171e-05,
+      "loss": 0.9177,
+      "step": 9611
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3697471510681198e-05,
+      "loss": 0.9402,
+      "step": 9612
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3688599439177591e-05,
+      "loss": 21.5088,
+      "step": 9613
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.367972978618181e-05,
+      "loss": 1.0031,
+      "step": 9614
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3670862552284658e-05,
+      "loss": 0.2036,
+      "step": 9615
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3661997738076681e-05,
+      "loss": 0.2017,
+      "step": 9616
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3653135344148332e-05,
+      "loss": 0.9499,
+      "step": 9617
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.364427537108992e-05,
+      "loss": 0.9529,
+      "step": 9618
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.363541781949152e-05,
+      "loss": 20.247,
+      "step": 9619
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3626562689943107e-05,
+      "loss": 18.8793,
+      "step": 9620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.361770998303446e-05,
+      "loss": 18.6326,
+      "step": 9621
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3608859699355225e-05,
+      "loss": 0.9761,
+      "step": 9622
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.360001183949486e-05,
+      "loss": 19.2071,
+      "step": 9623
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3591166404042666e-05,
+      "loss": 19.1329,
+      "step": 9624
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3582323393587832e-05,
+      "loss": 0.1651,
+      "step": 9625
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3573482808719301e-05,
+      "loss": 0.9586,
+      "step": 9626
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3564644650025893e-05,
+      "loss": 0.9595,
+      "step": 9627
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3555808918096318e-05,
+      "loss": 0.9291,
+      "step": 9628
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3546975613519036e-05,
+      "loss": 0.1957,
+      "step": 9629
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3538144736882391e-05,
+      "loss": 18.6887,
+      "step": 9630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3529316288774568e-05,
+      "loss": 0.2099,
+      "step": 9631
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3520490269783581e-05,
+      "loss": 19.1129,
+      "step": 9632
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3511666680497286e-05,
+      "loss": 0.9511,
+      "step": 9633
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3502845521503371e-05,
+      "loss": 19.6339,
+      "step": 9634
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3494026793389364e-05,
+      "loss": 0.9673,
+      "step": 9635
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.348521049674264e-05,
+      "loss": 19.1559,
+      "step": 9636
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3476396632150402e-05,
+      "loss": 0.9305,
+      "step": 9637
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3467585200199695e-05,
+      "loss": 0.9593,
+      "step": 9638
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3458776201477396e-05,
+      "loss": 0.2111,
+      "step": 9639
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3449969636570237e-05,
+      "loss": 0.9528,
+      "step": 9640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3441165506064763e-05,
+      "loss": 19.4766,
+      "step": 9641
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3432363810547383e-05,
+      "loss": 20.7167,
+      "step": 9642
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3423564550604334e-05,
+      "loss": 0.2054,
+      "step": 9643
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3414767726821642e-05,
+      "loss": 0.1763,
+      "step": 9644
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3405973339785277e-05,
+      "loss": 0.2084,
+      "step": 9645
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3397181390080953e-05,
+      "loss": 0.9529,
+      "step": 9646
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3388391878294266e-05,
+      "loss": 0.1899,
+      "step": 9647
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3379604805010632e-05,
+      "loss": 18.573,
+      "step": 9648
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3370820170815318e-05,
+      "loss": 0.2028,
+      "step": 9649
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3362037976293412e-05,
+      "loss": 19.9752,
+      "step": 9650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3353258222029858e-05,
+      "loss": 0.9099,
+      "step": 9651
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.334448090860942e-05,
+      "loss": 19.8581,
+      "step": 9652
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.333570603661673e-05,
+      "loss": 19.472,
+      "step": 9653
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3326933606636182e-05,
+      "loss": 0.9295,
+      "step": 9654
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3318163619252116e-05,
+      "loss": 0.9521,
+      "step": 9655
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3309396075048646e-05,
+      "loss": 0.2184,
+      "step": 9656
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.330063097460969e-05,
+      "loss": 17.336,
+      "step": 9657
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.329186831851909e-05,
+      "loss": 0.9265,
+      "step": 9658
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3283108107360476e-05,
+      "loss": 0.9359,
+      "step": 9659
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.327435034171729e-05,
+      "loss": 18.1446,
+      "step": 9660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.326559502217284e-05,
+      "loss": 0.923,
+      "step": 9661
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3256842149310317e-05,
+      "loss": 20.483,
+      "step": 9662
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3248091723712653e-05,
+      "loss": 18.3175,
+      "step": 9663
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3239343745962679e-05,
+      "loss": 0.957,
+      "step": 9664
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.323059821664308e-05,
+      "loss": 0.99,
+      "step": 9665
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3221855136336319e-05,
+      "loss": 0.1852,
+      "step": 9666
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.321311450562473e-05,
+      "loss": 0.9473,
+      "step": 9667
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3204376325090467e-05,
+      "loss": 18.0747,
+      "step": 9668
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3195640595315584e-05,
+      "loss": 17.5878,
+      "step": 9669
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.318690731688187e-05,
+      "loss": 0.8929,
+      "step": 9670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3178176490371003e-05,
+      "loss": 1.0079,
+      "step": 9671
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3169448116364542e-05,
+      "loss": 0.209,
+      "step": 9672
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3160722195443792e-05,
+      "loss": 19.7638,
+      "step": 9673
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.315199872818994e-05,
+      "loss": 0.9665,
+      "step": 9674
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3143277715184055e-05,
+      "loss": 18.196,
+      "step": 9675
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3134559157006948e-05,
+      "loss": 19.0817,
+      "step": 9676
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.312584305423934e-05,
+      "loss": 0.939,
+      "step": 9677
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.311712940746176e-05,
+      "loss": 22.0026,
+      "step": 9678
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.310841821725457e-05,
+      "loss": 0.1779,
+      "step": 9679
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.309970948419798e-05,
+      "loss": 19.2857,
+      "step": 9680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3091003208872039e-05,
+      "loss": 18.5597,
+      "step": 9681
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3082299391856612e-05,
+      "loss": 0.9308,
+      "step": 9682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3073598033731426e-05,
+      "loss": 0.9306,
+      "step": 9683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3064899135076025e-05,
+      "loss": 19.7668,
+      "step": 9684
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.30562026964698e-05,
+      "loss": 0.9151,
+      "step": 9685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3047508718491974e-05,
+      "loss": 19.0452,
+      "step": 9686
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3038817201721599e-05,
+      "loss": 0.2113,
+      "step": 9687
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.303012814673758e-05,
+      "loss": 19.1785,
+      "step": 9688
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3021441554118642e-05,
+      "loss": 0.9385,
+      "step": 9689
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3012757424443356e-05,
+      "loss": 0.926,
+      "step": 9690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3004075758290118e-05,
+      "loss": 0.9557,
+      "step": 9691
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2995396556237182e-05,
+      "loss": 0.2056,
+      "step": 9692
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2986719818862608e-05,
+      "loss": 0.9641,
+      "step": 9693
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2978045546744316e-05,
+      "loss": 18.2418,
+      "step": 9694
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.296937374046005e-05,
+      "loss": 0.9456,
+      "step": 9695
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2960704400587392e-05,
+      "loss": 20.0732,
+      "step": 9696
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2952037527703764e-05,
+      "loss": 0.2028,
+      "step": 9697
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.294337312238641e-05,
+      "loss": 0.9478,
+      "step": 9698
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2934711185212434e-05,
+      "loss": 19.157,
+      "step": 9699
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2926051716758764e-05,
+      "loss": 18.4765,
+      "step": 9700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2917394717602121e-05,
+      "loss": 17.3262,
+      "step": 9701
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2908740188319146e-05,
+      "loss": 17.9182,
+      "step": 9702
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2900088129486277e-05,
+      "loss": 21.8247,
+      "step": 9703
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2891438541679723e-05,
+      "loss": 0.19,
+      "step": 9704
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2882791425475637e-05,
+      "loss": 19.8598,
+      "step": 9705
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2874146781449965e-05,
+      "loss": 18.7587,
+      "step": 9706
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2865504610178442e-05,
+      "loss": 0.9126,
+      "step": 9707
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2856864912236677e-05,
+      "loss": 19.4845,
+      "step": 9708
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2848227688200143e-05,
+      "loss": 21.3351,
+      "step": 9709
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2839592938644129e-05,
+      "loss": 19.9937,
+      "step": 9710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2830960664143692e-05,
+      "loss": 0.2005,
+      "step": 9711
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2822330865273835e-05,
+      "loss": 19.0784,
+      "step": 9712
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.281370354260934e-05,
+      "loss": 19.4291,
+      "step": 9713
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2805078696724781e-05,
+      "loss": 0.1965,
+      "step": 9714
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2796456328194655e-05,
+      "loss": 19.1007,
+      "step": 9715
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.278783643759326e-05,
+      "loss": 0.9799,
+      "step": 9716
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2779219025494687e-05,
+      "loss": 0.1892,
+      "step": 9717
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2770604092472888e-05,
+      "loss": 0.2443,
+      "step": 9718
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2761991639101712e-05,
+      "loss": 19.6721,
+      "step": 9719
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2753381665954739e-05,
+      "loss": 0.9246,
+      "step": 9720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.274477417360544e-05,
+      "loss": 0.9228,
+      "step": 9721
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2736169162627148e-05,
+      "loss": 0.9823,
+      "step": 9722
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2727566633592957e-05,
+      "loss": 17.546,
+      "step": 9723
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2718966587075854e-05,
+      "loss": 19.275,
+      "step": 9724
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2710369023648633e-05,
+      "loss": 20.3015,
+      "step": 9725
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2701773943883943e-05,
+      "loss": 0.9463,
+      "step": 9726
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.269318134835425e-05,
+      "loss": 0.9402,
+      "step": 9727
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2684591237631853e-05,
+      "loss": 0.1729,
+      "step": 9728
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2676003612288911e-05,
+      "loss": 0.1883,
+      "step": 9729
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2667418472897386e-05,
+      "loss": 19.5244,
+      "step": 9730
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2658835820029092e-05,
+      "loss": 0.1699,
+      "step": 9731
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2650255654255672e-05,
+      "loss": 0.1687,
+      "step": 9732
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2641677976148609e-05,
+      "loss": 0.9209,
+      "step": 9733
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2633102786279216e-05,
+      "loss": 0.9284,
+      "step": 9734
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2624530085218633e-05,
+      "loss": 18.7987,
+      "step": 9735
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2615959873537847e-05,
+      "loss": 0.9363,
+      "step": 9736
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.260739215180768e-05,
+      "loss": 0.2528,
+      "step": 9737
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2598826920598772e-05,
+      "loss": 0.2057,
+      "step": 9738
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.259026418048161e-05,
+      "loss": 0.1666,
+      "step": 9739
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2581703932026512e-05,
+      "loss": 0.97,
+      "step": 9740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2573146175803635e-05,
+      "loss": 17.4212,
+      "step": 9741
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2564590912382961e-05,
+      "loss": 19.6077,
+      "step": 9742
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2556038142334309e-05,
+      "loss": 17.2954,
+      "step": 9743
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2547487866227336e-05,
+      "loss": 0.9316,
+      "step": 9744
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2538940084631535e-05,
+      "loss": 18.9435,
+      "step": 9745
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2530394798116213e-05,
+      "loss": 19.1662,
+      "step": 9746
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2521852007250556e-05,
+      "loss": 18.4428,
+      "step": 9747
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2513311712603499e-05,
+      "loss": 20.2443,
+      "step": 9748
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2504773914743916e-05,
+      "loss": 0.9828,
+      "step": 9749
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2496238614240463e-05,
+      "loss": 20.0744,
+      "step": 9750
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2487705811661583e-05,
+      "loss": 18.7164,
+      "step": 9751
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2479175507575646e-05,
+      "loss": 20.2835,
+      "step": 9752
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2470647702550798e-05,
+      "loss": 18.4778,
+      "step": 9753
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2462122397155023e-05,
+      "loss": 21.315,
+      "step": 9754
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2453599591956156e-05,
+      "loss": 18.0968,
+      "step": 9755
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2445079287521844e-05,
+      "loss": 0.9356,
+      "step": 9756
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2436561484419607e-05,
+      "loss": 19.2736,
+      "step": 9757
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2428046183216712e-05,
+      "loss": 0.2002,
+      "step": 9758
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2419533384480375e-05,
+      "loss": 0.9822,
+      "step": 9759
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2411023088777579e-05,
+      "loss": 20.2537,
+      "step": 9760
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2402515296675105e-05,
+      "loss": 19.9843,
+      "step": 9761
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.239401000873966e-05,
+      "loss": 0.9288,
+      "step": 9762
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2385507225537734e-05,
+      "loss": 0.9206,
+      "step": 9763
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2377006947635621e-05,
+      "loss": 0.2143,
+      "step": 9764
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.236850917559948e-05,
+      "loss": 19.0071,
+      "step": 9765
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.236001390999535e-05,
+      "loss": 20.6253,
+      "step": 9766
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2351521151389005e-05,
+      "loss": 0.9706,
+      "step": 9767
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2343030900346103e-05,
+      "loss": 19.8425,
+      "step": 9768
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2334543157432182e-05,
+      "loss": 19.2569,
+      "step": 9769
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.232605792321252e-05,
+      "loss": 0.9893,
+      "step": 9770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2317575198252285e-05,
+      "loss": 0.2229,
+      "step": 9771
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2309094983116465e-05,
+      "loss": 18.7092,
+      "step": 9772
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2300617278369886e-05,
+      "loss": 18.9415,
+      "step": 9773
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2292142084577202e-05,
+      "loss": 20.6128,
+      "step": 9774
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2283669402302878e-05,
+      "loss": 0.9373,
+      "step": 9775
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.227519923211129e-05,
+      "loss": 20.2522,
+      "step": 9776
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2266731574566536e-05,
+      "loss": 0.9254,
+      "step": 9777
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2258266430232613e-05,
+      "loss": 0.9044,
+      "step": 9778
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.224980379967337e-05,
+      "loss": 0.8901,
+      "step": 9779
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2241343683452422e-05,
+      "loss": 0.9336,
+      "step": 9780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.223288608213326e-05,
+      "loss": 18.745,
+      "step": 9781
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2224430996279207e-05,
+      "loss": 20.4148,
+      "step": 9782
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2215978426453406e-05,
+      "loss": 20.4733,
+      "step": 9783
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.220752837321884e-05,
+      "loss": 0.2504,
+      "step": 9784
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2199080837138322e-05,
+      "loss": 0.9302,
+      "step": 9785
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2190635818774492e-05,
+      "loss": 20.223,
+      "step": 9786
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2182193318689839e-05,
+      "loss": 19.0433,
+      "step": 9787
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2173753337446659e-05,
+      "loss": 0.1855,
+      "step": 9788
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2165315875607097e-05,
+      "loss": 0.2356,
+      "step": 9789
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2156880933733133e-05,
+      "loss": 1.0154,
+      "step": 9790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2148448512386573e-05,
+      "loss": 17.5047,
+      "step": 9791
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.214001861212905e-05,
+      "loss": 20.0091,
+      "step": 9792
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2131591233522038e-05,
+      "loss": 18.3077,
+      "step": 9793
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2123166377126854e-05,
+      "loss": 19.2138,
+      "step": 9794
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2114744043504583e-05,
+      "loss": 0.946,
+      "step": 9795
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2106324233216248e-05,
+      "loss": 18.2654,
+      "step": 9796
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2097906946822618e-05,
+      "loss": 0.9383,
+      "step": 9797
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2089492184884333e-05,
+      "loss": 0.9647,
+      "step": 9798
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2081079947961854e-05,
+      "loss": 17.4681,
+      "step": 9799
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2072670236615474e-05,
+      "loss": 17.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2064263051405316e-05,
+      "loss": 19.4335,
+      "step": 9801
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2055858392891345e-05,
+      "loss": 18.9455,
+      "step": 9802
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2047456261633343e-05,
+      "loss": 0.2059,
+      "step": 9803
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2039056658190946e-05,
+      "loss": 0.9536,
+      "step": 9804
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2030659583123565e-05,
+      "loss": 18.5119,
+      "step": 9805
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2022265036990533e-05,
+      "loss": 19.7196,
+      "step": 9806
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2013873020350952e-05,
+      "loss": 18.5455,
+      "step": 9807
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.200548353376374e-05,
+      "loss": 18.7492,
+      "step": 9808
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1997096577787714e-05,
+      "loss": 19.6501,
+      "step": 9809
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.198871215298148e-05,
+      "loss": 0.9514,
+      "step": 9810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1980330259903456e-05,
+      "loss": 20.9865,
+      "step": 9811
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.197195089911191e-05,
+      "loss": 20.4198,
+      "step": 9812
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1963574071164995e-05,
+      "loss": 0.207,
+      "step": 9813
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1955199776620607e-05,
+      "loss": 0.1688,
+      "step": 9814
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.19468280160365e-05,
+      "loss": 17.8712,
+      "step": 9815
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1938458789970309e-05,
+      "loss": 0.1683,
+      "step": 9816
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1930092098979462e-05,
+      "loss": 0.9336,
+      "step": 9817
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.192172794362118e-05,
+      "loss": 19.4804,
+      "step": 9818
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1913366324452591e-05,
+      "loss": 22.0635,
+      "step": 9819
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1905007242030625e-05,
+      "loss": 17.614,
+      "step": 9820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1896650696911998e-05,
+      "loss": 18.8141,
+      "step": 9821
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1888296689653306e-05,
+      "loss": 19.1167,
+      "step": 9822
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1879945220810996e-05,
+      "loss": 19.0569,
+      "step": 9823
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1871596290941278e-05,
+      "loss": 0.9166,
+      "step": 9824
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1863249900600232e-05,
+      "loss": 20.5932,
+      "step": 9825
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.18549060503438e-05,
+      "loss": 0.174,
+      "step": 9826
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.184656474072769e-05,
+      "loss": 0.9621,
+      "step": 9827
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1838225972307477e-05,
+      "loss": 0.1628,
+      "step": 9828
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1829889745638562e-05,
+      "loss": 20.54,
+      "step": 9829
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1821556061276185e-05,
+      "loss": 20.4222,
+      "step": 9830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1813224919775396e-05,
+      "loss": 0.9151,
+      "step": 9831
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1804896321691101e-05,
+      "loss": 18.5606,
+      "step": 9832
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.179657026757801e-05,
+      "loss": 20.6954,
+      "step": 9833
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1788246757990685e-05,
+      "loss": 0.1929,
+      "step": 9834
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.177992579348351e-05,
+      "loss": 18.3286,
+      "step": 9835
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1771607374610699e-05,
+      "loss": 18.2085,
+      "step": 9836
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1763291501926293e-05,
+      "loss": 0.1896,
+      "step": 9837
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1754978175984176e-05,
+      "loss": 19.1042,
+      "step": 9838
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1746667397338052e-05,
+      "loss": 19.8218,
+      "step": 9839
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1738359166541447e-05,
+      "loss": 0.19,
+      "step": 9840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.173005348414774e-05,
+      "loss": 19.0508,
+      "step": 9841
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1721750350710126e-05,
+      "loss": 0.1554,
+      "step": 9842
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1713449766781625e-05,
+      "loss": 0.9278,
+      "step": 9843
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1705151732915104e-05,
+      "loss": 0.9039,
+      "step": 9844
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.169685624966324e-05,
+      "loss": 19.0125,
+      "step": 9845
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1688563317578565e-05,
+      "loss": 19.926,
+      "step": 9846
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.168027293721341e-05,
+      "loss": 20.4021,
+      "step": 9847
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.167198510911997e-05,
+      "loss": 20.0965,
+      "step": 9848
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1663699833850238e-05,
+      "loss": 0.9088,
+      "step": 9849
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1655417111956058e-05,
+      "loss": 0.9603,
+      "step": 9850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1647136943989111e-05,
+      "loss": 19.7009,
+      "step": 9851
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1638859330500856e-05,
+      "loss": 18.8883,
+      "step": 9852
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.163058427204266e-05,
+      "loss": 19.9035,
+      "step": 9853
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1622311769165678e-05,
+      "loss": 19.2522,
+      "step": 9854
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1614041822420863e-05,
+      "loss": 0.1656,
+      "step": 9855
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1605774432359063e-05,
+      "loss": 20.7658,
+      "step": 9856
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1597509599530926e-05,
+      "loss": 0.9342,
+      "step": 9857
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1589247324486897e-05,
+      "loss": 19.4197,
+      "step": 9858
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1580987607777311e-05,
+      "loss": 18.7046,
+      "step": 9859
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1572730449952296e-05,
+      "loss": 18.3304,
+      "step": 9860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1564475851561835e-05,
+      "loss": 20.6627,
+      "step": 9861
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.155622381315567e-05,
+      "loss": 18.5701,
+      "step": 9862
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1547974335283473e-05,
+      "loss": 0.9607,
+      "step": 9863
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.15397274184947e-05,
+      "loss": 18.4531,
+      "step": 9864
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1531483063338588e-05,
+      "loss": 0.206,
+      "step": 9865
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1523241270364293e-05,
+      "loss": 19.9936,
+      "step": 9866
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1515002040120759e-05,
+      "loss": 17.8681,
+      "step": 9867
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.150676537315672e-05,
+      "loss": 0.9601,
+      "step": 9868
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1498531270020785e-05,
+      "loss": 18.547,
+      "step": 9869
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1490299731261427e-05,
+      "loss": 0.216,
+      "step": 9870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1482070757426856e-05,
+      "loss": 0.9595,
+      "step": 9871
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.147384434906516e-05,
+      "loss": 0.201,
+      "step": 9872
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.14656205067243e-05,
+      "loss": 0.9298,
+      "step": 9873
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1457399230951987e-05,
+      "loss": 0.1911,
+      "step": 9874
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.14491805222958e-05,
+      "loss": 19.1242,
+      "step": 9875
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.144096438130315e-05,
+      "loss": 0.2235,
+      "step": 9876
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.143275080852126e-05,
+      "loss": 0.9603,
+      "step": 9877
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1424539804497203e-05,
+      "loss": 20.0671,
+      "step": 9878
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1416331369777867e-05,
+      "loss": 17.4298,
+      "step": 9879
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1408125504909967e-05,
+      "loss": 0.9531,
+      "step": 9880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1399922210440061e-05,
+      "loss": 0.1572,
+      "step": 9881
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1391721486914503e-05,
+      "loss": 0.9111,
+      "step": 9882
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1383523334879548e-05,
+      "loss": 20.174,
+      "step": 9883
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1375327754881182e-05,
+      "loss": 17.57,
+      "step": 9884
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1367134747465291e-05,
+      "loss": 21.0052,
+      "step": 9885
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1358944313177567e-05,
+      "loss": 20.1891,
+      "step": 9886
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1350756452563522e-05,
+      "loss": 0.9366,
+      "step": 9887
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1342571166168509e-05,
+      "loss": 18.0702,
+      "step": 9888
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1334388454537708e-05,
+      "loss": 19.1724,
+      "step": 9889
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.132620831821613e-05,
+      "loss": 0.9163,
+      "step": 9890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1318030757748605e-05,
+      "loss": 18.8839,
+      "step": 9891
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1309855773679794e-05,
+      "loss": 17.7216,
+      "step": 9892
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1301683366554195e-05,
+      "loss": 18.6739,
+      "step": 9893
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1293513536916123e-05,
+      "loss": 18.5229,
+      "step": 9894
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1285346285309729e-05,
+      "loss": 19.4427,
+      "step": 9895
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1277181612278986e-05,
+      "loss": 0.1971,
+      "step": 9896
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1269019518367708e-05,
+      "loss": 20.5131,
+      "step": 9897
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.126086000411954e-05,
+      "loss": 20.2116,
+      "step": 9898
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1252703070077892e-05,
+      "loss": 0.1939,
+      "step": 9899
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1244548716786102e-05,
+      "loss": 18.2482,
+      "step": 9900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1236396944787275e-05,
+      "loss": 20.1554,
+      "step": 9901
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.122824775462436e-05,
+      "loss": 20.7312,
+      "step": 9902
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1220101146840123e-05,
+      "loss": 0.2223,
+      "step": 9903
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1211957121977174e-05,
+      "loss": 17.4302,
+      "step": 9904
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1203815680577934e-05,
+      "loss": 19.9204,
+      "step": 9905
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1195676823184665e-05,
+      "loss": 0.1991,
+      "step": 9906
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1187540550339454e-05,
+      "loss": 0.1968,
+      "step": 9907
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.117940686258423e-05,
+      "loss": 18.4484,
+      "step": 9908
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1171275760460687e-05,
+      "loss": 0.2024,
+      "step": 9909
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1163147244510441e-05,
+      "loss": 0.9766,
+      "step": 9910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1155021315274884e-05,
+      "loss": 18.0153,
+      "step": 9911
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1146897973295205e-05,
+      "loss": 0.9828,
+      "step": 9912
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1138777219112495e-05,
+      "loss": 0.9113,
+      "step": 9913
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1130659053267628e-05,
+      "loss": 18.3135,
+      "step": 9914
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1122543476301295e-05,
+      "loss": 19.9673,
+      "step": 9915
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1114430488754025e-05,
+      "loss": 0.1995,
+      "step": 9916
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1106320091166223e-05,
+      "loss": 1.0008,
+      "step": 9917
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1098212284078036e-05,
+      "loss": 0.1608,
+      "step": 9918
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1090107068029482e-05,
+      "loss": 18.3765,
+      "step": 9919
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.108200444356045e-05,
+      "loss": 19.5675,
+      "step": 9920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1073904411210567e-05,
+      "loss": 18.8171,
+      "step": 9921
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1065806971519338e-05,
+      "loss": 0.9056,
+      "step": 9922
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1057712125026116e-05,
+      "loss": 20.2691,
+      "step": 9923
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.104961987227005e-05,
+      "loss": 0.2118,
+      "step": 9924
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1041530213790102e-05,
+      "loss": 20.1611,
+      "step": 9925
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1033443150125073e-05,
+      "loss": 0.2156,
+      "step": 9926
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1025358681813646e-05,
+      "loss": 18.3066,
+      "step": 9927
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1017276809394245e-05,
+      "loss": 18.4457,
+      "step": 9928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1009197533405153e-05,
+      "loss": 0.959,
+      "step": 9929
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1001120854384528e-05,
+      "loss": 0.9261,
+      "step": 9930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.099304677287028e-05,
+      "loss": 0.1902,
+      "step": 9931
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.098497528940019e-05,
+      "loss": 19.071,
+      "step": 9932
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0976906404511861e-05,
+      "loss": 18.7154,
+      "step": 9933
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0968840118742713e-05,
+      "loss": 17.2723,
+      "step": 9934
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0960776432629999e-05,
+      "loss": 19.4274,
+      "step": 9935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0952715346710796e-05,
+      "loss": 0.964,
+      "step": 9936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0944656861522018e-05,
+      "loss": 19.4339,
+      "step": 9937
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0936600977600387e-05,
+      "loss": 0.2067,
+      "step": 9938
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0928547695482467e-05,
+      "loss": 19.1514,
+      "step": 9939
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0920497015704645e-05,
+      "loss": 19.4531,
+      "step": 9940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0912448938803133e-05,
+      "loss": 19.548,
+      "step": 9941
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0904403465313973e-05,
+      "loss": 0.9603,
+      "step": 9942
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0896360595773019e-05,
+      "loss": 0.214,
+      "step": 9943
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.088832033071598e-05,
+      "loss": 19.8854,
+      "step": 9944
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0880282670678366e-05,
+      "loss": 17.5281,
+      "step": 9945
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0872247616195525e-05,
+      "loss": 20.0895,
+      "step": 9946
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0864215167802622e-05,
+      "loss": 0.9592,
+      "step": 9947
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0856185326034662e-05,
+      "loss": 0.2003,
+      "step": 9948
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0848158091426475e-05,
+      "loss": 18.2961,
+      "step": 9949
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0840133464512697e-05,
+      "loss": 19.1589,
+      "step": 9950
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0832111445827819e-05,
+      "loss": 19.4119,
+      "step": 9951
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0824092035906142e-05,
+      "loss": 0.2041,
+      "step": 9952
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0816075235281791e-05,
+      "loss": 18.0092,
+      "step": 9953
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0808061044488726e-05,
+      "loss": 20.6141,
+      "step": 9954
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0800049464060741e-05,
+      "loss": 0.1897,
+      "step": 9955
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0792040494531403e-05,
+      "loss": 20.3989,
+      "step": 9956
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.078403413643419e-05,
+      "loss": 0.9657,
+      "step": 9957
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.077603039030236e-05,
+      "loss": 18.636,
+      "step": 9958
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0768029256668966e-05,
+      "loss": 0.2074,
+      "step": 9959
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0760030736066951e-05,
+      "loss": 17.9778,
+      "step": 9960
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0752034829029056e-05,
+      "loss": 0.9831,
+      "step": 9961
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.074404153608783e-05,
+      "loss": 0.1924,
+      "step": 9962
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0736050857775643e-05,
+      "loss": 19.7132,
+      "step": 9963
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0728062794624772e-05,
+      "loss": 18.4306,
+      "step": 9964
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0720077347167202e-05,
+      "loss": 0.188,
+      "step": 9965
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0712094515934813e-05,
+      "loss": 17.7476,
+      "step": 9966
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0704114301459317e-05,
+      "loss": 17.2287,
+      "step": 9967
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0696136704272242e-05,
+      "loss": 0.9453,
+      "step": 9968
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0688161724904888e-05,
+      "loss": 0.2026,
+      "step": 9969
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0680189363888459e-05,
+      "loss": 18.7541,
+      "step": 9970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0672219621753965e-05,
+      "loss": 19.7309,
+      "step": 9971
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0664252499032196e-05,
+      "loss": 0.9324,
+      "step": 9972
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0656287996253789e-05,
+      "loss": 18.8237,
+      "step": 9973
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.064832611394927e-05,
+      "loss": 0.2272,
+      "step": 9974
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0640366852648897e-05,
+      "loss": 0.2432,
+      "step": 9975
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0632410212882782e-05,
+      "loss": 19.1391,
+      "step": 9976
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0624456195180926e-05,
+      "loss": 20.3242,
+      "step": 9977
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0616504800073058e-05,
+      "loss": 19.3416,
+      "step": 9978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0608556028088795e-05,
+      "loss": 19.2805,
+      "step": 9979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0600609879757556e-05,
+      "loss": 0.2106,
+      "step": 9980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0592666355608599e-05,
+      "loss": 0.2053,
+      "step": 9981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.058472545617099e-05,
+      "loss": 19.0617,
+      "step": 9982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0576787181973636e-05,
+      "loss": 0.9625,
+      "step": 9983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0568851533545265e-05,
+      "loss": 19.495,
+      "step": 9984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0560918511414424e-05,
+      "loss": 18.5796,
+      "step": 9985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0552988116109475e-05,
+      "loss": 0.2284,
+      "step": 9986
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0545060348158663e-05,
+      "loss": 17.8055,
+      "step": 9987
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0537135208089971e-05,
+      "loss": 0.9354,
+      "step": 9988
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0529212696431267e-05,
+      "loss": 0.1964,
+      "step": 9989
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.052129281371022e-05,
+      "loss": 20.0886,
+      "step": 9990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0513375560454342e-05,
+      "loss": 16.9482,
+      "step": 9991
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0505460937190948e-05,
+      "loss": 0.2222,
+      "step": 9992
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0497548944447195e-05,
+      "loss": 0.9099,
+      "step": 9993
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0489639582750055e-05,
+      "loss": 18.6705,
+      "step": 9994
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.048173285262633e-05,
+      "loss": 0.9207,
+      "step": 9995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0473828754602643e-05,
+      "loss": 19.622,
+      "step": 9996
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0465927289205452e-05,
+      "loss": 17.6386,
+      "step": 9997
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0458028456961016e-05,
+      "loss": 0.159,
+      "step": 9998
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0450132258395445e-05,
+      "loss": 1.006,
+      "step": 9999
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0442238694034662e-05,
+      "loss": 19.6281,
+      "step": 10000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0434347764404412e-05,
+      "loss": 18.5007,
+      "step": 10001
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.042645947003028e-05,
+      "loss": 19.1974,
+      "step": 10002
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0418573811437631e-05,
+      "loss": 18.3778,
+      "step": 10003
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0410690789151718e-05,
+      "loss": 0.9256,
+      "step": 10004
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0402810403697582e-05,
+      "loss": 19.5069,
+      "step": 10005
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0394932655600064e-05,
+      "loss": 18.5178,
+      "step": 10006
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.03870575453839e-05,
+      "loss": 0.193,
+      "step": 10007
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0379185073573588e-05,
+      "loss": 18.5937,
+      "step": 10008
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0371315240693475e-05,
+      "loss": 19.7573,
+      "step": 10009
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0363448047267721e-05,
+      "loss": 0.2334,
+      "step": 10010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0355583493820332e-05,
+      "loss": 0.9587,
+      "step": 10011
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0347721580875126e-05,
+      "loss": 0.9071,
+      "step": 10012
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.03398623089557e-05,
+      "loss": 19.097,
+      "step": 10013
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.033200567858557e-05,
+      "loss": 20.0732,
+      "step": 10014
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0324151690288014e-05,
+      "loss": 0.9629,
+      "step": 10015
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0316300344586111e-05,
+      "loss": 19.5915,
+      "step": 10016
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0308451642002836e-05,
+      "loss": 19.0459,
+      "step": 10017
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0300605583060941e-05,
+      "loss": 17.3578,
+      "step": 10018
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0292762168282994e-05,
+      "loss": 19.9456,
+      "step": 10019
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0284921398191394e-05,
+      "loss": 0.9421,
+      "step": 10020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0277083273308425e-05,
+      "loss": 0.913,
+      "step": 10021
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0269247794156084e-05,
+      "loss": 19.8557,
+      "step": 10022
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0261414961256267e-05,
+      "loss": 20.5107,
+      "step": 10023
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.025358477513071e-05,
+      "loss": 19.5004,
+      "step": 10024
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0245757236300896e-05,
+      "loss": 16.8691,
+      "step": 10025
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0237932345288187e-05,
+      "loss": 0.9282,
+      "step": 10026
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0230110102613782e-05,
+      "loss": 18.9729,
+      "step": 10027
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0222290508798654e-05,
+      "loss": 21.3516,
+      "step": 10028
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0214473564363624e-05,
+      "loss": 19.9162,
+      "step": 10029
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0206659269829333e-05,
+      "loss": 0.9659,
+      "step": 10030
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0198847625716284e-05,
+      "loss": 21.2524,
+      "step": 10031
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.019103863254473e-05,
+      "loss": 0.1949,
+      "step": 10032
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0183232290834788e-05,
+      "loss": 20.5946,
+      "step": 10033
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.017542860110644e-05,
+      "loss": 0.9615,
+      "step": 10034
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0167627563879396e-05,
+      "loss": 0.2358,
+      "step": 10035
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0159829179673264e-05,
+      "loss": 19.6597,
+      "step": 10036
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0152033449007453e-05,
+      "loss": 0.9063,
+      "step": 10037
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0144240372401193e-05,
+      "loss": 0.9012,
+      "step": 10038
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0136449950373534e-05,
+      "loss": 0.9638,
+      "step": 10039
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.012866218344336e-05,
+      "loss": 0.9567,
+      "step": 10040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0120877072129376e-05,
+      "loss": 0.1842,
+      "step": 10041
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0113094616950098e-05,
+      "loss": 19.8969,
+      "step": 10042
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0105314818423877e-05,
+      "loss": 0.9301,
+      "step": 10043
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0097537677068885e-05,
+      "loss": 19.2858,
+      "step": 10044
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0089763193403112e-05,
+      "loss": 19.5613,
+      "step": 10045
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0081991367944382e-05,
+      "loss": 19.8157,
+      "step": 10046
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0074222201210331e-05,
+      "loss": 18.7917,
+      "step": 10047
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0066455693718418e-05,
+      "loss": 20.6231,
+      "step": 10048
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.005869184598593e-05,
+      "loss": 19.8861,
+      "step": 10049
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0050930658529977e-05,
+      "loss": 0.9649,
+      "step": 10050
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.004317213186749e-05,
+      "loss": 0.9209,
+      "step": 10051
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0035416266515225e-05,
+      "loss": 0.157,
+      "step": 10052
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.002766306298975e-05,
+      "loss": 18.5421,
+      "step": 10053
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0019912521807478e-05,
+      "loss": 17.0364,
+      "step": 10054
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0012164643484617e-05,
+      "loss": 19.6786,
+      "step": 10055
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0004419428537221e-05,
+      "loss": 18.5151,
+      "step": 10056
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.996676877481154e-06,
+      "loss": 0.2041,
+      "step": 10057
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.98893699083211e-06,
+      "loss": 18.3058,
+      "step": 10058
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.981199769105604e-06,
+      "loss": 0.2072,
+      "step": 10059
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.973465212816946e-06,
+      "loss": 0.8901,
+      "step": 10060
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.965733322481319e-06,
+      "loss": 19.9159,
+      "step": 10061
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.95800409861371e-06,
+      "loss": 20.5522,
+      "step": 10062
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.95027754172888e-06,
+      "loss": 21.3849,
+      "step": 10063
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.942553652341496e-06,
+      "loss": 0.9049,
+      "step": 10064
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.934832430966e-06,
+      "loss": 0.2178,
+      "step": 10065
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.92711387811664e-06,
+      "loss": 18.9326,
+      "step": 10066
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.919397994307505e-06,
+      "loss": 18.9972,
+      "step": 10067
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.911684780052543e-06,
+      "loss": 0.9162,
+      "step": 10068
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.90397423586546e-06,
+      "loss": 20.2394,
+      "step": 10069
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.896266362259809e-06,
+      "loss": 0.9299,
+      "step": 10070
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.888561159748993e-06,
+      "loss": 0.2344,
+      "step": 10071
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.880858628846224e-06,
+      "loss": 0.2083,
+      "step": 10072
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.873158770064479e-06,
+      "loss": 18.5077,
+      "step": 10073
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.865461583916647e-06,
+      "loss": 17.7956,
+      "step": 10074
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.857767070915391e-06,
+      "loss": 18.8601,
+      "step": 10075
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.85007523157318e-06,
+      "loss": 21.074,
+      "step": 10076
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.842386066402325e-06,
+      "loss": 19.5776,
+      "step": 10077
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.834699575915002e-06,
+      "loss": 0.1574,
+      "step": 10078
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.827015760623125e-06,
+      "loss": 18.8251,
+      "step": 10079
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.819334621038473e-06,
+      "loss": 0.2274,
+      "step": 10080
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.811656157672684e-06,
+      "loss": 0.9292,
+      "step": 10081
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.803980371037136e-06,
+      "loss": 0.962,
+      "step": 10082
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.796307261643089e-06,
+      "loss": 0.2264,
+      "step": 10083
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.788636830001613e-06,
+      "loss": 19.4643,
+      "step": 10084
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.780969076623587e-06,
+      "loss": 0.1889,
+      "step": 10085
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.773304002019717e-06,
+      "loss": 18.6323,
+      "step": 10086
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.765641606700543e-06,
+      "loss": 0.1738,
+      "step": 10087
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.757981891176405e-06,
+      "loss": 0.9087,
+      "step": 10088
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.750324855957482e-06,
+      "loss": 0.9902,
+      "step": 10089
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.742670501553768e-06,
+      "loss": 0.1831,
+      "step": 10090
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.735018828475079e-06,
+      "loss": 0.9251,
+      "step": 10091
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.727369837231048e-06,
+      "loss": 19.8018,
+      "step": 10092
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.719723528331132e-06,
+      "loss": 18.2015,
+      "step": 10093
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.712079902284616e-06,
+      "loss": 19.2118,
+      "step": 10094
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.704438959600597e-06,
+      "loss": 20.3864,
+      "step": 10095
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.696800700788006e-06,
+      "loss": 17.8522,
+      "step": 10096
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.689165126355575e-06,
+      "loss": 0.9095,
+      "step": 10097
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.681532236811875e-06,
+      "loss": 0.2021,
+      "step": 10098
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.673902032665288e-06,
+      "loss": 20.9194,
+      "step": 10099
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.666274514424024e-06,
+      "loss": 0.1944,
+      "step": 10100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.65864968259611e-06,
+      "loss": 19.7645,
+      "step": 10101
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.651027537689395e-06,
+      "loss": 0.9363,
+      "step": 10102
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.643408080211552e-06,
+      "loss": 19.0402,
+      "step": 10103
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.635791310670068e-06,
+      "loss": 19.2053,
+      "step": 10104
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.62817722957226e-06,
+      "loss": 18.5104,
+      "step": 10105
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.620565837425271e-06,
+      "loss": 0.1932,
+      "step": 10106
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.612957134736011e-06,
+      "loss": 18.037,
+      "step": 10107
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.605351122011309e-06,
+      "loss": 18.4406,
+      "step": 10108
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.597747799757745e-06,
+      "loss": 0.2083,
+      "step": 10109
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.5901471684817e-06,
+      "loss": 0.8948,
+      "step": 10110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.582549228689457e-06,
+      "loss": 0.9362,
+      "step": 10111
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.574953980887069e-06,
+      "loss": 0.1733,
+      "step": 10112
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.567361425580379e-06,
+      "loss": 0.9566,
+      "step": 10113
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.559771563275121e-06,
+      "loss": 18.3277,
+      "step": 10114
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.552184394476809e-06,
+      "loss": 0.9442,
+      "step": 10115
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.544599919690789e-06,
+      "loss": 0.9427,
+      "step": 10116
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.537018139422182e-06,
+      "loss": 0.9359,
+      "step": 10117
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.52943905417602e-06,
+      "loss": 19.0382,
+      "step": 10118
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.521862664457093e-06,
+      "loss": 0.251,
+      "step": 10119
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.51428897076999e-06,
+      "loss": 18.9284,
+      "step": 10120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.506717973619195e-06,
+      "loss": 19.7718,
+      "step": 10121
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.499149673508972e-06,
+      "loss": 18.7696,
+      "step": 10122
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.491584070943372e-06,
+      "loss": 0.1772,
+      "step": 10123
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.484021166426311e-06,
+      "loss": 17.2037,
+      "step": 10124
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.476460960461537e-06,
+      "loss": 19.7991,
+      "step": 10125
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.468903453552569e-06,
+      "loss": 20.2815,
+      "step": 10126
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.461348646202767e-06,
+      "loss": 0.2118,
+      "step": 10127
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.453796538915355e-06,
+      "loss": 18.2531,
+      "step": 10128
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.446247132193298e-06,
+      "loss": 0.9609,
+      "step": 10129
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.438700426539426e-06,
+      "loss": 18.4754,
+      "step": 10130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.431156422456417e-06,
+      "loss": 16.1114,
+      "step": 10131
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.423615120446705e-06,
+      "loss": 0.1883,
+      "step": 10132
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.416076521012584e-06,
+      "loss": 18.082,
+      "step": 10133
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.408540624656164e-06,
+      "loss": 20.0991,
+      "step": 10134
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.401007431879367e-06,
+      "loss": 18.1026,
+      "step": 10135
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.393476943183938e-06,
+      "loss": 0.9547,
+      "step": 10136
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.385949159071434e-06,
+      "loss": 0.2142,
+      "step": 10137
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.378424080043274e-06,
+      "loss": 0.9657,
+      "step": 10138
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.370901706600627e-06,
+      "loss": 0.9476,
+      "step": 10139
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.363382039244529e-06,
+      "loss": 18.699,
+      "step": 10140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.355865078475828e-06,
+      "loss": 0.2112,
+      "step": 10141
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.348350824795188e-06,
+      "loss": 19.7102,
+      "step": 10142
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.34083927870309e-06,
+      "loss": 0.2045,
+      "step": 10143
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.33333044069984e-06,
+      "loss": 0.863,
+      "step": 10144
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.325824311285564e-06,
+      "loss": 0.2213,
+      "step": 10145
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.318320890960198e-06,
+      "loss": 18.4785,
+      "step": 10146
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.310820180223512e-06,
+      "loss": 19.5255,
+      "step": 10147
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.303322179575086e-06,
+      "loss": 20.0492,
+      "step": 10148
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.295826889514314e-06,
+      "loss": 17.8316,
+      "step": 10149
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.28833431054043e-06,
+      "loss": 0.2096,
+      "step": 10150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.28084444315247e-06,
+      "loss": 0.1617,
+      "step": 10151
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.273357287849288e-06,
+      "loss": 0.9084,
+      "step": 10152
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.26587284512957e-06,
+      "loss": 18.6626,
+      "step": 10153
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.25839111549181e-06,
+      "loss": 20.1425,
+      "step": 10154
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.250912099434334e-06,
+      "loss": 0.9256,
+      "step": 10155
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.243435797455263e-06,
+      "loss": 19.7065,
+      "step": 10156
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.235962210052574e-06,
+      "loss": 0.1938,
+      "step": 10157
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.228491337724027e-06,
+      "loss": 0.9136,
+      "step": 10158
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.221023180967225e-06,
+      "loss": 0.9377,
+      "step": 10159
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.213557740279577e-06,
+      "loss": 20.949,
+      "step": 10160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.206095016158317e-06,
+      "loss": 18.9045,
+      "step": 10161
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.198635009100503e-06,
+      "loss": 18.1305,
+      "step": 10162
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.191177719603012e-06,
+      "loss": 0.2244,
+      "step": 10163
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.183723148162493e-06,
+      "loss": 0.9416,
+      "step": 10164
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.176271295275502e-06,
+      "loss": 0.2105,
+      "step": 10165
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.168822161438367e-06,
+      "loss": 18.918,
+      "step": 10166
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.161375747147188e-06,
+      "loss": 19.817,
+      "step": 10167
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.153932052897973e-06,
+      "loss": 18.282,
+      "step": 10168
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.146491079186508e-06,
+      "loss": 0.9077,
+      "step": 10169
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.139052826508376e-06,
+      "loss": 17.5477,
+      "step": 10170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.13161729535899e-06,
+      "loss": 17.3852,
+      "step": 10171
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.124184486233628e-06,
+      "loss": 0.9944,
+      "step": 10172
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.11675439962732e-06,
+      "loss": 18.9953,
+      "step": 10173
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.109327036034943e-06,
+      "loss": 19.043,
+      "step": 10174
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.101902395951229e-06,
+      "loss": 18.1678,
+      "step": 10175
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.094480479870649e-06,
+      "loss": 20.2491,
+      "step": 10176
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.087061288287546e-06,
+      "loss": 18.6745,
+      "step": 10177
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.079644821696099e-06,
+      "loss": 0.1819,
+      "step": 10178
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.072231080590276e-06,
+      "loss": 19.7756,
+      "step": 10179
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.064820065463842e-06,
+      "loss": 18.1949,
+      "step": 10180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.057411776810398e-06,
+      "loss": 0.9071,
+      "step": 10181
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.050006215123419e-06,
+      "loss": 0.1955,
+      "step": 10182
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.042603380896109e-06,
+      "loss": 0.2166,
+      "step": 10183
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.035203274621523e-06,
+      "loss": 0.1739,
+      "step": 10184
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.027805896792591e-06,
+      "loss": 0.206,
+      "step": 10185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.020411247901972e-06,
+      "loss": 0.2006,
+      "step": 10186
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.01301932844219e-06,
+      "loss": 0.2342,
+      "step": 10187
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.005630138905591e-06,
+      "loss": 17.0449,
+      "step": 10188
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.998243679784324e-06,
+      "loss": 17.8452,
+      "step": 10189
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.990859951570357e-06,
+      "loss": 0.9312,
+      "step": 10190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.983478954755492e-06,
+      "loss": 0.888,
+      "step": 10191
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.976100689831329e-06,
+      "loss": 17.6413,
+      "step": 10192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.968725157289298e-06,
+      "loss": 0.8868,
+      "step": 10193
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.96135235762064e-06,
+      "loss": 0.1593,
+      "step": 10194
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.953982291316421e-06,
+      "loss": 17.9104,
+      "step": 10195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.946614958867522e-06,
+      "loss": 18.8533,
+      "step": 10196
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.939250360764645e-06,
+      "loss": 0.1961,
+      "step": 10197
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.931888497498297e-06,
+      "loss": 18.925,
+      "step": 10198
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.924529369558815e-06,
+      "loss": 19.269,
+      "step": 10199
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.917172977436356e-06,
+      "loss": 0.225,
+      "step": 10200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.909819321620888e-06,
+      "loss": 0.9699,
+      "step": 10201
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.9024684026022e-06,
+      "loss": 0.1877,
+      "step": 10202
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.89512022086989e-06,
+      "loss": 0.9311,
+      "step": 10203
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.887774776913388e-06,
+      "loss": 20.9237,
+      "step": 10204
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.880432071221933e-06,
+      "loss": 0.9257,
+      "step": 10205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.87309210428458e-06,
+      "loss": 0.2365,
+      "step": 10206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.865754876590216e-06,
+      "loss": 21.4405,
+      "step": 10207
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.85842038862752e-06,
+      "loss": 0.9509,
+      "step": 10208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.851088640885007e-06,
+      "loss": 0.9713,
+      "step": 10209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.843759633851029e-06,
+      "loss": 0.1633,
+      "step": 10210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.83643336801368e-06,
+      "loss": 0.2053,
+      "step": 10211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.829109843860966e-06,
+      "loss": 0.2118,
+      "step": 10212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.82178906188067e-06,
+      "loss": 0.9445,
+      "step": 10213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.81447102256035e-06,
+      "loss": 0.2107,
+      "step": 10214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.807155726387462e-06,
+      "loss": 18.2226,
+      "step": 10215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.79984317384923e-06,
+      "loss": 18.1438,
+      "step": 10216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.792533365432676e-06,
+      "loss": 19.8432,
+      "step": 10217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.785226301624705e-06,
+      "loss": 20.0446,
+      "step": 10218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.777921982911996e-06,
+      "loss": 20.662,
+      "step": 10219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.770620409781028e-06,
+      "loss": 0.1999,
+      "step": 10220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.763321582718116e-06,
+      "loss": 0.2034,
+      "step": 10221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.756025502209425e-06,
+      "loss": 20.4822,
+      "step": 10222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.748732168740904e-06,
+      "loss": 0.925,
+      "step": 10223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.741441582798293e-06,
+      "loss": 0.1799,
+      "step": 10224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.734153744867208e-06,
+      "loss": 0.9595,
+      "step": 10225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.726868655433056e-06,
+      "loss": 19.2866,
+      "step": 10226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.719586314981031e-06,
+      "loss": 19.8392,
+      "step": 10227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.712306723996178e-06,
+      "loss": 0.1847,
+      "step": 10228
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.705029882963383e-06,
+      "loss": 0.1869,
+      "step": 10229
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.697755792367279e-06,
+      "loss": 0.2283,
+      "step": 10230
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.690484452692355e-06,
+      "loss": 16.8624,
+      "step": 10231
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.683215864422955e-06,
+      "loss": 0.201,
+      "step": 10232
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.675950028043161e-06,
+      "loss": 21.9557,
+      "step": 10233
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.66868694403693e-06,
+      "loss": 19.3075,
+      "step": 10234
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.661426612888007e-06,
+      "loss": 0.1953,
+      "step": 10235
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.654169035079972e-06,
+      "loss": 20.632,
+      "step": 10236
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.646914211096208e-06,
+      "loss": 0.2048,
+      "step": 10237
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.639662141419924e-06,
+      "loss": 21.1276,
+      "step": 10238
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.632412826534142e-06,
+      "loss": 21.0458,
+      "step": 10239
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.625166266921697e-06,
+      "loss": 18.6242,
+      "step": 10240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.61792246306523e-06,
+      "loss": 18.9223,
+      "step": 10241
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.610681415447258e-06,
+      "loss": 0.9651,
+      "step": 10242
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.60344312455002e-06,
+      "loss": 0.1898,
+      "step": 10243
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.596207590855637e-06,
+      "loss": 20.5965,
+      "step": 10244
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.58897481484603e-06,
+      "loss": 17.9899,
+      "step": 10245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.581744797002933e-06,
+      "loss": 0.9375,
+      "step": 10246
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.574517537807897e-06,
+      "loss": 0.1752,
+      "step": 10247
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.567293037742297e-06,
+      "loss": 18.0861,
+      "step": 10248
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.560071297287314e-06,
+      "loss": 0.9035,
+      "step": 10249
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.552852316923955e-06,
+      "loss": 20.0495,
+      "step": 10250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.54563609713303e-06,
+      "loss": 0.215,
+      "step": 10251
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.538422638395178e-06,
+      "loss": 19.5777,
+      "step": 10252
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.531211941190841e-06,
+      "loss": 20.0169,
+      "step": 10253
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.524004006000296e-06,
+      "loss": 0.9131,
+      "step": 10254
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.516798833303625e-06,
+      "loss": 20.2886,
+      "step": 10255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.509596423580712e-06,
+      "loss": 0.9816,
+      "step": 10256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.502396777311283e-06,
+      "loss": 19.5548,
+      "step": 10257
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.495199894974865e-06,
+      "loss": 19.4453,
+      "step": 10258
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.4880057770508e-06,
+      "loss": 19.4611,
+      "step": 10259
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.480814424018268e-06,
+      "loss": 0.2045,
+      "step": 10260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.473625836356208e-06,
+      "loss": 0.1839,
+      "step": 10261
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.466440014543447e-06,
+      "loss": 0.9462,
+      "step": 10262
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.459256959058587e-06,
+      "loss": 0.9655,
+      "step": 10263
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.45207667038005e-06,
+      "loss": 0.1896,
+      "step": 10264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.444899148986068e-06,
+      "loss": 0.9191,
+      "step": 10265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.437724395354713e-06,
+      "loss": 0.9001,
+      "step": 10266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.43055240996386e-06,
+      "loss": 18.8655,
+      "step": 10267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.423383193291162e-06,
+      "loss": 0.172,
+      "step": 10268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.41621674581416e-06,
+      "loss": 17.3771,
+      "step": 10269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.409053068010164e-06,
+      "loss": 17.5702,
+      "step": 10270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.401892160356284e-06,
+      "loss": 19.0154,
+      "step": 10271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.394734023329497e-06,
+      "loss": 21.7873,
+      "step": 10272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.387578657406576e-06,
+      "loss": 0.9039,
+      "step": 10273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.380426063064073e-06,
+      "loss": 17.5162,
+      "step": 10274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.373276240778383e-06,
+      "loss": 0.946,
+      "step": 10275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.366129191025757e-06,
+      "loss": 19.4973,
+      "step": 10276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.358984914282181e-06,
+      "loss": 0.9354,
+      "step": 10277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.351843411023496e-06,
+      "loss": 19.6051,
+      "step": 10278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.344704681725402e-06,
+      "loss": 0.9334,
+      "step": 10279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.337568726863331e-06,
+      "loss": 20.7756,
+      "step": 10280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.330435546912574e-06,
+      "loss": 0.1845,
+      "step": 10281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.32330514234827e-06,
+      "loss": 18.7874,
+      "step": 10282
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.316177513645296e-06,
+      "loss": 18.7687,
+      "step": 10283
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.309052661278398e-06,
+      "loss": 0.2029,
+      "step": 10284
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.301930585722112e-06,
+      "loss": 19.5723,
+      "step": 10285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.294811287450843e-06,
+      "loss": 19.4322,
+      "step": 10286
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.28769476693873e-06,
+      "loss": 19.2916,
+      "step": 10287
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.280581024659766e-06,
+      "loss": 0.1819,
+      "step": 10288
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.273470061087795e-06,
+      "loss": 18.4007,
+      "step": 10289
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.266361876696405e-06,
+      "loss": 0.2132,
+      "step": 10290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.259256471959049e-06,
+      "loss": 19.8802,
+      "step": 10291
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.25215384734897e-06,
+      "loss": 19.8924,
+      "step": 10292
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.245054003339247e-06,
+      "loss": 19.1469,
+      "step": 10293
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.237956940402757e-06,
+      "loss": 0.8467,
+      "step": 10294
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.230862659012194e-06,
+      "loss": 0.9426,
+      "step": 10295
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.22377115964008e-06,
+      "loss": 18.7869,
+      "step": 10296
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.216682442758732e-06,
+      "loss": 0.2078,
+      "step": 10297
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.2095965088403e-06,
+      "loss": 0.9608,
+      "step": 10298
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.202513358356733e-06,
+      "loss": 0.9984,
+      "step": 10299
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.195432991779806e-06,
+      "loss": 0.9524,
+      "step": 10300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.1883554095811e-06,
+      "loss": 18.9423,
+      "step": 10301
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.181280612232017e-06,
+      "loss": 18.5339,
+      "step": 10302
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.174208600203776e-06,
+      "loss": 0.96,
+      "step": 10303
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.167139373967398e-06,
+      "loss": 19.0841,
+      "step": 10304
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.160072933993729e-06,
+      "loss": 0.9328,
+      "step": 10305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.153009280753427e-06,
+      "loss": 0.1758,
+      "step": 10306
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.14594841471697e-06,
+      "loss": 0.194,
+      "step": 10307
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.138890336354632e-06,
+      "loss": 0.9011,
+      "step": 10308
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.131835046136527e-06,
+      "loss": 20.3525,
+      "step": 10309
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.124782544532562e-06,
+      "loss": 19.1674,
+      "step": 10310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.11773283201247e-06,
+      "loss": 0.9093,
+      "step": 10311
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.11068590904579e-06,
+      "loss": 18.0008,
+      "step": 10312
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.103641776101883e-06,
+      "loss": 18.6804,
+      "step": 10313
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.096600433649937e-06,
+      "loss": 20.7781,
+      "step": 10314
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.089561882158892e-06,
+      "loss": 20.9946,
+      "step": 10315
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.082526122097594e-06,
+      "loss": 0.9032,
+      "step": 10316
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.075493153934648e-06,
+      "loss": 0.221,
+      "step": 10317
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.068462978138459e-06,
+      "loss": 19.1291,
+      "step": 10318
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.061435595177292e-06,
+      "loss": 18.0348,
+      "step": 10319
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.054411005519213e-06,
+      "loss": 19.4432,
+      "step": 10320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.04738920963205e-06,
+      "loss": 17.8851,
+      "step": 10321
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.040370207983527e-06,
+      "loss": 20.0785,
+      "step": 10322
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.033354001041143e-06,
+      "loss": 0.1969,
+      "step": 10323
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.026340589272179e-06,
+      "loss": 18.4405,
+      "step": 10324
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.019329973143768e-06,
+      "loss": 0.9417,
+      "step": 10325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.012322153122887e-06,
+      "loss": 0.9345,
+      "step": 10326
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.005317129676242e-06,
+      "loss": 19.3435,
+      "step": 10327
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.99831490327041e-06,
+      "loss": 0.9076,
+      "step": 10328
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.991315474371797e-06,
+      "loss": 17.9843,
+      "step": 10329
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.984318843446593e-06,
+      "loss": 0.9553,
+      "step": 10330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.977325010960778e-06,
+      "loss": 20.451,
+      "step": 10331
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.970333977380184e-06,
+      "loss": 1.0306,
+      "step": 10332
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.963345743170476e-06,
+      "loss": 18.1439,
+      "step": 10333
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.956360308797062e-06,
+      "loss": 0.2085,
+      "step": 10334
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.949377674725217e-06,
+      "loss": 20.2056,
+      "step": 10335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.94239784142004e-06,
+      "loss": 0.9392,
+      "step": 10336
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.935420809346395e-06,
+      "loss": 17.6097,
+      "step": 10337
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.928446578968989e-06,
+      "loss": 18.0993,
+      "step": 10338
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.921475150752344e-06,
+      "loss": 18.786,
+      "step": 10339
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.914506525160786e-06,
+      "loss": 0.2042,
+      "step": 10340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.907540702658456e-06,
+      "loss": 18.2944,
+      "step": 10341
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.900577683709315e-06,
+      "loss": 19.7311,
+      "step": 10342
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.893617468777132e-06,
+      "loss": 0.9412,
+      "step": 10343
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.886660058325485e-06,
+      "loss": 19.7933,
+      "step": 10344
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.879705452817771e-06,
+      "loss": 19.2494,
+      "step": 10345
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.872753652717203e-06,
+      "loss": 0.1715,
+      "step": 10346
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.865804658486797e-06,
+      "loss": 18.2446,
+      "step": 10347
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.858858470589392e-06,
+      "loss": 20.8826,
+      "step": 10348
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.851915089487638e-06,
+      "loss": 0.8853,
+      "step": 10349
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.844974515643994e-06,
+      "loss": 18.5812,
+      "step": 10350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.838036749520733e-06,
+      "loss": 19.0951,
+      "step": 10351
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.831101791579948e-06,
+      "loss": 19.4367,
+      "step": 10352
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.824169642283536e-06,
+      "loss": 18.9203,
+      "step": 10353
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.81724030209321e-06,
+      "loss": 19.7378,
+      "step": 10354
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.81031377147049e-06,
+      "loss": 19.2539,
+      "step": 10355
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.80339005087673e-06,
+      "loss": 0.1775,
+      "step": 10356
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.796469140773067e-06,
+      "loss": 0.2019,
+      "step": 10357
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.789551041620475e-06,
+      "loss": 19.5824,
+      "step": 10358
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.78263575387973e-06,
+      "loss": 0.8871,
+      "step": 10359
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.775723278011421e-06,
+      "loss": 19.1992,
+      "step": 10360
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.768813614475957e-06,
+      "loss": 0.9433,
+      "step": 10361
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.761906763733529e-06,
+      "loss": 0.2291,
+      "step": 10362
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.755002726244192e-06,
+      "loss": 17.9323,
+      "step": 10363
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.748101502467797e-06,
+      "loss": 18.7949,
+      "step": 10364
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.741203092863952e-06,
+      "loss": 0.2092,
+      "step": 10365
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.734307497892162e-06,
+      "loss": 19.0093,
+      "step": 10366
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.727414718011704e-06,
+      "loss": 18.8532,
+      "step": 10367
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.72052475368164e-06,
+      "loss": 19.3664,
+      "step": 10368
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.713637605360901e-06,
+      "loss": 0.2212,
+      "step": 10369
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.706753273508194e-06,
+      "loss": 0.9815,
+      "step": 10370
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.699871758582067e-06,
+      "loss": 0.223,
+      "step": 10371
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.69299306104081e-06,
+      "loss": 0.1891,
+      "step": 10372
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.686117181342628e-06,
+      "loss": 0.1901,
+      "step": 10373
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.679244119945472e-06,
+      "loss": 0.9719,
+      "step": 10374
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.672373877307093e-06,
+      "loss": 0.9332,
+      "step": 10375
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.665506453885118e-06,
+      "loss": 19.8373,
+      "step": 10376
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.658641850136944e-06,
+      "loss": 18.6268,
+      "step": 10377
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.65178006651976e-06,
+      "loss": 17.8229,
+      "step": 10378
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.644921103490598e-06,
+      "loss": 19.2294,
+      "step": 10379
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.638064961506335e-06,
+      "loss": 19.251,
+      "step": 10380
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.631211641023578e-06,
+      "loss": 0.1907,
+      "step": 10381
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.624361142498792e-06,
+      "loss": 18.7666,
+      "step": 10382
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.617513466388293e-06,
+      "loss": 17.2903,
+      "step": 10383
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.610668613148131e-06,
+      "loss": 19.9961,
+      "step": 10384
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.603826583234197e-06,
+      "loss": 19.2332,
+      "step": 10385
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.596987377102243e-06,
+      "loss": 0.1692,
+      "step": 10386
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.590150995207762e-06,
+      "loss": 0.9669,
+      "step": 10387
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.583317438006093e-06,
+      "loss": 20.8571,
+      "step": 10388
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.576486705952385e-06,
+      "loss": 0.9127,
+      "step": 10389
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.5696587995016e-06,
+      "loss": 0.933,
+      "step": 10390
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.562833719108503e-06,
+      "loss": 0.2052,
+      "step": 10391
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.5560114652276635e-06,
+      "loss": 0.9395,
+      "step": 10392
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.549192038313513e-06,
+      "loss": 19.2153,
+      "step": 10393
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.54237543882022e-06,
+      "loss": 19.0414,
+      "step": 10394
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.535561667201818e-06,
+      "loss": 0.9251,
+      "step": 10395
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.528750723912137e-06,
+      "loss": 0.9034,
+      "step": 10396
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.521942609404809e-06,
+      "loss": 18.0295,
+      "step": 10397
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.515137324133287e-06,
+      "loss": 0.1917,
+      "step": 10398
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.508334868550843e-06,
+      "loss": 0.1849,
+      "step": 10399
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.501535243110546e-06,
+      "loss": 0.9223,
+      "step": 10400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.4947384482652875e-06,
+      "loss": 0.9695,
+      "step": 10401
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.487944484467757e-06,
+      "loss": 19.9621,
+      "step": 10402
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.481153352170473e-06,
+      "loss": 19.4396,
+      "step": 10403
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.474365051825749e-06,
+      "loss": 18.8633,
+      "step": 10404
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.46757958388572e-06,
+      "loss": 20.575,
+      "step": 10405
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.460796948802329e-06,
+      "loss": 0.19,
+      "step": 10406
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.454017147027337e-06,
+      "loss": 19.667,
+      "step": 10407
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.447240179012305e-06,
+      "loss": 16.5688,
+      "step": 10408
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.44046604520861e-06,
+      "loss": 17.5708,
+      "step": 10409
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.433694746067438e-06,
+      "loss": 0.9599,
+      "step": 10410
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.426926282039809e-06,
+      "loss": 19.0089,
+      "step": 10411
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.4201606535764966e-06,
+      "loss": 17.4171,
+      "step": 10412
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.413397861128152e-06,
+      "loss": 21.3056,
+      "step": 10413
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.406637905145203e-06,
+      "loss": 18.6338,
+      "step": 10414
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.399880786077889e-06,
+      "loss": 17.8393,
+      "step": 10415
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.393126504376274e-06,
+      "loss": 20.0167,
+      "step": 10416
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.386375060490213e-06,
+      "loss": 18.7069,
+      "step": 10417
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.379626454869404e-06,
+      "loss": 18.5063,
+      "step": 10418
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.372880687963296e-06,
+      "loss": 19.2322,
+      "step": 10419
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.366137760221231e-06,
+      "loss": 0.8733,
+      "step": 10420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.359397672092306e-06,
+      "loss": 19.1635,
+      "step": 10421
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.352660424025421e-06,
+      "loss": 18.2206,
+      "step": 10422
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.345926016469329e-06,
+      "loss": 0.9231,
+      "step": 10423
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.339194449872588e-06,
+      "loss": 19.0617,
+      "step": 10424
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.332465724683507e-06,
+      "loss": 0.9477,
+      "step": 10425
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.325739841350282e-06,
+      "loss": 19.196,
+      "step": 10426
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.319016800320899e-06,
+      "loss": 19.7063,
+      "step": 10427
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.312296602043112e-06,
+      "loss": 21.12,
+      "step": 10428
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.305579246964523e-06,
+      "loss": 0.9734,
+      "step": 10429
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.298864735532568e-06,
+      "loss": 19.1234,
+      "step": 10430
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.292153068194435e-06,
+      "loss": 0.9665,
+      "step": 10431
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.285444245397144e-06,
+      "loss": 18.8357,
+      "step": 10432
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.27873826758757e-06,
+      "loss": 0.947,
+      "step": 10433
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.272035135212352e-06,
+      "loss": 18.6173,
+      "step": 10434
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.265334848717931e-06,
+      "loss": 0.1994,
+      "step": 10435
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.258637408550578e-06,
+      "loss": 0.2047,
+      "step": 10436
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.251942815156404e-06,
+      "loss": 0.207,
+      "step": 10437
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.245251068981268e-06,
+      "loss": 0.1467,
+      "step": 10438
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.238562170470869e-06,
+      "loss": 17.4949,
+      "step": 10439
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.231876120070758e-06,
+      "loss": 19.1421,
+      "step": 10440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.225192918226214e-06,
+      "loss": 0.212,
+      "step": 10441
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.218512565382396e-06,
+      "loss": 0.1874,
+      "step": 10442
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.211835061984229e-06,
+      "loss": 19.8893,
+      "step": 10443
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.20516040847648e-06,
+      "loss": 17.9092,
+      "step": 10444
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.198488605303699e-06,
+      "loss": 21.4149,
+      "step": 10445
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.19181965291027e-06,
+      "loss": 20.6735,
+      "step": 10446
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.18515355174037e-06,
+      "loss": 18.6356,
+      "step": 10447
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.178490302237995e-06,
+      "loss": 21.0554,
+      "step": 10448
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.17182990484695e-06,
+      "loss": 20.3264,
+      "step": 10449
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.165172360010841e-06,
+      "loss": 0.1787,
+      "step": 10450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.158517668173098e-06,
+      "loss": 0.9043,
+      "step": 10451
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.151865829776955e-06,
+      "loss": 0.9184,
+      "step": 10452
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.1452168452654515e-06,
+      "loss": 18.3597,
+      "step": 10453
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.138570715081438e-06,
+      "loss": 20.2465,
+      "step": 10454
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.131927439667585e-06,
+      "loss": 19.7691,
+      "step": 10455
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.125287019466364e-06,
+      "loss": 0.1952,
+      "step": 10456
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.1186494549200565e-06,
+      "loss": 19.135,
+      "step": 10457
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.112014746470752e-06,
+      "loss": 19.215,
+      "step": 10458
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.105382894560359e-06,
+      "loss": 18.2863,
+      "step": 10459
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.098753899630578e-06,
+      "loss": 0.9043,
+      "step": 10460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.092127762122947e-06,
+      "loss": 19.2359,
+      "step": 10461
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.085504482478789e-06,
+      "loss": 20.3638,
+      "step": 10462
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.07888406113924e-06,
+      "loss": 0.1977,
+      "step": 10463
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.072266498545266e-06,
+      "loss": 18.7043,
+      "step": 10464
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.065651795137623e-06,
+      "loss": 20.3772,
+      "step": 10465
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.059039951356855e-06,
+      "loss": 0.9606,
+      "step": 10466
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.05243096764337e-06,
+      "loss": 0.9247,
+      "step": 10467
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.045824844437366e-06,
+      "loss": 20.743,
+      "step": 10468
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.039221582178795e-06,
+      "loss": 0.1725,
+      "step": 10469
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.0326211813075135e-06,
+      "loss": 17.9896,
+      "step": 10470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.02602364226313e-06,
+      "loss": 18.1112,
+      "step": 10471
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.0194289654850365e-06,
+      "loss": 22.0363,
+      "step": 10472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.012837151412505e-06,
+      "loss": 0.9495,
+      "step": 10473
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.006248200484583e-06,
+      "loss": 0.2031,
+      "step": 10474
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.999662113140099e-06,
+      "loss": 0.2033,
+      "step": 10475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.993078889817717e-06,
+      "loss": 18.6263,
+      "step": 10476
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.986498530955937e-06,
+      "loss": 0.9581,
+      "step": 10477
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.979921036993042e-06,
+      "loss": 20.8289,
+      "step": 10478
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.973346408367082e-06,
+      "loss": 17.7077,
+      "step": 10479
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.966774645516e-06,
+      "loss": 0.1695,
+      "step": 10480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.9602057488774965e-06,
+      "loss": 17.0021,
+      "step": 10481
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.953639718889076e-06,
+      "loss": 0.98,
+      "step": 10482
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.947076555988069e-06,
+      "loss": 0.9818,
+      "step": 10483
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.94051626061164e-06,
+      "loss": 0.9351,
+      "step": 10484
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.933958833196702e-06,
+      "loss": 20.0027,
+      "step": 10485
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.927404274180011e-06,
+      "loss": 0.9137,
+      "step": 10486
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.920852583998166e-06,
+      "loss": 0.9576,
+      "step": 10487
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.914303763087504e-06,
+      "loss": 0.2051,
+      "step": 10488
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.907757811884208e-06,
+      "loss": 18.8053,
+      "step": 10489
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.901214730824301e-06,
+      "loss": 20.2304,
+      "step": 10490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8946745203435525e-06,
+      "loss": 0.2298,
+      "step": 10491
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8881371808775794e-06,
+      "loss": 0.2022,
+      "step": 10492
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.881602712861801e-06,
+      "loss": 0.9589,
+      "step": 10493
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.875071116731441e-06,
+      "loss": 0.9262,
+      "step": 10494
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.868542392921535e-06,
+      "loss": 0.9261,
+      "step": 10495
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8620165418669224e-06,
+      "loss": 19.6668,
+      "step": 10496
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.855493564002263e-06,
+      "loss": 19.8129,
+      "step": 10497
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.848973459762015e-06,
+      "loss": 0.9644,
+      "step": 10498
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.842456229580441e-06,
+      "loss": 0.1804,
+      "step": 10499
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.835941873891627e-06,
+      "loss": 0.1676,
+      "step": 10500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.829430393129454e-06,
+      "loss": 20.4839,
+      "step": 10501
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.822921787727621e-06,
+      "loss": 0.9536,
+      "step": 10502
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.816416058119623e-06,
+      "loss": 0.9625,
+      "step": 10503
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.809913204738788e-06,
+      "loss": 0.918,
+      "step": 10504
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.803413228018218e-06,
+      "loss": 18.7201,
+      "step": 10505
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.796916128390851e-06,
+      "loss": 19.1643,
+      "step": 10506
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.790421906289423e-06,
+      "loss": 0.2091,
+      "step": 10507
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.783930562146484e-06,
+      "loss": 0.9253,
+      "step": 10508
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.777442096394376e-06,
+      "loss": 0.2221,
+      "step": 10509
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.770956509465271e-06,
+      "loss": 0.1835,
+      "step": 10510
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.764473801791138e-06,
+      "loss": 0.1864,
+      "step": 10511
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.75799397380375e-06,
+      "loss": 0.2127,
+      "step": 10512
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.751517025934701e-06,
+      "loss": 19.0457,
+      "step": 10513
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.745042958615377e-06,
+      "loss": 17.933,
+      "step": 10514
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.738571772276997e-06,
+      "loss": 0.2174,
+      "step": 10515
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.732103467350536e-06,
+      "loss": 19.5187,
+      "step": 10516
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.725638044266858e-06,
+      "loss": 17.913,
+      "step": 10517
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.719175503456565e-06,
+      "loss": 18.2157,
+      "step": 10518
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.712715845350093e-06,
+      "loss": 0.1736,
+      "step": 10519
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.706259070377696e-06,
+      "loss": 17.6078,
+      "step": 10520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.6998051789694145e-06,
+      "loss": 19.9771,
+      "step": 10521
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.693354171555128e-06,
+      "loss": 19.3702,
+      "step": 10522
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.686906048564462e-06,
+      "loss": 1.0115,
+      "step": 10523
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.680460810426925e-06,
+      "loss": 18.8739,
+      "step": 10524
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.674018457571807e-06,
+      "loss": 0.929,
+      "step": 10525
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.667578990428158e-06,
+      "loss": 0.2437,
+      "step": 10526
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.6611424094249146e-06,
+      "loss": 17.1721,
+      "step": 10527
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.654708714990781e-06,
+      "loss": 20.9133,
+      "step": 10528
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.648277907554235e-06,
+      "loss": 17.6979,
+      "step": 10529
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.641849987543636e-06,
+      "loss": 0.9286,
+      "step": 10530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.635424955387115e-06,
+      "loss": 0.9348,
+      "step": 10531
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.62900281151258e-06,
+      "loss": 18.0608,
+      "step": 10532
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.622583556347778e-06,
+      "loss": 17.8673,
+      "step": 10533
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.6161671903203e-06,
+      "loss": 17.6774,
+      "step": 10534
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.609753713857464e-06,
+      "loss": 19.4113,
+      "step": 10535
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.603343127386441e-06,
+      "loss": 18.7526,
+      "step": 10536
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.596935431334239e-06,
+      "loss": 0.9632,
+      "step": 10537
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.590530626127611e-06,
+      "loss": 0.1783,
+      "step": 10538
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.584128712193155e-06,
+      "loss": 0.1709,
+      "step": 10539
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.577729689957252e-06,
+      "loss": 0.9437,
+      "step": 10540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.571333559846143e-06,
+      "loss": 18.4079,
+      "step": 10541
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.564940322285812e-06,
+      "loss": 18.1458,
+      "step": 10542
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.558549977702078e-06,
+      "loss": 0.1705,
+      "step": 10543
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.5521625265205885e-06,
+      "loss": 0.2299,
+      "step": 10544
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.5457779691667555e-06,
+      "loss": 0.9693,
+      "step": 10545
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.539396306065832e-06,
+      "loss": 20.2819,
+      "step": 10546
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.53301753764286e-06,
+      "loss": 17.3269,
+      "step": 10547
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.5266416643226965e-06,
+      "loss": 20.6741,
+      "step": 10548
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.5202686865300065e-06,
+      "loss": 16.5512,
+      "step": 10549
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.513898604689261e-06,
+      "loss": 21.066,
+      "step": 10550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.507531419224739e-06,
+      "loss": 0.9255,
+      "step": 10551
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.501167130560515e-06,
+      "loss": 18.8077,
+      "step": 10552
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.494805739120485e-06,
+      "loss": 20.2781,
+      "step": 10553
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.4884472453283495e-06,
+      "loss": 0.1903,
+      "step": 10554
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.482091649607613e-06,
+      "loss": 0.1956,
+      "step": 10555
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.475738952381588e-06,
+      "loss": 0.9398,
+      "step": 10556
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.469389154073391e-06,
+      "loss": 0.9973,
+      "step": 10557
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.46304225510595e-06,
+      "loss": 21.1024,
+      "step": 10558
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.456698255901999e-06,
+      "loss": 0.9384,
+      "step": 10559
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.450357156884074e-06,
+      "loss": 20.4066,
+      "step": 10560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.444018958474524e-06,
+      "loss": 18.5192,
+      "step": 10561
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.437683661095506e-06,
+      "loss": 20.4573,
+      "step": 10562
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.4313512651689726e-06,
+      "loss": 0.9301,
+      "step": 10563
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.425021771116696e-06,
+      "loss": 0.9449,
+      "step": 10564
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.418695179360251e-06,
+      "loss": 0.2117,
+      "step": 10565
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.412371490321012e-06,
+      "loss": 0.2006,
+      "step": 10566
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.406050704420174e-06,
+      "loss": 0.9514,
+      "step": 10567
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.399732822078725e-06,
+      "loss": 17.4523,
+      "step": 10568
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.393417843717475e-06,
+      "loss": 0.1835,
+      "step": 10569
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.387105769757001e-06,
+      "loss": 18.6565,
+      "step": 10570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.38079660061775e-06,
+      "loss": 19.223,
+      "step": 10571
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.374490336719946e-06,
+      "loss": 17.1507,
+      "step": 10572
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.3681869784835705e-06,
+      "loss": 18.8014,
+      "step": 10573
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.361886526328498e-06,
+      "loss": 18.4991,
+      "step": 10574
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.355588980674365e-06,
+      "loss": 0.9739,
+      "step": 10575
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.349294341940593e-06,
+      "loss": 0.9184,
+      "step": 10576
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.34300261054645e-06,
+      "loss": 0.9395,
+      "step": 10577
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.336713786911008e-06,
+      "loss": 0.9642,
+      "step": 10578
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.33042787145311e-06,
+      "loss": 0.2176,
+      "step": 10579
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.324144864591419e-06,
+      "loss": 19.9377,
+      "step": 10580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.317864766744447e-06,
+      "loss": 18.4051,
+      "step": 10581
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.3115875783304524e-06,
+      "loss": 17.8335,
+      "step": 10582
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.305313299767518e-06,
+      "loss": 22.9642,
+      "step": 10583
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.299041931473565e-06,
+      "loss": 19.0535,
+      "step": 10584
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.292773473866293e-06,
+      "loss": 18.5983,
+      "step": 10585
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.286507927363189e-06,
+      "loss": 20.3907,
+      "step": 10586
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.2802452923815706e-06,
+      "loss": 19.0189,
+      "step": 10587
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.273985569338586e-06,
+      "loss": 0.2041,
+      "step": 10588
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.267728758651132e-06,
+      "loss": 0.1731,
+      "step": 10589
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.2614748607359445e-06,
+      "loss": 0.2007,
+      "step": 10590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.255223876009586e-06,
+      "loss": 18.0909,
+      "step": 10591
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.24897580488838e-06,
+      "loss": 19.8251,
+      "step": 10592
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.242730647788469e-06,
+      "loss": 0.193,
+      "step": 10593
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.236488405125845e-06,
+      "loss": 17.9778,
+      "step": 10594
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.230249077316236e-06,
+      "loss": 0.9743,
+      "step": 10595
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.224012664775219e-06,
+      "loss": 20.6231,
+      "step": 10596
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.217779167918176e-06,
+      "loss": 0.9493,
+      "step": 10597
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.211548587160276e-06,
+      "loss": 18.8868,
+      "step": 10598
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.2053209229165184e-06,
+      "loss": 0.2202,
+      "step": 10599
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.199096175601682e-06,
+      "loss": 0.9041,
+      "step": 10600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.192874345630367e-06,
+      "loss": 18.7763,
+      "step": 10601
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.1866554334169765e-06,
+      "loss": 19.6309,
+      "step": 10602
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.180439439375724e-06,
+      "loss": 18.1231,
+      "step": 10603
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.174226363920616e-06,
+      "loss": 18.4175,
+      "step": 10604
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.168016207465477e-06,
+      "loss": 0.1847,
+      "step": 10605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1618089704239325e-06,
+      "loss": 18.8025,
+      "step": 10606
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.155604653209407e-06,
+      "loss": 0.9709,
+      "step": 10607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.149403256235148e-06,
+      "loss": 18.4939,
+      "step": 10608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.143204779914185e-06,
+      "loss": 20.1985,
+      "step": 10609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.137009224659374e-06,
+      "loss": 0.892,
+      "step": 10610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.130816590883365e-06,
+      "loss": 0.91,
+      "step": 10611
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.124626878998618e-06,
+      "loss": 18.6156,
+      "step": 10612
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.118440089417393e-06,
+      "loss": 16.9839,
+      "step": 10613
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1122562225517595e-06,
+      "loss": 20.348,
+      "step": 10614
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.106075278813595e-06,
+      "loss": 0.9059,
+      "step": 10615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.099897258614573e-06,
+      "loss": 0.1626,
+      "step": 10616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.093722162366184e-06,
+      "loss": 0.2357,
+      "step": 10617
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.087549990479719e-06,
+      "loss": 18.5608,
+      "step": 10618
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.081380743366277e-06,
+      "loss": 0.8796,
+      "step": 10619
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0752144214367314e-06,
+      "loss": 0.975,
+      "step": 10620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.069051025101818e-06,
+      "loss": 20.0427,
+      "step": 10621
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.06289055477205e-06,
+      "loss": 0.2138,
+      "step": 10622
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.056733010857712e-06,
+      "loss": 20.0528,
+      "step": 10623
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.050578393768952e-06,
+      "loss": 0.9334,
+      "step": 10624
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.04442670391569e-06,
+      "loss": 0.8882,
+      "step": 10625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.03827794170767e-06,
+      "loss": 0.2073,
+      "step": 10626
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.032132107554389e-06,
+      "loss": 0.9425,
+      "step": 10627
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0259892018652235e-06,
+      "loss": 18.3489,
+      "step": 10628
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.019849225049317e-06,
+      "loss": 18.4145,
+      "step": 10629
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.013712177515596e-06,
+      "loss": 19.6026,
+      "step": 10630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0075780596728356e-06,
+      "loss": 19.3736,
+      "step": 10631
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.001446871929611e-06,
+      "loss": 0.9731,
+      "step": 10632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.995318614694245e-06,
+      "loss": 0.9244,
+      "step": 10633
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.989193288374945e-06,
+      "loss": 18.8588,
+      "step": 10634
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.983070893379689e-06,
+      "loss": 0.2261,
+      "step": 10635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.976951430116229e-06,
+      "loss": 20.4797,
+      "step": 10636
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.970834898992151e-06,
+      "loss": 18.9155,
+      "step": 10637
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.964721300414877e-06,
+      "loss": 18.9362,
+      "step": 10638
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.958610634791573e-06,
+      "loss": 0.2342,
+      "step": 10639
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.952502902529234e-06,
+      "loss": 0.9651,
+      "step": 10640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.946398104034695e-06,
+      "loss": 17.9785,
+      "step": 10641
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9402962397145326e-06,
+      "loss": 0.9206,
+      "step": 10642
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.934197309975165e-06,
+      "loss": 20.444,
+      "step": 10643
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.928101315222823e-06,
+      "loss": 0.1626,
+      "step": 10644
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.922008255863515e-06,
+      "loss": 0.2176,
+      "step": 10645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.915918132303072e-06,
+      "loss": 0.2132,
+      "step": 10646
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.909830944947114e-06,
+      "loss": 0.9361,
+      "step": 10647
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.903746694201107e-06,
+      "loss": 0.2121,
+      "step": 10648
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.897665380470257e-06,
+      "loss": 0.9656,
+      "step": 10649
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.891587004159621e-06,
+      "loss": 18.5024,
+      "step": 10650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.885511565674045e-06,
+      "loss": 0.9404,
+      "step": 10651
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.879439065418185e-06,
+      "loss": 20.0208,
+      "step": 10652
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.873369503796494e-06,
+      "loss": 0.1879,
+      "step": 10653
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.867302881213232e-06,
+      "loss": 0.9275,
+      "step": 10654
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.861239198072471e-06,
+      "loss": 20.6211,
+      "step": 10655
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.855178454778071e-06,
+      "loss": 0.9355,
+      "step": 10656
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.849120651733714e-06,
+      "loss": 0.9515,
+      "step": 10657
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.843065789342872e-06,
+      "loss": 18.0076,
+      "step": 10658
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.837013868008828e-06,
+      "loss": 0.225,
+      "step": 10659
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.830964888134671e-06,
+      "loss": 0.981,
+      "step": 10660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.82491885012329e-06,
+      "loss": 18.888,
+      "step": 10661
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.818875754377379e-06,
+      "loss": 0.8932,
+      "step": 10662
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8128356012994375e-06,
+      "loss": 20.4559,
+      "step": 10663
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.806798391291762e-06,
+      "loss": 0.9399,
+      "step": 10664
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.800764124756463e-06,
+      "loss": 0.9333,
+      "step": 10665
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.794732802095465e-06,
+      "loss": 0.9143,
+      "step": 10666
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7887044237104444e-06,
+      "loss": 19.6163,
+      "step": 10667
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.782678990002954e-06,
+      "loss": 18.2654,
+      "step": 10668
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7766565013743e-06,
+      "loss": 0.2108,
+      "step": 10669
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.770636958225617e-06,
+      "loss": 20.5301,
+      "step": 10670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.764620360957823e-06,
+      "loss": 19.5737,
+      "step": 10671
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.758606709971664e-06,
+      "loss": 17.4653,
+      "step": 10672
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.752596005667677e-06,
+      "loss": 18.4117,
+      "step": 10673
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.746588248446183e-06,
+      "loss": 0.9009,
+      "step": 10674
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.740583438707342e-06,
+      "loss": 0.9215,
+      "step": 10675
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.734581576851117e-06,
+      "loss": 19.6681,
+      "step": 10676
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.728582663277221e-06,
+      "loss": 0.9612,
+      "step": 10677
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.722586698385241e-06,
+      "loss": 0.9653,
+      "step": 10678
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.716593682574539e-06,
+      "loss": 19.5927,
+      "step": 10679
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.710603616244242e-06,
+      "loss": 17.7932,
+      "step": 10680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7046164997933525e-06,
+      "loss": 0.2462,
+      "step": 10681
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.698632333620635e-06,
+      "loss": 0.9405,
+      "step": 10682
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.692651118124648e-06,
+      "loss": 17.2074,
+      "step": 10683
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.686672853703767e-06,
+      "loss": 0.9451,
+      "step": 10684
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.680697540756197e-06,
+      "loss": 0.9902,
+      "step": 10685
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6747251796798964e-06,
+      "loss": 20.2753,
+      "step": 10686
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.668755770872652e-06,
+      "loss": 19.6915,
+      "step": 10687
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6627893147320755e-06,
+      "loss": 19.2144,
+      "step": 10688
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.656825811655553e-06,
+      "loss": 0.9207,
+      "step": 10689
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.650865262040272e-06,
+      "loss": 20.4465,
+      "step": 10690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.644907666283222e-06,
+      "loss": 20.2966,
+      "step": 10691
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.638953024781246e-06,
+      "loss": 0.248,
+      "step": 10692
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.633001337930916e-06,
+      "loss": 0.9426,
+      "step": 10693
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.627052606128641e-06,
+      "loss": 0.9147,
+      "step": 10694
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6211068297706725e-06,
+      "loss": 20.001,
+      "step": 10695
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.615164009252982e-06,
+      "loss": 18.8822,
+      "step": 10696
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6092241449714036e-06,
+      "loss": 0.1616,
+      "step": 10697
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.603287237321581e-06,
+      "loss": 0.1667,
+      "step": 10698
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.597353286698909e-06,
+      "loss": 19.4474,
+      "step": 10699
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.591422293498633e-06,
+      "loss": 19.9893,
+      "step": 10700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.585494258115781e-06,
+      "loss": 21.1519,
+      "step": 10701
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5795691809451875e-06,
+      "loss": 0.9465,
+      "step": 10702
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.573647062381488e-06,
+      "loss": 17.1822,
+      "step": 10703
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.567727902819131e-06,
+      "loss": 0.2175,
+      "step": 10704
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.561811702652348e-06,
+      "loss": 18.2665,
+      "step": 10705
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.555898462275194e-06,
+      "loss": 0.1877,
+      "step": 10706
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.549988182081523e-06,
+      "loss": 20.3528,
+      "step": 10707
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.544080862464973e-06,
+      "loss": 19.4824,
+      "step": 10708
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5381765038190095e-06,
+      "loss": 17.9408,
+      "step": 10709
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.532275106536883e-06,
+      "loss": 0.9263,
+      "step": 10710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.526376671011663e-06,
+      "loss": 18.4371,
+      "step": 10711
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5204811976362016e-06,
+      "loss": 0.899,
+      "step": 10712
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.514588686803174e-06,
+      "loss": 20.6517,
+      "step": 10713
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.508699138905049e-06,
+      "loss": 0.1966,
+      "step": 10714
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5028125543340915e-06,
+      "loss": 17.9723,
+      "step": 10715
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.496928933482376e-06,
+      "loss": 17.6464,
+      "step": 10716
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.491048276741784e-06,
+      "loss": 19.9736,
+      "step": 10717
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.48517058450399e-06,
+      "loss": 0.1713,
+      "step": 10718
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.479295857160477e-06,
+      "loss": 17.4536,
+      "step": 10719
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.473424095102531e-06,
+      "loss": 18.9159,
+      "step": 10720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.467555298721233e-06,
+      "loss": 0.2101,
+      "step": 10721
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.461689468407483e-06,
+      "loss": 18.3818,
+      "step": 10722
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.455826604551972e-06,
+      "loss": 17.3627,
+      "step": 10723
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.449966707545167e-06,
+      "loss": 0.1893,
+      "step": 10724
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.444109777777395e-06,
+      "loss": 18.6425,
+      "step": 10725
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.438255815638754e-06,
+      "loss": 0.1973,
+      "step": 10726
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.432404821519116e-06,
+      "loss": 0.8968,
+      "step": 10727
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.426556795808213e-06,
+      "loss": 18.7921,
+      "step": 10728
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.420711738895551e-06,
+      "loss": 0.9381,
+      "step": 10729
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.414869651170424e-06,
+      "loss": 18.3234,
+      "step": 10730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.4090305330219304e-06,
+      "loss": 0.9229,
+      "step": 10731
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.403194384839011e-06,
+      "loss": 17.4931,
+      "step": 10732
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.397361207010382e-06,
+      "loss": 0.9431,
+      "step": 10733
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3915309999245215e-06,
+      "loss": 0.2169,
+      "step": 10734
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.385703763969785e-06,
+      "loss": 19.4538,
+      "step": 10735
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.379879499534291e-06,
+      "loss": 19.5291,
+      "step": 10736
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.374058207005944e-06,
+      "loss": 0.974,
+      "step": 10737
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.368239886772469e-06,
+      "loss": 19.127,
+      "step": 10738
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.362424539221422e-06,
+      "loss": 0.1808,
+      "step": 10739
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3566121647401045e-06,
+      "loss": 0.942,
+      "step": 10740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.350802763715646e-06,
+      "loss": 0.9098,
+      "step": 10741
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.344996336535002e-06,
+      "loss": 19.8717,
+      "step": 10742
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3391928835848925e-06,
+      "loss": 0.9162,
+      "step": 10743
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3333924052518465e-06,
+      "loss": 18.6467,
+      "step": 10744
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.327594901922228e-06,
+      "loss": 19.8316,
+      "step": 10745
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.321800373982155e-06,
+      "loss": 19.9076,
+      "step": 10746
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.316008821817581e-06,
+      "loss": 19.2936,
+      "step": 10747
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.31022024581424e-06,
+      "loss": 18.8376,
+      "step": 10748
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.304434646357692e-06,
+      "loss": 0.9902,
+      "step": 10749
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.298652023833273e-06,
+      "loss": 0.9405,
+      "step": 10750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2928723786261404e-06,
+      "loss": 18.7064,
+      "step": 10751
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2870957111212426e-06,
+      "loss": 0.1828,
+      "step": 10752
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.281322021703328e-06,
+      "loss": 17.5574,
+      "step": 10753
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.27555131075696e-06,
+      "loss": 19.1719,
+      "step": 10754
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.269783578666487e-06,
+      "loss": 20.4783,
+      "step": 10755
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2640188258160696e-06,
+      "loss": 0.9259,
+      "step": 10756
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.258257052589666e-06,
+      "loss": 19.1055,
+      "step": 10757
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.252498259371042e-06,
+      "loss": 0.2033,
+      "step": 10758
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.246742446543751e-06,
+      "loss": 19.9931,
+      "step": 10759
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.240989614491165e-06,
+      "loss": 0.911,
+      "step": 10760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.235239763596445e-06,
+      "loss": 0.2018,
+      "step": 10761
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.22949289424256e-06,
+      "loss": 0.9065,
+      "step": 10762
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.223749006812279e-06,
+      "loss": 0.2267,
+      "step": 10763
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.218008101688171e-06,
+      "loss": 19.8278,
+      "step": 10764
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.212270179252604e-06,
+      "loss": 0.1856,
+      "step": 10765
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2065352398877495e-06,
+      "loss": 0.9482,
+      "step": 10766
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.200803283975586e-06,
+      "loss": 0.9406,
+      "step": 10767
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.195074311897885e-06,
+      "loss": 20.1956,
+      "step": 10768
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.189348324036225e-06,
+      "loss": 0.2492,
+      "step": 10769
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.18362532077199e-06,
+      "loss": 0.955,
+      "step": 10770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.177905302486336e-06,
+      "loss": 16.8591,
+      "step": 10771
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.172188269560257e-06,
+      "loss": 0.189,
+      "step": 10772
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1664742223745456e-06,
+      "loss": 18.0361,
+      "step": 10773
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.160763161309767e-06,
+      "loss": 0.21,
+      "step": 10774
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.155055086746308e-06,
+      "loss": 19.9477,
+      "step": 10775
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1493499990643565e-06,
+      "loss": 0.9422,
+      "step": 10776
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.143647898643911e-06,
+      "loss": 0.924,
+      "step": 10777
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.137948785864715e-06,
+      "loss": 17.7626,
+      "step": 10778
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.132252661106402e-06,
+      "loss": 19.0191,
+      "step": 10779
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.126559524748348e-06,
+      "loss": 19.239,
+      "step": 10780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.120869377169712e-06,
+      "loss": 0.9091,
+      "step": 10781
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.115182218749526e-06,
+      "loss": 20.0313,
+      "step": 10782
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.109498049866568e-06,
+      "loss": 19.2188,
+      "step": 10783
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.103816870899414e-06,
+      "loss": 18.2878,
+      "step": 10784
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.098138682226472e-06,
+      "loss": 18.1622,
+      "step": 10785
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.092463484225945e-06,
+      "loss": 19.6376,
+      "step": 10786
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.086791277275804e-06,
+      "loss": 19.1019,
+      "step": 10787
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.081122061753846e-06,
+      "loss": 17.8626,
+      "step": 10788
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.075455838037696e-06,
+      "loss": 18.9613,
+      "step": 10789
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.06979260650472e-06,
+      "loss": 0.904,
+      "step": 10790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.06413236753212e-06,
+      "loss": 0.1884,
+      "step": 10791
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.058475121496919e-06,
+      "loss": 0.9346,
+      "step": 10792
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.052820868775887e-06,
+      "loss": 0.2015,
+      "step": 10793
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.047169609745639e-06,
+      "loss": 0.9236,
+      "step": 10794
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.041521344782557e-06,
+      "loss": 19.1914,
+      "step": 10795
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.035876074262874e-06,
+      "loss": 0.1969,
+      "step": 10796
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.030233798562567e-06,
+      "loss": 19.5248,
+      "step": 10797
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.02459451805743e-06,
+      "loss": 17.8298,
+      "step": 10798
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.018958233123094e-06,
+      "loss": 18.1963,
+      "step": 10799
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.01332494413494e-06,
+      "loss": 18.341,
+      "step": 10800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.00769465146817e-06,
+      "loss": 18.3684,
+      "step": 10801
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.002067355497808e-06,
+      "loss": 20.4313,
+      "step": 10802
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.996443056598643e-06,
+      "loss": 0.1679,
+      "step": 10803
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9908217551452754e-06,
+      "loss": 0.9037,
+      "step": 10804
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.985203451512116e-06,
+      "loss": 0.9712,
+      "step": 10805
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.979588146073371e-06,
+      "loss": 0.9576,
+      "step": 10806
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9739758392030485e-06,
+      "loss": 0.8956,
+      "step": 10807
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.968366531274943e-06,
+      "loss": 0.9172,
+      "step": 10808
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.962760222662671e-06,
+      "loss": 0.1709,
+      "step": 10809
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9571569137396356e-06,
+      "loss": 18.9354,
+      "step": 10810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.951556604879048e-06,
+      "loss": 0.9428,
+      "step": 10811
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.945959296453906e-06,
+      "loss": 0.9218,
+      "step": 10812
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.940364988837015e-06,
+      "loss": 20.1631,
+      "step": 10813
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9347736824009945e-06,
+      "loss": 19.9086,
+      "step": 10814
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.929185377518236e-06,
+      "loss": 0.9288,
+      "step": 10815
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9236000745609635e-06,
+      "loss": 19.7768,
+      "step": 10816
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.918017773901168e-06,
+      "loss": 0.2112,
+      "step": 10817
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.912438475910669e-06,
+      "loss": 18.0346,
+      "step": 10818
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.906862180961064e-06,
+      "loss": 0.9409,
+      "step": 10819
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.901288889423766e-06,
+      "loss": 17.857,
+      "step": 10820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.89571860166998e-06,
+      "loss": 18.2449,
+      "step": 10821
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.890151318070713e-06,
+      "loss": 19.2112,
+      "step": 10822
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.884587038996774e-06,
+      "loss": 20.898,
+      "step": 10823
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8790257648187675e-06,
+      "loss": 19.2382,
+      "step": 10824
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.873467495907102e-06,
+      "loss": 0.2004,
+      "step": 10825
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.867912232631982e-06,
+      "loss": 19.0584,
+      "step": 10826
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.862359975363428e-06,
+      "loss": 0.932,
+      "step": 10827
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.856810724471211e-06,
+      "loss": 0.2027,
+      "step": 10828
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.851264480324969e-06,
+      "loss": 19.0408,
+      "step": 10829
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.845721243294104e-06,
+      "loss": 17.767,
+      "step": 10830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8401810137478025e-06,
+      "loss": 18.7353,
+      "step": 10831
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8346437920550824e-06,
+      "loss": 18.9135,
+      "step": 10832
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.829109578584762e-06,
+      "loss": 17.9184,
+      "step": 10833
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.823578373705423e-06,
+      "loss": 18.5946,
+      "step": 10834
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.818050177785466e-06,
+      "loss": 17.8426,
+      "step": 10835
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.812524991193123e-06,
+      "loss": 18.8454,
+      "step": 10836
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8070028142963744e-06,
+      "loss": 0.9308,
+      "step": 10837
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.801483647463017e-06,
+      "loss": 19.0746,
+      "step": 10838
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.795967491060677e-06,
+      "loss": 20.2461,
+      "step": 10839
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7904543454567515e-06,
+      "loss": 18.0646,
+      "step": 10840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.784944211018421e-06,
+      "loss": 20.1682,
+      "step": 10841
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.779437088112698e-06,
+      "loss": 1.0188,
+      "step": 10842
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.773932977106393e-06,
+      "loss": 17.2969,
+      "step": 10843
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.768431878366092e-06,
+      "loss": 0.1769,
+      "step": 10844
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.762933792258184e-06,
+      "loss": 20.0892,
+      "step": 10845
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.757438719148899e-06,
+      "loss": 17.0479,
+      "step": 10846
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.75194665940421e-06,
+      "loss": 18.6141,
+      "step": 10847
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.746457613389904e-06,
+      "loss": 0.9787,
+      "step": 10848
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7409715814716145e-06,
+      "loss": 18.9363,
+      "step": 10849
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7354885640147064e-06,
+      "loss": 17.1031,
+      "step": 10850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.730008561384381e-06,
+      "loss": 19.7063,
+      "step": 10851
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.724531573945634e-06,
+      "loss": 18.9471,
+      "step": 10852
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.719057602063259e-06,
+      "loss": 0.2081,
+      "step": 10853
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.713586646101842e-06,
+      "loss": 17.549,
+      "step": 10854
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7081187064257795e-06,
+      "loss": 19.0037,
+      "step": 10855
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.702653783399264e-06,
+      "loss": 0.9584,
+      "step": 10856
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.697191877386276e-06,
+      "loss": 0.971,
+      "step": 10857
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.691732988750614e-06,
+      "loss": 0.9297,
+      "step": 10858
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6862771178558525e-06,
+      "loss": 0.9775,
+      "step": 10859
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.680824265065392e-06,
+      "loss": 0.1998,
+      "step": 10860
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.675374430742413e-06,
+      "loss": 19.587,
+      "step": 10861
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6699276152498915e-06,
+      "loss": 0.9817,
+      "step": 10862
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.664483818950621e-06,
+      "loss": 0.8986,
+      "step": 10863
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.659043042207184e-06,
+      "loss": 0.1535,
+      "step": 10864
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.653605285381951e-06,
+      "loss": 17.3476,
+      "step": 10865
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.648170548837111e-06,
+      "loss": 20.5801,
+      "step": 10866
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.64273883293464e-06,
+      "loss": 17.4457,
+      "step": 10867
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.637310138036316e-06,
+      "loss": 18.8537,
+      "step": 10868
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.631884464503716e-06,
+      "loss": 16.8137,
+      "step": 10869
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.626461812698213e-06,
+      "loss": 0.942,
+      "step": 10870
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.621042182980978e-06,
+      "loss": 17.8547,
+      "step": 10871
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.61562557571299e-06,
+      "loss": 0.1838,
+      "step": 10872
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.61021199125502e-06,
+      "loss": 0.9246,
+      "step": 10873
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.604801429967642e-06,
+      "loss": 0.9189,
+      "step": 10874
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5993938922112016e-06,
+      "loss": 17.6927,
+      "step": 10875
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5939893783458865e-06,
+      "loss": 18.2512,
+      "step": 10876
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.588587888731671e-06,
+      "loss": 19.2231,
+      "step": 10877
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.583189423728285e-06,
+      "loss": 17.5014,
+      "step": 10878
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.577793983695322e-06,
+      "loss": 0.9266,
+      "step": 10879
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.572401568992135e-06,
+      "loss": 0.1684,
+      "step": 10880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.567012179977892e-06,
+      "loss": 20.4541,
+      "step": 10881
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.561625817011523e-06,
+      "loss": 20.4776,
+      "step": 10882
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.556242480451816e-06,
+      "loss": 18.3372,
+      "step": 10883
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.550862170657322e-06,
+      "loss": 17.4818,
+      "step": 10884
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.545484887986368e-06,
+      "loss": 19.3605,
+      "step": 10885
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.540110632797135e-06,
+      "loss": 0.2451,
+      "step": 10886
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.534739405447569e-06,
+      "loss": 0.9294,
+      "step": 10887
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.529371206295396e-06,
+      "loss": 0.8904,
+      "step": 10888
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.524006035698192e-06,
+      "loss": 0.9107,
+      "step": 10889
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.518643894013291e-06,
+      "loss": 0.968,
+      "step": 10890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.513284781597838e-06,
+      "loss": 19.1481,
+      "step": 10891
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.507928698808756e-06,
+      "loss": 19.3204,
+      "step": 10892
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.502575646002815e-06,
+      "loss": 0.2395,
+      "step": 10893
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.497225623536539e-06,
+      "loss": 18.9049,
+      "step": 10894
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.49187863176625e-06,
+      "loss": 19.3981,
+      "step": 10895
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.48653467104812e-06,
+      "loss": 0.9611,
+      "step": 10896
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.481193741738049e-06,
+      "loss": 0.9819,
+      "step": 10897
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4758558441917735e-06,
+      "loss": 17.6899,
+      "step": 10898
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.47052097876483e-06,
+      "loss": 19.2907,
+      "step": 10899
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4651891458125426e-06,
+      "loss": 19.7269,
+      "step": 10900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.459860345690031e-06,
+      "loss": 16.4062,
+      "step": 10901
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.454534578752217e-06,
+      "loss": 0.9349,
+      "step": 10902
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.44921184535384e-06,
+      "loss": 19.061,
+      "step": 10903
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.443892145849398e-06,
+      "loss": 20.9758,
+      "step": 10904
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.43857548059321e-06,
+      "loss": 0.9003,
+      "step": 10905
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.433261849939413e-06,
+      "loss": 0.2117,
+      "step": 10906
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.427951254241886e-06,
+      "loss": 19.2444,
+      "step": 10907
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.422643693854361e-06,
+      "loss": 0.1812,
+      "step": 10908
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4173391691303416e-06,
+      "loss": 17.9,
+      "step": 10909
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.412037680423131e-06,
+      "loss": 19.2274,
+      "step": 10910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4067392280858325e-06,
+      "loss": 0.2214,
+      "step": 10911
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.401443812471351e-06,
+      "loss": 20.6839,
+      "step": 10912
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.396151433932383e-06,
+      "loss": 0.2117,
+      "step": 10913
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3908620928214296e-06,
+      "loss": 18.268,
+      "step": 10914
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.385575789490781e-06,
+      "loss": 19.4509,
+      "step": 10915
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3802925242925345e-06,
+      "loss": 19.1064,
+      "step": 10916
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.375012297578573e-06,
+      "loss": 18.7579,
+      "step": 10917
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.36973510970059e-06,
+      "loss": 19.7288,
+      "step": 10918
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.364460961010069e-06,
+      "loss": 0.2363,
+      "step": 10919
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.359189851858297e-06,
+      "loss": 0.8925,
+      "step": 10920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.353921782596354e-06,
+      "loss": 17.9268,
+      "step": 10921
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.348656753575092e-06,
+      "loss": 18.5335,
+      "step": 10922
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.343394765145226e-06,
+      "loss": 0.9205,
+      "step": 10923
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.338135817657202e-06,
+      "loss": 17.1525,
+      "step": 10924
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.332879911461307e-06,
+      "loss": 19.9198,
+      "step": 10925
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.327627046907595e-06,
+      "loss": 21.0516,
+      "step": 10926
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.322377224345942e-06,
+      "loss": 18.5655,
+      "step": 10927
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.317130444126005e-06,
+      "loss": 0.9497,
+      "step": 10928
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3118867065972455e-06,
+      "loss": 0.9311,
+      "step": 10929
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.306646012108917e-06,
+      "loss": 0.9466,
+      "step": 10930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.301408361010084e-06,
+      "loss": 0.9432,
+      "step": 10931
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.296173753649579e-06,
+      "loss": 0.9384,
+      "step": 10932
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.290942190376068e-06,
+      "loss": 0.9323,
+      "step": 10933
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.285713671538005e-06,
+      "loss": 19.8163,
+      "step": 10934
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.280488197483606e-06,
+      "loss": 0.9005,
+      "step": 10935
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.275265768560932e-06,
+      "loss": 19.2062,
+      "step": 10936
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.270046385117826e-06,
+      "loss": 0.9317,
+      "step": 10937
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.264830047501906e-06,
+      "loss": 18.9934,
+      "step": 10938
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.259616756060603e-06,
+      "loss": 19.065,
+      "step": 10939
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.254406511141173e-06,
+      "loss": 0.8913,
+      "step": 10940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.249199313090618e-06,
+      "loss": 0.9382,
+      "step": 10941
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2439951622557535e-06,
+      "loss": 0.1717,
+      "step": 10942
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.238794058983242e-06,
+      "loss": 1.0126,
+      "step": 10943
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.233596003619455e-06,
+      "loss": 19.4266,
+      "step": 10944
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2284009965106345e-06,
+      "loss": 0.9026,
+      "step": 10945
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.223209038002768e-06,
+      "loss": 16.5461,
+      "step": 10946
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.218020128441702e-06,
+      "loss": 0.1987,
+      "step": 10947
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.212834268173011e-06,
+      "loss": 0.957,
+      "step": 10948
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2076514575421e-06,
+      "loss": 0.9533,
+      "step": 10949
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.202471696894184e-06,
+      "loss": 19.9585,
+      "step": 10950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.197294986574246e-06,
+      "loss": 19.4984,
+      "step": 10951
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.192121326927073e-06,
+      "loss": 19.204,
+      "step": 10952
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.186950718297283e-06,
+      "loss": 20.971,
+      "step": 10953
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.181783161029234e-06,
+      "loss": 0.9,
+      "step": 10954
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.176618655467118e-06,
+      "loss": 0.984,
+      "step": 10955
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.17145720195492e-06,
+      "loss": 18.5891,
+      "step": 10956
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1662988008364104e-06,
+      "loss": 20.9748,
+      "step": 10957
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.161143452455163e-06,
+      "loss": 1.0115,
+      "step": 10958
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.155991157154554e-06,
+      "loss": 0.2058,
+      "step": 10959
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.150841915277742e-06,
+      "loss": 17.4546,
+      "step": 10960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1456957271676975e-06,
+      "loss": 0.9297,
+      "step": 10961
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.14055259316718e-06,
+      "loss": 0.9585,
+      "step": 10962
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.135412513618736e-06,
+      "loss": 19.441,
+      "step": 10963
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.130275488864732e-06,
+      "loss": 0.208,
+      "step": 10964
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.125141519247311e-06,
+      "loss": 0.905,
+      "step": 10965
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.120010605108421e-06,
+      "loss": 0.9705,
+      "step": 10966
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.114882746789806e-06,
+      "loss": 0.9423,
+      "step": 10967
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.109757944633003e-06,
+      "loss": 18.9086,
+      "step": 10968
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1046361989793444e-06,
+      "loss": 0.8619,
+      "step": 10969
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0995175101699644e-06,
+      "loss": 19.045,
+      "step": 10970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0944018785458005e-06,
+      "loss": 18.1882,
+      "step": 10971
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.089289304447563e-06,
+      "loss": 18.8516,
+      "step": 10972
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0841797882157805e-06,
+      "loss": 0.9511,
+      "step": 10973
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.079073330190769e-06,
+      "loss": 20.2052,
+      "step": 10974
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0739699307126455e-06,
+      "loss": 18.7877,
+      "step": 10975
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0688695901213215e-06,
+      "loss": 19.8292,
+      "step": 10976
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.063772308756492e-06,
+      "loss": 19.9313,
+      "step": 10977
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.058678086957679e-06,
+      "loss": 19.7222,
+      "step": 10978
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.053586925064151e-06,
+      "loss": 18.9719,
+      "step": 10979
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.048498823415037e-06,
+      "loss": 18.1323,
+      "step": 10980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.043413782349214e-06,
+      "loss": 0.1884,
+      "step": 10981
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.038331802205353e-06,
+      "loss": 19.5686,
+      "step": 10982
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.033252883321958e-06,
+      "loss": 0.9543,
+      "step": 10983
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.028177026037316e-06,
+      "loss": 0.9513,
+      "step": 10984
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.023104230689478e-06,
+      "loss": 0.197,
+      "step": 10985
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0180344976163245e-06,
+      "loss": 19.5928,
+      "step": 10986
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.012967827155529e-06,
+      "loss": 0.9398,
+      "step": 10987
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.007904219644559e-06,
+      "loss": 20.0407,
+      "step": 10988
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.002843675420659e-06,
+      "loss": 0.9279,
+      "step": 10989
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.997786194820896e-06,
+      "loss": 20.0611,
+      "step": 10990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.99273177818213e-06,
+      "loss": 20.142,
+      "step": 10991
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.987680425840978e-06,
+      "loss": 19.6426,
+      "step": 10992
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.982632138133913e-06,
+      "loss": 0.1869,
+      "step": 10993
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9775869153971745e-06,
+      "loss": 0.9243,
+      "step": 10994
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.972544757966778e-06,
+      "loss": 0.2219,
+      "step": 10995
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.967505666178556e-06,
+      "loss": 20.1152,
+      "step": 10996
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.962469640368166e-06,
+      "loss": 0.961,
+      "step": 10997
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.957436680870996e-06,
+      "loss": 0.9372,
+      "step": 10998
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.952406788022267e-06,
+      "loss": 0.2165,
+      "step": 10999
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.947379962157022e-06,
+      "loss": 0.9538,
+      "step": 11000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.942356203610043e-06,
+      "loss": 19.9801,
+      "step": 11001
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.937335512715945e-06,
+      "loss": 0.1703,
+      "step": 11002
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.932317889809129e-06,
+      "loss": 19.1208,
+      "step": 11003
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.927303335223792e-06,
+      "loss": 0.979,
+      "step": 11004
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.92229184929393e-06,
+      "loss": 21.7113,
+      "step": 11005
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.917283432353324e-06,
+      "loss": 0.1991,
+      "step": 11006
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9122780847355654e-06,
+      "loss": 0.9561,
+      "step": 11007
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.90727580677403e-06,
+      "loss": 19.8438,
+      "step": 11008
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.90227659880189e-06,
+      "loss": 18.813,
+      "step": 11009
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.89728046115212e-06,
+      "loss": 18.7881,
+      "step": 11010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.892287394157484e-06,
+      "loss": 0.9604,
+      "step": 11011
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.887297398150547e-06,
+      "loss": 19.6208,
+      "step": 11012
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.882310473463658e-06,
+      "loss": 0.1943,
+      "step": 11013
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.877326620428978e-06,
+      "loss": 0.8871,
+      "step": 11014
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.872345839378449e-06,
+      "loss": 20.6121,
+      "step": 11015
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.867368130643817e-06,
+      "loss": 18.7649,
+      "step": 11016
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.862393494556615e-06,
+      "loss": 19.178,
+      "step": 11017
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.857421931448191e-06,
+      "loss": 0.9441,
+      "step": 11018
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8524534416496564e-06,
+      "loss": 0.1618,
+      "step": 11019
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.847488025491947e-06,
+      "loss": 18.575,
+      "step": 11020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.842525683305781e-06,
+      "loss": 21.0563,
+      "step": 11021
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8375664154216675e-06,
+      "loss": 0.9462,
+      "step": 11022
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8326102221699245e-06,
+      "loss": 1.0085,
+      "step": 11023
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.827657103880656e-06,
+      "loss": 19.0702,
+      "step": 11024
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8227070608837675e-06,
+      "loss": 0.1905,
+      "step": 11025
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.817760093508932e-06,
+      "loss": 0.959,
+      "step": 11026
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8128162020856616e-06,
+      "loss": 0.2218,
+      "step": 11027
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8078753869432494e-06,
+      "loss": 0.933,
+      "step": 11028
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.802937648410748e-06,
+      "loss": 0.9345,
+      "step": 11029
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7980029868170564e-06,
+      "loss": 19.1495,
+      "step": 11030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.793071402490844e-06,
+      "loss": 21.2777,
+      "step": 11031
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7881428957605713e-06,
+      "loss": 0.2024,
+      "step": 11032
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.783217466954503e-06,
+      "loss": 19.8024,
+      "step": 11033
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.778295116400693e-06,
+      "loss": 17.7971,
+      "step": 11034
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.773375844427002e-06,
+      "loss": 18.9566,
+      "step": 11035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.76845965136105e-06,
+      "loss": 0.9871,
+      "step": 11036
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7635465375303047e-06,
+      "loss": 19.1043,
+      "step": 11037
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.758636503262003e-06,
+      "loss": 0.19,
+      "step": 11038
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.75372954888315e-06,
+      "loss": 20.3357,
+      "step": 11039
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.748825674720602e-06,
+      "loss": 0.9939,
+      "step": 11040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7439248811009687e-06,
+      "loss": 0.9355,
+      "step": 11041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.739027168350662e-06,
+      "loss": 0.964,
+      "step": 11042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7341325367958813e-06,
+      "loss": 18.6485,
+      "step": 11043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.729240986762661e-06,
+      "loss": 17.4685,
+      "step": 11044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7243525185767737e-06,
+      "loss": 18.171,
+      "step": 11045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7194671325638198e-06,
+      "loss": 19.2981,
+      "step": 11046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7145848290492057e-06,
+      "loss": 18.8816,
+      "step": 11047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7097056083580995e-06,
+      "loss": 19.6991,
+      "step": 11048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7048294708154853e-06,
+      "loss": 0.1863,
+      "step": 11049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.69995641674612e-06,
+      "loss": 0.945,
+      "step": 11050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6950864464746105e-06,
+      "loss": 17.6252,
+      "step": 11051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.690219560325281e-06,
+      "loss": 0.9455,
+      "step": 11052
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6853557586222996e-06,
+      "loss": 0.2195,
+      "step": 11053
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6804950416896355e-06,
+      "loss": 18.4161,
+      "step": 11054
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6756374098510183e-06,
+      "loss": 18.9984,
+      "step": 11055
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.670782863429978e-06,
+      "loss": 19.7723,
+      "step": 11056
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6659314027498835e-06,
+      "loss": 0.227,
+      "step": 11057
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6610830281338327e-06,
+      "loss": 19.5142,
+      "step": 11058
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6562377399047663e-06,
+      "loss": 0.1865,
+      "step": 11059
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6513955383853936e-06,
+      "loss": 0.977,
+      "step": 11060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.64655642389824e-06,
+      "loss": 0.9197,
+      "step": 11061
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6417203967655967e-06,
+      "loss": 19.2681,
+      "step": 11062
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6368874573095846e-06,
+      "loss": 18.47,
+      "step": 11063
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6320576058520793e-06,
+      "loss": 17.4336,
+      "step": 11064
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6272308427147904e-06,
+      "loss": 18.5905,
+      "step": 11065
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6224071682191884e-06,
+      "loss": 0.9358,
+      "step": 11066
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.617586582686566e-06,
+      "loss": 19.4055,
+      "step": 11067
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6127690864379836e-06,
+      "loss": 18.7357,
+      "step": 11068
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.607954679794312e-06,
+      "loss": 0.9529,
+      "step": 11069
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.603143363076217e-06,
+      "loss": 0.2001,
+      "step": 11070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.598335136604153e-06,
+      "loss": 0.8873,
+      "step": 11071
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5935300006983697e-06,
+      "loss": 0.2293,
+      "step": 11072
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5887279556789167e-06,
+      "loss": 16.9267,
+      "step": 11073
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5839290018656213e-06,
+      "loss": 19.1381,
+      "step": 11074
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5791331395781226e-06,
+      "loss": 19.13,
+      "step": 11075
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.574340369135848e-06,
+      "loss": 0.9318,
+      "step": 11076
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.56955069085802e-06,
+      "loss": 19.6641,
+      "step": 11077
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.564764105063645e-06,
+      "loss": 18.6253,
+      "step": 11078
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5599806120715453e-06,
+      "loss": 0.9499,
+      "step": 11079
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5552002122003104e-06,
+      "loss": 18.6843,
+      "step": 11080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5504229057683414e-06,
+      "loss": 0.1956,
+      "step": 11081
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5456486930938393e-06,
+      "loss": 20.7481,
+      "step": 11082
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5408775744947664e-06,
+      "loss": 18.0537,
+      "step": 11083
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5361095502889242e-06,
+      "loss": 21.1214,
+      "step": 11084
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5313446207938804e-06,
+      "loss": 19.1002,
+      "step": 11085
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5265827863269817e-06,
+      "loss": 0.2018,
+      "step": 11086
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.521824047205413e-06,
+      "loss": 18.0535,
+      "step": 11087
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.517068403746121e-06,
+      "loss": 21.1436,
+      "step": 11088
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.512315856265852e-06,
+      "loss": 17.4471,
+      "step": 11089
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5075664050811306e-06,
+      "loss": 17.956,
+      "step": 11090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.502820050508332e-06,
+      "loss": 0.18,
+      "step": 11091
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4980767928635527e-06,
+      "loss": 17.9371,
+      "step": 11092
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.493336632462718e-06,
+      "loss": 0.919,
+      "step": 11093
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.488599569621559e-06,
+      "loss": 21.8525,
+      "step": 11094
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.483865604655584e-06,
+      "loss": 17.1524,
+      "step": 11095
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.47913473788008e-06,
+      "loss": 0.9281,
+      "step": 11096
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4744069696101664e-06,
+      "loss": 18.64,
+      "step": 11097
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4696823001607314e-06,
+      "loss": 17.9422,
+      "step": 11098
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4649607298464447e-06,
+      "loss": 20.9864,
+      "step": 11099
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.460242258981783e-06,
+      "loss": 18.936,
+      "step": 11100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.45552688788105e-06,
+      "loss": 16.8852,
+      "step": 11101
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.450814616858278e-06,
+      "loss": 0.9238,
+      "step": 11102
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.446105446227332e-06,
+      "loss": 0.9505,
+      "step": 11103
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4413993763018847e-06,
+      "loss": 18.1334,
+      "step": 11104
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4366964073953677e-06,
+      "loss": 0.9384,
+      "step": 11105
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.431996539821014e-06,
+      "loss": 0.1881,
+      "step": 11106
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.427299773891868e-06,
+      "loss": 18.9831,
+      "step": 11107
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4226061099207507e-06,
+      "loss": 19.9678,
+      "step": 11108
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.41791554822028e-06,
+      "loss": 0.8987,
+      "step": 11109
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4132280891028767e-06,
+      "loss": 0.9251,
+      "step": 11110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.408543732880742e-06,
+      "loss": 0.9364,
+      "step": 11111
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.403862479865877e-06,
+      "loss": 18.6295,
+      "step": 11112
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3991843303700756e-06,
+      "loss": 19.3046,
+      "step": 11113
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.394509284704922e-06,
+      "loss": 20.1284,
+      "step": 11114
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3898373431817955e-06,
+      "loss": 18.8722,
+      "step": 11115
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3851685061118686e-06,
+      "loss": 18.637,
+      "step": 11116
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3805027738061156e-06,
+      "loss": 20.2701,
+      "step": 11117
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3758401465752875e-06,
+      "loss": 18.3686,
+      "step": 11118
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3711806247299417e-06,
+      "loss": 18.3842,
+      "step": 11119
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3665242085804183e-06,
+      "loss": 0.9505,
+      "step": 11120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.361870898436864e-06,
+      "loss": 0.9494,
+      "step": 11121
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.357220694609203e-06,
+      "loss": 20.6339,
+      "step": 11122
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.352573597407166e-06,
+      "loss": 20.5133,
+      "step": 11123
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3479296071402654e-06,
+      "loss": 19.2894,
+      "step": 11124
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3432887241178214e-06,
+      "loss": 0.9547,
+      "step": 11125
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3386509486489304e-06,
+      "loss": 0.9687,
+      "step": 11126
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.334016281042496e-06,
+      "loss": 18.7689,
+      "step": 11127
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.329384721607204e-06,
+      "loss": 0.9428,
+      "step": 11128
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3247562706515465e-06,
+      "loss": 0.2267,
+      "step": 11129
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3201309284837724e-06,
+      "loss": 0.1725,
+      "step": 11130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.315508695411984e-06,
+      "loss": 18.1052,
+      "step": 11131
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3108895717440358e-06,
+      "loss": 0.1984,
+      "step": 11132
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3062735577875646e-06,
+      "loss": 17.6947,
+      "step": 11133
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.301660653850036e-06,
+      "loss": 0.8917,
+      "step": 11134
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2970508602386873e-06,
+      "loss": 18.6044,
+      "step": 11135
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.292444177260562e-06,
+      "loss": 17.8532,
+      "step": 11136
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2878406052224587e-06,
+      "loss": 0.9503,
+      "step": 11137
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.283240144431016e-06,
+      "loss": 0.2199,
+      "step": 11138
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2786427951926544e-06,
+      "loss": 17.7853,
+      "step": 11139
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.274048557813553e-06,
+      "loss": 20.1357,
+      "step": 11140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.269457432599732e-06,
+      "loss": 21.0096,
+      "step": 11141
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2648694198569805e-06,
+      "loss": 0.9296,
+      "step": 11142
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2602845198908603e-06,
+      "loss": 18.3437,
+      "step": 11143
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2557027330067658e-06,
+      "loss": 0.2414,
+      "step": 11144
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2511240595098747e-06,
+      "loss": 19.8616,
+      "step": 11145
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.246548499705121e-06,
+      "loss": 17.7267,
+      "step": 11146
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2419760538972666e-06,
+      "loss": 19.9354,
+      "step": 11147
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2374067223908733e-06,
+      "loss": 20.4149,
+      "step": 11148
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2328405054902644e-06,
+      "loss": 0.9446,
+      "step": 11149
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.228277403499569e-06,
+      "loss": 19.6953,
+      "step": 11150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2237174167227324e-06,
+      "loss": 0.1719,
+      "step": 11151
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2191605454634456e-06,
+      "loss": 0.9295,
+      "step": 11152
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2146067900252263e-06,
+      "loss": 18.8377,
+      "step": 11153
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2100561507113823e-06,
+      "loss": 0.896,
+      "step": 11154
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.205508627824999e-06,
+      "loss": 20.0585,
+      "step": 11155
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.200964221668967e-06,
+      "loss": 19.7989,
+      "step": 11156
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.196422932545956e-06,
+      "loss": 20.5413,
+      "step": 11157
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1918847607584567e-06,
+      "loss": 19.0047,
+      "step": 11158
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1873497066087165e-06,
+      "loss": 18.8883,
+      "step": 11159
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1828177703987883e-06,
+      "loss": 0.1567,
+      "step": 11160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.178288952430536e-06,
+      "loss": 0.1698,
+      "step": 11161
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.173763253005585e-06,
+      "loss": 19.1432,
+      "step": 11162
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1692406724253776e-06,
+      "loss": 0.9545,
+      "step": 11163
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1647212109911285e-06,
+      "loss": 20.1803,
+      "step": 11164
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1602048690038633e-06,
+      "loss": 19.01,
+      "step": 11165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.155691646764386e-06,
+      "loss": 18.3081,
+      "step": 11166
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.151181544573306e-06,
+      "loss": 0.9326,
+      "step": 11167
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.146674562731011e-06,
+      "loss": 0.1788,
+      "step": 11168
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1421707015376835e-06,
+      "loss": 20.666,
+      "step": 11169
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1376699612933104e-06,
+      "loss": 19.1524,
+      "step": 11170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.133172342297652e-06,
+      "loss": 0.9878,
+      "step": 11171
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1286778448502808e-06,
+      "loss": 18.4606,
+      "step": 11172
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.124186469250545e-06,
+      "loss": 0.9038,
+      "step": 11173
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1196982157975895e-06,
+      "loss": 0.8992,
+      "step": 11174
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1152130847903583e-06,
+      "loss": 0.9039,
+      "step": 11175
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.11073107652759e-06,
+      "loss": 19.9185,
+      "step": 11176
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1062521913077793e-06,
+      "loss": 0.9632,
+      "step": 11177
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.101776429429265e-06,
+      "loss": 0.9721,
+      "step": 11178
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0973037911901427e-06,
+      "loss": 19.0201,
+      "step": 11179
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.092834276888318e-06,
+      "loss": 0.9178,
+      "step": 11180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0883678868214806e-06,
+      "loss": 0.1937,
+      "step": 11181
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0839046212871094e-06,
+      "loss": 0.9441,
+      "step": 11182
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0794444805824776e-06,
+      "loss": 20.1595,
+      "step": 11183
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0749874650046527e-06,
+      "loss": 0.9014,
+      "step": 11184
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0705335748504928e-06,
+      "loss": 0.9208,
+      "step": 11185
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0660828104166596e-06,
+      "loss": 18.4791,
+      "step": 11186
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.061635171999566e-06,
+      "loss": 19.9613,
+      "step": 11187
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0571906598954703e-06,
+      "loss": 0.1987,
+      "step": 11188
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.052749274400396e-06,
+      "loss": 18.0403,
+      "step": 11189
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.048311015810135e-06,
+      "loss": 19.7584,
+      "step": 11190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0438758844203274e-06,
+      "loss": 0.9727,
+      "step": 11191
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0394438805263658e-06,
+      "loss": 21.3874,
+      "step": 11192
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0350150044234305e-06,
+      "loss": 0.1722,
+      "step": 11193
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0305892564065075e-06,
+      "loss": 20.3861,
+      "step": 11194
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0261666367703888e-06,
+      "loss": 0.1745,
+      "step": 11195
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0217471458096225e-06,
+      "loss": 18.861,
+      "step": 11196
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.017330783818567e-06,
+      "loss": 19.2499,
+      "step": 11197
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.012917551091393e-06,
+      "loss": 17.2774,
+      "step": 11198
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.008507447922021e-06,
+      "loss": 16.1712,
+      "step": 11199
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0041004746041822e-06,
+      "loss": 18.6046,
+      "step": 11200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.99969663143142e-06,
+      "loss": 17.9468,
+      "step": 11201
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9952959186970553e-06,
+      "loss": 20.6983,
+      "step": 11202
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.99089833669417e-06,
+      "loss": 0.9067,
+      "step": 11203
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9865038857156693e-06,
+      "loss": 0.9299,
+      "step": 11204
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.982112566054268e-06,
+      "loss": 19.9091,
+      "step": 11205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.977724378002422e-06,
+      "loss": 0.9666,
+      "step": 11206
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9733393218524084e-06,
+      "loss": 20.7584,
+      "step": 11207
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.96895739789631e-06,
+      "loss": 0.1828,
+      "step": 11208
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.964578606425966e-06,
+      "loss": 19.2469,
+      "step": 11209
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.960202947733032e-06,
+      "loss": 0.9408,
+      "step": 11210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9558304221089405e-06,
+      "loss": 18.0849,
+      "step": 11211
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.951461029844932e-06,
+      "loss": 17.202,
+      "step": 11212
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9470947712320175e-06,
+      "loss": 19.235,
+      "step": 11213
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.94273164656102e-06,
+      "loss": 16.6014,
+      "step": 11214
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.938371656122535e-06,
+      "loss": 19.0407,
+      "step": 11215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9340148002069633e-06,
+      "loss": 19.3544,
+      "step": 11216
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9296610791044955e-06,
+      "loss": 0.9588,
+      "step": 11217
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.925310493105099e-06,
+      "loss": 0.989,
+      "step": 11218
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9209630424985535e-06,
+      "loss": 0.91,
+      "step": 11219
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9166187275744107e-06,
+      "loss": 19.2969,
+      "step": 11220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.912277548622028e-06,
+      "loss": 0.1929,
+      "step": 11221
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.907939505930546e-06,
+      "loss": 18.5257,
+      "step": 11222
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9036045997889005e-06,
+      "loss": 0.214,
+      "step": 11223
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8992728304858164e-06,
+      "loss": 0.2148,
+      "step": 11224
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8949441983098015e-06,
+      "loss": 17.5431,
+      "step": 11225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.890618703549175e-06,
+      "loss": 19.0518,
+      "step": 11226
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8862963464920235e-06,
+      "loss": 17.3448,
+      "step": 11227
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8819771274262497e-06,
+      "loss": 19.2053,
+      "step": 11228
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8776610466395183e-06,
+      "loss": 18.5404,
+      "step": 11229
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.87334810441931e-06,
+      "loss": 18.737,
+      "step": 11230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8690383010528844e-06,
+      "loss": 17.8349,
+      "step": 11231
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.864731636827295e-06,
+      "loss": 18.7255,
+      "step": 11232
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.860428112029395e-06,
+      "loss": 0.9527,
+      "step": 11233
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.85612772694579e-06,
+      "loss": 0.9656,
+      "step": 11234
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8518304818629384e-06,
+      "loss": 0.9891,
+      "step": 11235
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8475363770670448e-06,
+      "loss": 19.8395,
+      "step": 11236
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8432454128440977e-06,
+      "loss": 19.2897,
+      "step": 11237
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.838957589479924e-06,
+      "loss": 0.1847,
+      "step": 11238
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.834672907260111e-06,
+      "loss": 0.9104,
+      "step": 11239
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.830391366470014e-06,
+      "loss": 19.2939,
+      "step": 11240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8261129673948174e-06,
+      "loss": 18.5684,
+      "step": 11241
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.821837710319486e-06,
+      "loss": 0.9316,
+      "step": 11242
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.817565595528776e-06,
+      "loss": 18.9693,
+      "step": 11243
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.81329662330721e-06,
+      "loss": 18.8295,
+      "step": 11244
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8090307939391435e-06,
+      "loss": 0.2211,
+      "step": 11245
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.804768107708694e-06,
+      "loss": 0.1888,
+      "step": 11246
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.800508564899762e-06,
+      "loss": 18.0668,
+      "step": 11247
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7962521657960706e-06,
+      "loss": 0.9005,
+      "step": 11248
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.791998910681115e-06,
+      "loss": 0.9634,
+      "step": 11249
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.787748799838169e-06,
+      "loss": 18.4181,
+      "step": 11250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7835018335503117e-06,
+      "loss": 0.9001,
+      "step": 11251
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7792580121004276e-06,
+      "loss": 20.0098,
+      "step": 11252
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7750173357711574e-06,
+      "loss": 0.9518,
+      "step": 11253
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.770779804844953e-06,
+      "loss": 21.3009,
+      "step": 11254
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7665454196040664e-06,
+      "loss": 19.0465,
+      "step": 11255
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7623141803305055e-06,
+      "loss": 0.9073,
+      "step": 11256
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.758086087306111e-06,
+      "loss": 17.1441,
+      "step": 11257
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.753861140812475e-06,
+      "loss": 18.5531,
+      "step": 11258
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7496393411310163e-06,
+      "loss": 0.1853,
+      "step": 11259
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7454206885429156e-06,
+      "loss": 0.9708,
+      "step": 11260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.741205183329154e-06,
+      "loss": 18.1387,
+      "step": 11261
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7369928257705123e-06,
+      "loss": 19.1876,
+      "step": 11262
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.732783616147544e-06,
+      "loss": 0.9429,
+      "step": 11263
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.728577554740602e-06,
+      "loss": 0.8911,
+      "step": 11264
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7243746418298466e-06,
+      "loss": 0.1871,
+      "step": 11265
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7201748776951862e-06,
+      "loss": 19.5663,
+      "step": 11266
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7159782626163643e-06,
+      "loss": 0.9566,
+      "step": 11267
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7117847968728847e-06,
+      "loss": 0.9639,
+      "step": 11268
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7075944807440524e-06,
+      "loss": 20.127,
+      "step": 11269
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.70340731450896e-06,
+      "loss": 20.4859,
+      "step": 11270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6992232984465015e-06,
+      "loss": 0.9536,
+      "step": 11271
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.695042432835343e-06,
+      "loss": 0.2052,
+      "step": 11272
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6908647179539558e-06,
+      "loss": 0.8975,
+      "step": 11273
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.68669015408059e-06,
+      "loss": 0.9376,
+      "step": 11274
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.682518741493295e-06,
+      "loss": 18.6802,
+      "step": 11275
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6783504804699046e-06,
+      "loss": 19.9936,
+      "step": 11276
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6741853712880405e-06,
+      "loss": 19.0074,
+      "step": 11277
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.67002341422512e-06,
+      "loss": 20.2476,
+      "step": 11278
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6658646095583552e-06,
+      "loss": 18.638,
+      "step": 11279
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6617089575647402e-06,
+      "loss": 0.9122,
+      "step": 11280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.657556458521049e-06,
+      "loss": 0.921,
+      "step": 11281
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6534071127038707e-06,
+      "loss": 0.9554,
+      "step": 11282
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.649260920389568e-06,
+      "loss": 0.9248,
+      "step": 11283
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6451178818542864e-06,
+      "loss": 0.2115,
+      "step": 11284
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6409779973739836e-06,
+      "loss": 18.333,
+      "step": 11285
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6368412672243936e-06,
+      "loss": 0.1953,
+      "step": 11286
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.632707691681041e-06,
+      "loss": 19.7229,
+      "step": 11287
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.628577271019239e-06,
+      "loss": 20.4669,
+      "step": 11288
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6244500055140895e-06,
+      "loss": 0.943,
+      "step": 11289
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.620325895440501e-06,
+      "loss": 18.6971,
+      "step": 11290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.616204941073136e-06,
+      "loss": 20.0484,
+      "step": 11291
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.612087142686487e-06,
+      "loss": 18.3172,
+      "step": 11292
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6079725005548227e-06,
+      "loss": 17.1435,
+      "step": 11293
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6038610149521747e-06,
+      "loss": 18.351,
+      "step": 11294
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.599752686152407e-06,
+      "loss": 0.9509,
+      "step": 11295
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5956475144291505e-06,
+      "loss": 18.0137,
+      "step": 11296
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.59154550005582e-06,
+      "loss": 0.9359,
+      "step": 11297
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5874466433056297e-06,
+      "loss": 18.305,
+      "step": 11298
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5833509444515957e-06,
+      "loss": 0.9128,
+      "step": 11299
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.579258403766499e-06,
+      "loss": 19.4666,
+      "step": 11300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5751690215229163e-06,
+      "loss": 18.856,
+      "step": 11301
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5710827979932406e-06,
+      "loss": 17.0212,
+      "step": 11302
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5669997334496156e-06,
+      "loss": 17.3004,
+      "step": 11303
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5629198281639843e-06,
+      "loss": 18.6293,
+      "step": 11304
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.558843082408119e-06,
+      "loss": 19.0519,
+      "step": 11305
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.554769496453524e-06,
+      "loss": 0.2222,
+      "step": 11306
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5506990705715217e-06,
+      "loss": 0.1973,
+      "step": 11307
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5466318050332227e-06,
+      "loss": 19.3494,
+      "step": 11308
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5425677001095385e-06,
+      "loss": 0.9181,
+      "step": 11309
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5385067560711464e-06,
+      "loss": 18.6137,
+      "step": 11310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5344489731885134e-06,
+      "loss": 16.2828,
+      "step": 11311
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.530394351731935e-06,
+      "loss": 0.973,
+      "step": 11312
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.526342891971445e-06,
+      "loss": 19.5234,
+      "step": 11313
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.522294594176894e-06,
+      "loss": 19.1841,
+      "step": 11314
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.518249458617916e-06,
+      "loss": 0.1931,
+      "step": 11315
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.514207485563941e-06,
+      "loss": 0.9393,
+      "step": 11316
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5101686752841856e-06,
+      "loss": 0.2208,
+      "step": 11317
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5061330280476413e-06,
+      "loss": 18.375,
+      "step": 11318
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.502100544123115e-06,
+      "loss": 0.9123,
+      "step": 11319
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4980712237791803e-06,
+      "loss": 20.0901,
+      "step": 11320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.494045067284212e-06,
+      "loss": 20.1872,
+      "step": 11321
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4900220749063673e-06,
+      "loss": 19.3022,
+      "step": 11322
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.486002246913599e-06,
+      "loss": 0.1886,
+      "step": 11323
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.481985583573643e-06,
+      "loss": 0.9761,
+      "step": 11324
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4779720851540343e-06,
+      "loss": 0.2001,
+      "step": 11325
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4739617519220824e-06,
+      "loss": 0.1624,
+      "step": 11326
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.469954584144901e-06,
+      "loss": 19.422,
+      "step": 11327
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4659505820893826e-06,
+      "loss": 17.855,
+      "step": 11328
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4619497460222184e-06,
+      "loss": 0.9514,
+      "step": 11329
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4579520762098683e-06,
+      "loss": 20.0655,
+      "step": 11330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4539575729186135e-06,
+      "loss": 0.9106,
+      "step": 11331
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4499662364144915e-06,
+      "loss": 17.8363,
+      "step": 11332
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4459780669633503e-06,
+      "loss": 0.1981,
+      "step": 11333
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.441993064830822e-06,
+      "loss": 19.2716,
+      "step": 11334
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4380112302823276e-06,
+      "loss": 19.2879,
+      "step": 11335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.434032563583066e-06,
+      "loss": 0.189,
+      "step": 11336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4300570649980538e-06,
+      "loss": 18.3835,
+      "step": 11337
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.42608473479205e-06,
+      "loss": 0.9953,
+      "step": 11338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.42211557322965e-06,
+      "loss": 18.416,
+      "step": 11339
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.418149580575224e-06,
+      "loss": 20.3224,
+      "step": 11340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.414186757092901e-06,
+      "loss": 0.1711,
+      "step": 11341
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4102271030466474e-06,
+      "loss": 19.0264,
+      "step": 11342
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.406270618700185e-06,
+      "loss": 0.9213,
+      "step": 11343
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4023173043170312e-06,
+      "loss": 19.8183,
+      "step": 11344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3983671601604916e-06,
+      "loss": 0.9658,
+      "step": 11345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3944201864936843e-06,
+      "loss": 17.7288,
+      "step": 11346
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3904763835794762e-06,
+      "loss": 18.9483,
+      "step": 11347
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3865357516805407e-06,
+      "loss": 18.7853,
+      "step": 11348
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.382598291059357e-06,
+      "loss": 18.5485,
+      "step": 11349
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3786640019781815e-06,
+      "loss": 19.2842,
+      "step": 11350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.374732884699027e-06,
+      "loss": 19.9742,
+      "step": 11351
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3708049394837562e-06,
+      "loss": 0.2361,
+      "step": 11352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3668801665939823e-06,
+      "loss": 0.9871,
+      "step": 11353
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.362958566291096e-06,
+      "loss": 19.1356,
+      "step": 11354
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3590401388362992e-06,
+      "loss": 21.1206,
+      "step": 11355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3551248844906004e-06,
+      "loss": 19.3101,
+      "step": 11356
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.351212803514746e-06,
+      "loss": 0.9021,
+      "step": 11357
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3473038961693004e-06,
+      "loss": 0.1971,
+      "step": 11358
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.343398162714644e-06,
+      "loss": 0.93,
+      "step": 11359
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.339495603410885e-06,
+      "loss": 18.6109,
+      "step": 11360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3355962185179604e-06,
+      "loss": 0.9334,
+      "step": 11361
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3317000082955955e-06,
+      "loss": 0.2161,
+      "step": 11362
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.327806973003288e-06,
+      "loss": 17.9861,
+      "step": 11363
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3239171129003424e-06,
+      "loss": 18.593,
+      "step": 11364
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3200304282458285e-06,
+      "loss": 0.188,
+      "step": 11365
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.316146919298623e-06,
+      "loss": 0.1946,
+      "step": 11366
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3122665863173854e-06,
+      "loss": 19.5456,
+      "step": 11367
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3083894295605646e-06,
+      "loss": 0.9199,
+      "step": 11368
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.304515449286393e-06,
+      "loss": 20.0724,
+      "step": 11369
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3006446457529086e-06,
+      "loss": 0.1636,
+      "step": 11370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2967770192179106e-06,
+      "loss": 18.1858,
+      "step": 11371
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.292912569939004e-06,
+      "loss": 0.952,
+      "step": 11372
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2890512981735833e-06,
+      "loss": 19.7317,
+      "step": 11373
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.285193204178826e-06,
+      "loss": 0.9204,
+      "step": 11374
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2813382882116986e-06,
+      "loss": 0.9779,
+      "step": 11375
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2774865505289512e-06,
+      "loss": 19.5972,
+      "step": 11376
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2736379913871398e-06,
+      "loss": 18.4187,
+      "step": 11377
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2697926110425815e-06,
+      "loss": 0.9679,
+      "step": 11378
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.265950409751405e-06,
+      "loss": 20.7977,
+      "step": 11379
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.262111387769522e-06,
+      "loss": 17.7807,
+      "step": 11380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.258275545352617e-06,
+      "loss": 19.6467,
+      "step": 11381
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.254442882756186e-06,
+      "loss": 0.9257,
+      "step": 11382
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.250613400235502e-06,
+      "loss": 0.925,
+      "step": 11383
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2467870980456274e-06,
+      "loss": 20.8101,
+      "step": 11384
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2429639764413924e-06,
+      "loss": 0.2236,
+      "step": 11385
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.239144035677454e-06,
+      "loss": 20.2847,
+      "step": 11386
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2353272760082423e-06,
+      "loss": 0.2071,
+      "step": 11387
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2315136976879536e-06,
+      "loss": 0.2277,
+      "step": 11388
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.227703300970596e-06,
+      "loss": 18.4528,
+      "step": 11389
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.223896086109978e-06,
+      "loss": 18.4408,
+      "step": 11390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2200920533596405e-06,
+      "loss": 19.6412,
+      "step": 11391
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2162912029729867e-06,
+      "loss": 0.8971,
+      "step": 11392
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2124935352031472e-06,
+      "loss": 17.7702,
+      "step": 11393
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2086990503030814e-06,
+      "loss": 0.945,
+      "step": 11394
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2049077485254975e-06,
+      "loss": 0.1671,
+      "step": 11395
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2011196301229386e-06,
+      "loss": 17.7184,
+      "step": 11396
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1973346953477027e-06,
+      "loss": 0.1836,
+      "step": 11397
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1935529444518654e-06,
+      "loss": 0.8977,
+      "step": 11398
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1897743776873316e-06,
+      "loss": 0.2321,
+      "step": 11399
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1859989953057714e-06,
+      "loss": 18.7249,
+      "step": 11400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1822267975586286e-06,
+      "loss": 19.3577,
+      "step": 11401
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1784577846971464e-06,
+      "loss": 20.1098,
+      "step": 11402
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1746919569723855e-06,
+      "loss": 19.9855,
+      "step": 11403
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1709293146351397e-06,
+      "loss": 0.2144,
+      "step": 11404
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1671698579360244e-06,
+      "loss": 19.9561,
+      "step": 11405
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1634135871254568e-06,
+      "loss": 21.0423,
+      "step": 11406
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1596605024535975e-06,
+      "loss": 0.1812,
+      "step": 11407
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.155910604170419e-06,
+      "loss": 19.5051,
+      "step": 11408
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1521638925257104e-06,
+      "loss": 19.3834,
+      "step": 11409
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1484203677689885e-06,
+      "loss": 19.4872,
+      "step": 11410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.144680030149604e-06,
+      "loss": 0.9426,
+      "step": 11411
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1409428799166742e-06,
+      "loss": 19.123,
+      "step": 11412
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1372089173191223e-06,
+      "loss": 21.8837,
+      "step": 11413
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1334781426056384e-06,
+      "loss": 19.0201,
+      "step": 11414
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.129750556024701e-06,
+      "loss": 19.803,
+      "step": 11415
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.126026157824612e-06,
+      "loss": 20.8936,
+      "step": 11416
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.122304948253406e-06,
+      "loss": 0.1703,
+      "step": 11417
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1185869275589465e-06,
+      "loss": 0.8873,
+      "step": 11418
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1148720959888677e-06,
+      "loss": 19.0072,
+      "step": 11419
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.111160453790595e-06,
+      "loss": 0.9387,
+      "step": 11420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1074520012113406e-06,
+      "loss": 0.9223,
+      "step": 11421
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1037467384981026e-06,
+      "loss": 20.6877,
+      "step": 11422
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1000446658976715e-06,
+      "loss": 20.5998,
+      "step": 11423
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0963457836566224e-06,
+      "loss": 18.5844,
+      "step": 11424
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.09265009202132e-06,
+      "loss": 0.2278,
+      "step": 11425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.088957591237911e-06,
+      "loss": 20.9035,
+      "step": 11426
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0852682815523326e-06,
+      "loss": 20.2165,
+      "step": 11427
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0815821632103105e-06,
+      "loss": 0.1907,
+      "step": 11428
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.077899236457359e-06,
+      "loss": 0.9467,
+      "step": 11429
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.074219501538777e-06,
+      "loss": 17.422,
+      "step": 11430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.070542958699656e-06,
+      "loss": 0.1859,
+      "step": 11431
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0668696081848628e-06,
+      "loss": 18.0937,
+      "step": 11432
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0631994502390673e-06,
+      "loss": 1.0097,
+      "step": 11433
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0595324851067134e-06,
+      "loss": 0.1821,
+      "step": 11434
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0558687130320386e-06,
+      "loss": 18.1,
+      "step": 11435
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.052208134259076e-06,
+      "loss": 19.5161,
+      "step": 11436
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0485507490316245e-06,
+      "loss": 0.9186,
+      "step": 11437
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.04489655759329e-06,
+      "loss": 0.8798,
+      "step": 11438
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.041245560187455e-06,
+      "loss": 19.5705,
+      "step": 11439
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0375977570572967e-06,
+      "loss": 0.1758,
+      "step": 11440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0339531484457776e-06,
+      "loss": 0.2471,
+      "step": 11441
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.03031173459563e-06,
+      "loss": 19.8943,
+      "step": 11442
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.026673515749411e-06,
+      "loss": 0.9349,
+      "step": 11443
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0230384921494317e-06,
+      "loss": 21.7428,
+      "step": 11444
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0194066640377986e-06,
+      "loss": 0.1963,
+      "step": 11445
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0157780316564123e-06,
+      "loss": 18.0181,
+      "step": 11446
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.012152595246963e-06,
+      "loss": 0.196,
+      "step": 11447
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.008530355050914e-06,
+      "loss": 0.1769,
+      "step": 11448
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0049113113095153e-06,
+      "loss": 19.2612,
+      "step": 11449
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0012954642638303e-06,
+      "loss": 0.9712,
+      "step": 11450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.997682814154683e-06,
+      "loss": 0.9061,
+      "step": 11451
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9940733612226815e-06,
+      "loss": 18.9747,
+      "step": 11452
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.990467105708249e-06,
+      "loss": 0.9623,
+      "step": 11453
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.986864047851572e-06,
+      "loss": 19.5143,
+      "step": 11454
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.983264187892625e-06,
+      "loss": 0.1999,
+      "step": 11455
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9796675260711885e-06,
+      "loss": 0.1858,
+      "step": 11456
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9760740626268147e-06,
+      "loss": 0.9773,
+      "step": 11457
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9724837977988352e-06,
+      "loss": 18.2863,
+      "step": 11458
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.968896731826375e-06,
+      "loss": 18.353,
+      "step": 11459
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.965312864948371e-06,
+      "loss": 18.2078,
+      "step": 11460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.961732197403504e-06,
+      "loss": 19.1692,
+      "step": 11461
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9581547294302616e-06,
+      "loss": 0.1906,
+      "step": 11462
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.954580461266947e-06,
+      "loss": 0.9621,
+      "step": 11463
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9510093931515983e-06,
+      "loss": 19.7972,
+      "step": 11464
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.947441525322069e-06,
+      "loss": 18.9102,
+      "step": 11465
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.943876858016003e-06,
+      "loss": 18.2101,
+      "step": 11466
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9403153914708206e-06,
+      "loss": 18.0731,
+      "step": 11467
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.936757125923727e-06,
+      "loss": 0.194,
+      "step": 11468
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.933202061611722e-06,
+      "loss": 17.9958,
+      "step": 11469
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.929650198771593e-06,
+      "loss": 0.9533,
+      "step": 11470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.926101537639907e-06,
+      "loss": 19.7469,
+      "step": 11471
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.922556078453025e-06,
+      "loss": 0.9843,
+      "step": 11472
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9190138214470854e-06,
+      "loss": 0.9376,
+      "step": 11473
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9154747668580218e-06,
+      "loss": 18.1887,
+      "step": 11474
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9119389149215517e-06,
+      "loss": 18.4986,
+      "step": 11475
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9084062658731804e-06,
+      "loss": 0.9569,
+      "step": 11476
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9048768199481982e-06,
+      "loss": 0.953,
+      "step": 11477
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9013505773816776e-06,
+      "loss": 18.8947,
+      "step": 11478
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.897827538408492e-06,
+      "loss": 0.1978,
+      "step": 11479
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8943077032632872e-06,
+      "loss": 19.197,
+      "step": 11480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8907910721804978e-06,
+      "loss": 0.9216,
+      "step": 11481
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.887277645394353e-06,
+      "loss": 19.2469,
+      "step": 11482
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8837674231388602e-06,
+      "loss": 0.2235,
+      "step": 11483
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8802604056478157e-06,
+      "loss": 0.9538,
+      "step": 11484
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8767565931548103e-06,
+      "loss": 19.729,
+      "step": 11485
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.873255985893202e-06,
+      "loss": 18.9372,
+      "step": 11486
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8697585840961596e-06,
+      "loss": 18.6325,
+      "step": 11487
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.866264387996619e-06,
+      "loss": 0.1808,
+      "step": 11488
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8627733978273055e-06,
+      "loss": 17.8691,
+      "step": 11489
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8592856138207492e-06,
+      "loss": 19.2007,
+      "step": 11490
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.855801036209248e-06,
+      "loss": 18.3285,
+      "step": 11491
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.852319665224872e-06,
+      "loss": 18.4944,
+      "step": 11492
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8488415010995241e-06,
+      "loss": 0.969,
+      "step": 11493
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8453665440648583e-06,
+      "loss": 0.9353,
+      "step": 11494
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8418947943523056e-06,
+      "loss": 0.1883,
+      "step": 11495
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.838426252193126e-06,
+      "loss": 16.0308,
+      "step": 11496
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8349609178183236e-06,
+      "loss": 19.8585,
+      "step": 11497
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8314987914587189e-06,
+      "loss": 0.9458,
+      "step": 11498
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8280398733448835e-06,
+      "loss": 18.7334,
+      "step": 11499
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8245841637072214e-06,
+      "loss": 19.7923,
+      "step": 11500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8211316627758878e-06,
+      "loss": 19.6575,
+      "step": 11501
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8176823707808322e-06,
+      "loss": 20.9425,
+      "step": 11502
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8142362879517982e-06,
+      "loss": 0.9369,
+      "step": 11503
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8107934145183193e-06,
+      "loss": 19.8645,
+      "step": 11504
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.807353750709695e-06,
+      "loss": 18.3652,
+      "step": 11505
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8039172967550143e-06,
+      "loss": 17.0712,
+      "step": 11506
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8004840528831834e-06,
+      "loss": 0.9757,
+      "step": 11507
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7970540193228636e-06,
+      "loss": 0.9296,
+      "step": 11508
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7936271963024941e-06,
+      "loss": 17.6504,
+      "step": 11509
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7902035840503484e-06,
+      "loss": 18.5847,
+      "step": 11510
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7867831827944326e-06,
+      "loss": 0.1609,
+      "step": 11511
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7833659927625645e-06,
+      "loss": 0.9309,
+      "step": 11512
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.779952014182351e-06,
+      "loss": 19.3427,
+      "step": 11513
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7765412472811771e-06,
+      "loss": 0.2096,
+      "step": 11514
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7731336922862108e-06,
+      "loss": 20.4433,
+      "step": 11515
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.769729349424415e-06,
+      "loss": 18.0568,
+      "step": 11516
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7663282189225305e-06,
+      "loss": 19.9263,
+      "step": 11517
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.762930301007093e-06,
+      "loss": 18.7006,
+      "step": 11518
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7595355959044157e-06,
+      "loss": 0.1906,
+      "step": 11519
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.756144103840618e-06,
+      "loss": 19.9121,
+      "step": 11520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7527558250415687e-06,
+      "loss": 0.1862,
+      "step": 11521
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7493707597329489e-06,
+      "loss": 20.1044,
+      "step": 11522
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.745988908140217e-06,
+      "loss": 0.2246,
+      "step": 11523
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.742610270488626e-06,
+      "loss": 20.0837,
+      "step": 11524
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7392348470032127e-06,
+      "loss": 0.9474,
+      "step": 11525
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7358626379087807e-06,
+      "loss": 18.5734,
+      "step": 11526
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.73249364342995e-06,
+      "loss": 18.1519,
+      "step": 11527
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7291278637911024e-06,
+      "loss": 18.166,
+      "step": 11528
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7257652992164142e-06,
+      "loss": 21.2438,
+      "step": 11529
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7224059499298561e-06,
+      "loss": 0.9768,
+      "step": 11530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7190498161551716e-06,
+      "loss": 19.1496,
+      "step": 11531
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7156968981158871e-06,
+      "loss": 18.88,
+      "step": 11532
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7123471960353355e-06,
+      "loss": 0.8882,
+      "step": 11533
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7090007101366157e-06,
+      "loss": 0.2249,
+      "step": 11534
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7056574406426217e-06,
+      "loss": 18.103,
+      "step": 11535
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7023173877760256e-06,
+      "loss": 19.072,
+      "step": 11536
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.698980551759294e-06,
+      "loss": 0.2177,
+      "step": 11537
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6956469328146817e-06,
+      "loss": 19.1429,
+      "step": 11538
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.692316531164212e-06,
+      "loss": 20.2443,
+      "step": 11539
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.688989347029707e-06,
+      "loss": 18.8291,
+      "step": 11540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6856653806327838e-06,
+      "loss": 19.3189,
+      "step": 11541
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.682344632194821e-06,
+      "loss": 0.916,
+      "step": 11542
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.679027101937003e-06,
+      "loss": 18.5316,
+      "step": 11543
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.675712790080286e-06,
+      "loss": 19.9213,
+      "step": 11544
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6724016968454326e-06,
+      "loss": 17.995,
+      "step": 11545
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6690938224529552e-06,
+      "loss": 19.1676,
+      "step": 11546
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.665789167123194e-06,
+      "loss": 18.8234,
+      "step": 11547
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.662487731076251e-06,
+      "loss": 0.9514,
+      "step": 11548
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6591895145320002e-06,
+      "loss": 0.9628,
+      "step": 11549
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6558945177101382e-06,
+      "loss": 0.9505,
+      "step": 11550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6526027408301226e-06,
+      "loss": 0.173,
+      "step": 11551
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6493141841111947e-06,
+      "loss": 20.6529,
+      "step": 11552
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6460288477723851e-06,
+      "loss": 18.854,
+      "step": 11553
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.642746732032524e-06,
+      "loss": 0.9162,
+      "step": 11554
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.639467837110209e-06,
+      "loss": 0.169,
+      "step": 11555
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6361921632238265e-06,
+      "loss": 20.9659,
+      "step": 11556
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6329197105915627e-06,
+      "loss": 0.1843,
+      "step": 11557
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.629650479431366e-06,
+      "loss": 18.9471,
+      "step": 11558
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6263844699609842e-06,
+      "loss": 0.9667,
+      "step": 11559
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.62312168239796e-06,
+      "loss": 0.9344,
+      "step": 11560
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6198621169595974e-06,
+      "loss": 19.1009,
+      "step": 11561
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6166057738630004e-06,
+      "loss": 17.9016,
+      "step": 11562
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6133526533250565e-06,
+      "loss": 0.172,
+      "step": 11563
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6101027555624538e-06,
+      "loss": 0.2046,
+      "step": 11564
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.60685608079163e-06,
+      "loss": 19.2713,
+      "step": 11565
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6036126292288289e-06,
+      "loss": 0.9556,
+      "step": 11566
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6003724010901056e-06,
+      "loss": 0.9736,
+      "step": 11567
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5971353965912428e-06,
+      "loss": 18.7461,
+      "step": 11568
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5939016159478571e-06,
+      "loss": 0.9127,
+      "step": 11569
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5906710593753315e-06,
+      "loss": 0.9195,
+      "step": 11570
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5874437270888332e-06,
+      "loss": 0.1713,
+      "step": 11571
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5842196193033176e-06,
+      "loss": 19.3497,
+      "step": 11572
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5809987362335243e-06,
+      "loss": 0.9079,
+      "step": 11573
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5777810780939873e-06,
+      "loss": 17.7888,
+      "step": 11574
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5745666450990016e-06,
+      "loss": 17.5287,
+      "step": 11575
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5713554374626794e-06,
+      "loss": 17.501,
+      "step": 11576
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.568147455398894e-06,
+      "loss": 0.95,
+      "step": 11577
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5649426991213134e-06,
+      "loss": 17.4612,
+      "step": 11578
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.561741168843389e-06,
+      "loss": 0.1854,
+      "step": 11579
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5585428647783617e-06,
+      "loss": 20.5264,
+      "step": 11580
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5553477871392442e-06,
+      "loss": 19.866,
+      "step": 11581
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5521559361388494e-06,
+      "loss": 0.9294,
+      "step": 11582
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5489673119897741e-06,
+      "loss": 0.9517,
+      "step": 11583
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5457819149043817e-06,
+      "loss": 17.5424,
+      "step": 11584
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5425997450948525e-06,
+      "loss": 0.1892,
+      "step": 11585
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5394208027731171e-06,
+      "loss": 19.725,
+      "step": 11586
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5362450881509172e-06,
+      "loss": 18.5233,
+      "step": 11587
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5330726014397668e-06,
+      "loss": 0.1577,
+      "step": 11588
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5299033428509746e-06,
+      "loss": 20.8375,
+      "step": 11589
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5267373125956164e-06,
+      "loss": 0.9163,
+      "step": 11590
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5235745108845734e-06,
+      "loss": 17.3979,
+      "step": 11591
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5204149379285048e-06,
+      "loss": 20.4965,
+      "step": 11592
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5172585939378369e-06,
+      "loss": 0.8845,
+      "step": 11593
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5141054791228127e-06,
+      "loss": 20.0925,
+      "step": 11594
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5109555936934472e-06,
+      "loss": 18.6987,
+      "step": 11595
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5078089378595228e-06,
+      "loss": 16.6238,
+      "step": 11596
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5046655118306275e-06,
+      "loss": 0.2389,
+      "step": 11597
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.501525315816138e-06,
+      "loss": 18.3214,
+      "step": 11598
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4983883500251872e-06,
+      "loss": 0.1875,
+      "step": 11599
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4952546146667246e-06,
+      "loss": 18.476,
+      "step": 11600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.492124109949472e-06,
+      "loss": 20.1287,
+      "step": 11601
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4889968360819295e-06,
+      "loss": 0.2205,
+      "step": 11602
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.485872793272386e-06,
+      "loss": 0.1876,
+      "step": 11603
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4827519817289248e-06,
+      "loss": 17.1499,
+      "step": 11604
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4796344016594132e-06,
+      "loss": 19.6282,
+      "step": 11605
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4765200532714685e-06,
+      "loss": 0.9776,
+      "step": 11606
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4734089367725522e-06,
+      "loss": 18.5756,
+      "step": 11607
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4703010523698658e-06,
+      "loss": 19.3603,
+      "step": 11608
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4671964002704042e-06,
+      "loss": 0.2238,
+      "step": 11609
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4640949806809523e-06,
+      "loss": 19.4326,
+      "step": 11610
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4609967938080948e-06,
+      "loss": 0.9572,
+      "step": 11611
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4579018398581668e-06,
+      "loss": 19.4079,
+      "step": 11612
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4548101190373086e-06,
+      "loss": 20.8263,
+      "step": 11613
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.451721631551456e-06,
+      "loss": 18.6909,
+      "step": 11614
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.448636377606305e-06,
+      "loss": 0.1939,
+      "step": 11615
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4455543574073526e-06,
+      "loss": 0.9237,
+      "step": 11616
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4424755711598736e-06,
+      "loss": 18.6212,
+      "step": 11617
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4394000190689316e-06,
+      "loss": 19.2243,
+      "step": 11618
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4363277013393683e-06,
+      "loss": 0.9014,
+      "step": 11619
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.43325861817582e-06,
+      "loss": 0.956,
+      "step": 11620
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.430192769782701e-06,
+      "loss": 16.292,
+      "step": 11621
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4271301563642036e-06,
+      "loss": 19.5424,
+      "step": 11622
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.42407077812432e-06,
+      "loss": 20.6771,
+      "step": 11623
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4210146352668207e-06,
+      "loss": 0.9718,
+      "step": 11624
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.417961727995254e-06,
+      "loss": 0.2114,
+      "step": 11625
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.414912056512957e-06,
+      "loss": 0.2121,
+      "step": 11626
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4118656210230508e-06,
+      "loss": 0.9454,
+      "step": 11627
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4088224217284452e-06,
+      "loss": 1.0015,
+      "step": 11628
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4057824588318391e-06,
+      "loss": 20.6279,
+      "step": 11629
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4027457325356929e-06,
+      "loss": 18.2081,
+      "step": 11630
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3997122430422781e-06,
+      "loss": 0.2069,
+      "step": 11631
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3966819905536332e-06,
+      "loss": 0.1796,
+      "step": 11632
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3936549752715856e-06,
+      "loss": 18.5314,
+      "step": 11633
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.390631197397757e-06,
+      "loss": 0.1785,
+      "step": 11634
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3876106571335367e-06,
+      "loss": 20.9111,
+      "step": 11635
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3845933546801138e-06,
+      "loss": 17.4498,
+      "step": 11636
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3815792902384494e-06,
+      "loss": 20.8459,
+      "step": 11637
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3785684640092945e-06,
+      "loss": 20.0526,
+      "step": 11638
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3755608761931938e-06,
+      "loss": 0.9965,
+      "step": 11639
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3725565269904429e-06,
+      "loss": 17.6897,
+      "step": 11640
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3695554166011704e-06,
+      "loss": 20.0167,
+      "step": 11641
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3665575452252555e-06,
+      "loss": 0.1903,
+      "step": 11642
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3635629130623607e-06,
+      "loss": 19.2679,
+      "step": 11643
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3605715203119597e-06,
+      "loss": 17.0174,
+      "step": 11644
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3575833671732873e-06,
+      "loss": 19.0864,
+      "step": 11645
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3545984538453515e-06,
+      "loss": 0.1889,
+      "step": 11646
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.351616780526982e-06,
+      "loss": 0.2105,
+      "step": 11647
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3486383474167696e-06,
+      "loss": 21.2436,
+      "step": 11648
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3456631547130894e-06,
+      "loss": 0.1891,
+      "step": 11649
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.342691202614088e-06,
+      "loss": 0.9719,
+      "step": 11650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3397224913177352e-06,
+      "loss": 22.2497,
+      "step": 11651
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3367570210217506e-06,
+      "loss": 19.3994,
+      "step": 11652
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3337947919236427e-06,
+      "loss": 0.2325,
+      "step": 11653
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3308358042207147e-06,
+      "loss": 0.941,
+      "step": 11654
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3278800581100591e-06,
+      "loss": 20.1946,
+      "step": 11655
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3249275537885243e-06,
+      "loss": 19.4846,
+      "step": 11656
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3219782914527634e-06,
+      "loss": 0.911,
+      "step": 11657
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.319032271299231e-06,
+      "loss": 18.9372,
+      "step": 11658
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3160894935241197e-06,
+      "loss": 19.8592,
+      "step": 11659
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3131499583234397e-06,
+      "loss": 18.3347,
+      "step": 11660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3102136658929954e-06,
+      "loss": 0.224,
+      "step": 11661
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3072806164283358e-06,
+      "loss": 18.4803,
+      "step": 11662
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3043508101248213e-06,
+      "loss": 18.5775,
+      "step": 11663
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3014242471776018e-06,
+      "loss": 18.4297,
+      "step": 11664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2985009277815819e-06,
+      "loss": 17.4667,
+      "step": 11665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2955808521314838e-06,
+      "loss": 18.8738,
+      "step": 11666
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.29266402042178e-06,
+      "loss": 18.552,
+      "step": 11667
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2897504328467703e-06,
+      "loss": 0.9434,
+      "step": 11668
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.286840089600494e-06,
+      "loss": 19.3863,
+      "step": 11669
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2839329908767961e-06,
+      "loss": 0.9407,
+      "step": 11670
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2810291368693107e-06,
+      "loss": 0.2035,
+      "step": 11671
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2781285277714439e-06,
+      "loss": 0.2323,
+      "step": 11672
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.275231163776386e-06,
+      "loss": 17.758,
+      "step": 11673
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2723370450771155e-06,
+      "loss": 0.981,
+      "step": 11674
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2694461718664007e-06,
+      "loss": 19.1022,
+      "step": 11675
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2665585443367767e-06,
+      "loss": 0.9522,
+      "step": 11676
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2636741626805837e-06,
+      "loss": 19.5705,
+      "step": 11677
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2607930270899294e-06,
+      "loss": 19.2141,
+      "step": 11678
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.25791513775671e-06,
+      "loss": 0.2071,
+      "step": 11679
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2550404948726114e-06,
+      "loss": 0.9367,
+      "step": 11680
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2521690986290913e-06,
+      "loss": 0.9634,
+      "step": 11681
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2493009492174024e-06,
+      "loss": 20.7543,
+      "step": 11682
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2464360468285752e-06,
+      "loss": 0.9388,
+      "step": 11683
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2435743916534292e-06,
+      "loss": 18.9414,
+      "step": 11684
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2407159838825566e-06,
+      "loss": 0.2546,
+      "step": 11685
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2378608237063494e-06,
+      "loss": 0.9057,
+      "step": 11686
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2350089113149665e-06,
+      "loss": 18.1033,
+      "step": 11687
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2321602468983617e-06,
+      "loss": 17.5136,
+      "step": 11688
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.229314830646272e-06,
+      "loss": 19.3719,
+      "step": 11689
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2264726627482125e-06,
+      "loss": 0.9106,
+      "step": 11690
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2236337433934819e-06,
+      "loss": 0.9366,
+      "step": 11691
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2207980727711731e-06,
+      "loss": 19.1331,
+      "step": 11692
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2179656510701464e-06,
+      "loss": 0.8941,
+      "step": 11693
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.215136478479062e-06,
+      "loss": 0.1805,
+      "step": 11694
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2123105551863467e-06,
+      "loss": 0.9419,
+      "step": 11695
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2094878813802335e-06,
+      "loss": 0.2178,
+      "step": 11696
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2066684572487107e-06,
+      "loss": 0.2071,
+      "step": 11697
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2038522829795728e-06,
+      "loss": 18.2498,
+      "step": 11698
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2010393587603974e-06,
+      "loss": 18.1896,
+      "step": 11699
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1982296847785179e-06,
+      "loss": 20.1347,
+      "step": 11700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1954232612210902e-06,
+      "loss": 19.1568,
+      "step": 11701
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1926200882750372e-06,
+      "loss": 0.186,
+      "step": 11702
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1898201661270369e-06,
+      "loss": 0.9502,
+      "step": 11703
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1870234949636073e-06,
+      "loss": 0.192,
+      "step": 11704
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.18423007497101e-06,
+      "loss": 0.9269,
+      "step": 11705
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.181439906335291e-06,
+      "loss": 0.9408,
+      "step": 11706
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1786529892422905e-06,
+      "loss": 19.4162,
+      "step": 11707
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.175869323877643e-06,
+      "loss": 17.687,
+      "step": 11708
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.173088910426745e-06,
+      "loss": 0.2078,
+      "step": 11709
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.170311749074776e-06,
+      "loss": 0.9569,
+      "step": 11710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1675378400067217e-06,
+      "loss": 18.7801,
+      "step": 11711
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1647671834073393e-06,
+      "loss": 0.9315,
+      "step": 11712
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1619997794611537e-06,
+      "loss": 21.0806,
+      "step": 11713
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1592356283524897e-06,
+      "loss": 0.9047,
+      "step": 11714
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.156474730265461e-06,
+      "loss": 0.1942,
+      "step": 11715
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.153717085383954e-06,
+      "loss": 0.9293,
+      "step": 11716
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1509626938916273e-06,
+      "loss": 17.8053,
+      "step": 11717
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.148211555971962e-06,
+      "loss": 19.4142,
+      "step": 11718
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1454636718081725e-06,
+      "loss": 0.9533,
+      "step": 11719
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1427190415832955e-06,
+      "loss": 19.1523,
+      "step": 11720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1399776654801242e-06,
+      "loss": 0.974,
+      "step": 11721
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1372395436812566e-06,
+      "loss": 18.7384,
+      "step": 11722
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1345046763690636e-06,
+      "loss": 20.3197,
+      "step": 11723
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1317730637256941e-06,
+      "loss": 0.9506,
+      "step": 11724
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.129044705933091e-06,
+      "loss": 18.4902,
+      "step": 11725
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1263196031729762e-06,
+      "loss": 20.5775,
+      "step": 11726
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1235977556268484e-06,
+      "loss": 19.59,
+      "step": 11727
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1208791634760019e-06,
+      "loss": 20.1223,
+      "step": 11728
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1181638269015027e-06,
+      "loss": 21.12,
+      "step": 11729
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1154517460842061e-06,
+      "loss": 0.9478,
+      "step": 11730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1127429212047513e-06,
+      "loss": 0.9513,
+      "step": 11731
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1100373524435549e-06,
+      "loss": 0.2312,
+      "step": 11732
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.107335039980828e-06,
+      "loss": 0.9314,
+      "step": 11733
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1046359839965493e-06,
+      "loss": 0.9656,
+      "step": 11734
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1019401846704912e-06,
+      "loss": 0.9933,
+      "step": 11735
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.099247642182205e-06,
+      "loss": 0.8725,
+      "step": 11736
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0965583567110305e-06,
+      "loss": 0.9118,
+      "step": 11737
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0938723284360797e-06,
+      "loss": 18.757,
+      "step": 11738
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0911895575362652e-06,
+      "loss": 18.7014,
+      "step": 11739
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0885100441902608e-06,
+      "loss": 0.1527,
+      "step": 11740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0858337885765402e-06,
+      "loss": 20.4301,
+      "step": 11741
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0831607908733555e-06,
+      "loss": 19.16,
+      "step": 11742
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0804910512587418e-06,
+      "loss": 21.7119,
+      "step": 11743
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0778245699105017e-06,
+      "loss": 18.1858,
+      "step": 11744
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0751613470062538e-06,
+      "loss": 0.2037,
+      "step": 11745
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0725013827233787e-06,
+      "loss": 18.1863,
+      "step": 11746
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0698446772390292e-06,
+      "loss": 0.9524,
+      "step": 11747
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.067191230730169e-06,
+      "loss": 18.622,
+      "step": 11748
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0645410433735236e-06,
+      "loss": 0.9557,
+      "step": 11749
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0618941153456018e-06,
+      "loss": 0.9499,
+      "step": 11750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0592504468227127e-06,
+      "loss": 0.9823,
+      "step": 11751
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0566100379809263e-06,
+      "loss": 18.5716,
+      "step": 11752
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0539728889961187e-06,
+      "loss": 16.8055,
+      "step": 11753
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0513390000439216e-06,
+      "loss": 0.2,
+      "step": 11754
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.048708371299778e-06,
+      "loss": 18.5807,
+      "step": 11755
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0460810029388923e-06,
+      "loss": 19.9223,
+      "step": 11756
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0434568951362578e-06,
+      "loss": 19.4037,
+      "step": 11757
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0408360480666568e-06,
+      "loss": 19.3614,
+      "step": 11758
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0382184619046498e-06,
+      "loss": 0.9243,
+      "step": 11759
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.035604136824575e-06,
+      "loss": 0.909,
+      "step": 11760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.03299307300056e-06,
+      "loss": 0.9879,
+      "step": 11761
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0303852706065265e-06,
+      "loss": 0.9362,
+      "step": 11762
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0277807298161467e-06,
+      "loss": 18.049,
+      "step": 11763
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0251794508029044e-06,
+      "loss": 17.9619,
+      "step": 11764
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0225814337400608e-06,
+      "loss": 17.8788,
+      "step": 11765
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0199866788006496e-06,
+      "loss": 18.1005,
+      "step": 11766
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.017395186157488e-06,
+      "loss": 18.0945,
+      "step": 11767
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0148069559831996e-06,
+      "loss": 0.9602,
+      "step": 11768
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0122219884501626e-06,
+      "loss": 18.5177,
+      "step": 11769
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0096402837305396e-06,
+      "loss": 0.9522,
+      "step": 11770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0070618419962984e-06,
+      "loss": 0.9082,
+      "step": 11771
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.004486663419163e-06,
+      "loss": 18.8026,
+      "step": 11772
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0019147481706625e-06,
+      "loss": 0.9184,
+      "step": 11773
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.99346096422088e-07,
+      "loss": 0.1788,
+      "step": 11774
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.967807083445358e-07,
+      "loss": 19.0692,
+      "step": 11775
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.942185841088637e-07,
+      "loss": 0.9551,
+      "step": 11776
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.916597238857183e-07,
+      "loss": 20.55,
+      "step": 11777
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.891041278455416e-07,
+      "loss": 18.4376,
+      "step": 11778
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.865517961585468e-07,
+      "loss": 18.346,
+      "step": 11779
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.840027289947207e-07,
+      "loss": 20.7681,
+      "step": 11780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.814569265238548e-07,
+      "loss": 0.9159,
+      "step": 11781
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.789143889155028e-07,
+      "loss": 0.229,
+      "step": 11782
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.763751163390123e-07,
+      "loss": 0.9486,
+      "step": 11783
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.738391089635147e-07,
+      "loss": 0.9853,
+      "step": 11784
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.713063669579192e-07,
+      "loss": 19.6576,
+      "step": 11785
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.687768904909079e-07,
+      "loss": 19.0047,
+      "step": 11786
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.66250679730968e-07,
+      "loss": 19.6472,
+      "step": 11787
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.637277348463536e-07,
+      "loss": 19.0673,
+      "step": 11788
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.61208056005103e-07,
+      "loss": 0.9516,
+      "step": 11789
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.58691643375037e-07,
+      "loss": 0.1901,
+      "step": 11790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.56178497123761e-07,
+      "loss": 18.5082,
+      "step": 11791
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.536686174186682e-07,
+      "loss": 0.9517,
+      "step": 11792
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.511620044269199e-07,
+      "loss": 0.1888,
+      "step": 11793
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.48658658315471e-07,
+      "loss": 19.1151,
+      "step": 11794
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.461585792510552e-07,
+      "loss": 0.1921,
+      "step": 11795
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.436617674001947e-07,
+      "loss": 20.565,
+      "step": 11796
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.411682229291785e-07,
+      "loss": 19.4635,
+      "step": 11797
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.386779460041017e-07,
+      "loss": 0.9337,
+      "step": 11798
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.36190936790815e-07,
+      "loss": 18.9578,
+      "step": 11799
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.337071954549804e-07,
+      "loss": 19.3727,
+      "step": 11800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.312267221620041e-07,
+      "loss": 0.9017,
+      "step": 11801
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.287495170771154e-07,
+      "loss": 0.9694,
+      "step": 11802
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.2627558036531e-07,
+      "loss": 19.3743,
+      "step": 11803
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.23804912191345e-07,
+      "loss": 0.1824,
+      "step": 11804
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.213375127197943e-07,
+      "loss": 19.665,
+      "step": 11805
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.188733821149986e-07,
+      "loss": 0.964,
+      "step": 11806
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.164125205410656e-07,
+      "loss": 16.8846,
+      "step": 11807
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.139549281619087e-07,
+      "loss": 18.3888,
+      "step": 11808
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.115006051412245e-07,
+      "loss": 18.153,
+      "step": 11809
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.090495516424713e-07,
+      "loss": 18.0443,
+      "step": 11810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.066017678288963e-07,
+      "loss": 0.918,
+      "step": 11811
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.041572538635468e-07,
+      "loss": 0.9315,
+      "step": 11812
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.017160099092315e-07,
+      "loss": 0.1861,
+      "step": 11813
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.992780361285424e-07,
+      "loss": 0.9537,
+      "step": 11814
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.968433326838777e-07,
+      "loss": 0.218,
+      "step": 11815
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.944118997373796e-07,
+      "loss": 19.3484,
+      "step": 11816
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.91983737451002e-07,
+      "loss": 18.6108,
+      "step": 11817
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.895588459864712e-07,
+      "loss": 0.2156,
+      "step": 11818
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.871372255053024e-07,
+      "loss": 18.1946,
+      "step": 11819
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.847188761687775e-07,
+      "loss": 20.509,
+      "step": 11820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.823037981379733e-07,
+      "loss": 0.8936,
+      "step": 11821
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.798919915737503e-07,
+      "loss": 19.2182,
+      "step": 11822
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.774834566367352e-07,
+      "loss": 18.9161,
+      "step": 11823
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.750781934873553e-07,
+      "loss": 0.2515,
+      "step": 11824
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.726762022858104e-07,
+      "loss": 0.1796,
+      "step": 11825
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.702774831920835e-07,
+      "loss": 0.9539,
+      "step": 11826
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.678820363659412e-07,
+      "loss": 0.912,
+      "step": 11827
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.654898619669283e-07,
+      "loss": 19.2291,
+      "step": 11828
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.631009601543838e-07,
+      "loss": 19.417,
+      "step": 11829
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.607153310874083e-07,
+      "loss": 17.8795,
+      "step": 11830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.583329749249025e-07,
+      "loss": 19.5047,
+      "step": 11831
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.55953891825545e-07,
+      "loss": 18.3917,
+      "step": 11832
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.535780819477867e-07,
+      "loss": 0.1778,
+      "step": 11833
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.51205545449868e-07,
+      "loss": 19.6028,
+      "step": 11834
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.488362824898177e-07,
+      "loss": 0.9335,
+      "step": 11835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.46470293225432e-07,
+      "loss": 18.2057,
+      "step": 11836
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.441075778143015e-07,
+      "loss": 0.9413,
+      "step": 11837
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.417481364137947e-07,
+      "loss": 0.2144,
+      "step": 11838
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.393919691810581e-07,
+      "loss": 0.2229,
+      "step": 11839
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.370390762730218e-07,
+      "loss": 0.9322,
+      "step": 11840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.346894578464049e-07,
+      "loss": 19.7483,
+      "step": 11841
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.323431140577043e-07,
+      "loss": 18.3996,
+      "step": 11842
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.300000450631896e-07,
+      "loss": 0.1837,
+      "step": 11843
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.276602510189191e-07,
+      "loss": 18.3705,
+      "step": 11844
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.253237320807461e-07,
+      "loss": 19.8185,
+      "step": 11845
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.229904884042849e-07,
+      "loss": 0.1848,
+      "step": 11846
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.206605201449447e-07,
+      "loss": 18.7907,
+      "step": 11847
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.183338274578956e-07,
+      "loss": 18.9563,
+      "step": 11848
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.160104104981303e-07,
+      "loss": 0.2278,
+      "step": 11849
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.136902694203919e-07,
+      "loss": 0.2116,
+      "step": 11850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.113734043792009e-07,
+      "loss": 18.9157,
+      "step": 11851
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.090598155288786e-07,
+      "loss": 19.443,
+      "step": 11852
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.067495030235295e-07,
+      "loss": 0.2357,
+      "step": 11853
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.044424670170192e-07,
+      "loss": 20.1566,
+      "step": 11854
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.021387076630083e-07,
+      "loss": 0.918,
+      "step": 11855
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.998382251149517e-07,
+      "loss": 0.9128,
+      "step": 11856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.975410195260491e-07,
+      "loss": 20.7127,
+      "step": 11857
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.952470910493171e-07,
+      "loss": 19.2121,
+      "step": 11858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.9295643983755e-07,
+      "loss": 0.9711,
+      "step": 11859
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.906690660433091e-07,
+      "loss": 18.7444,
+      "step": 11860
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.883849698189338e-07,
+      "loss": 18.2196,
+      "step": 11861
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.86104151316569e-07,
+      "loss": 0.9341,
+      "step": 11862
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.838266106881265e-07,
+      "loss": 18.0841,
+      "step": 11863
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.81552348085296e-07,
+      "loss": 0.9453,
+      "step": 11864
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.792813636595508e-07,
+      "loss": 19.6801,
+      "step": 11865
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.77013657562159e-07,
+      "loss": 19.2984,
+      "step": 11866
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.74749229944155e-07,
+      "loss": 0.1755,
+      "step": 11867
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.724880809563572e-07,
+      "loss": 17.5623,
+      "step": 11868
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.702302107493786e-07,
+      "loss": 18.7713,
+      "step": 11869
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.679756194735932e-07,
+      "loss": 18.5074,
+      "step": 11870
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.657243072791698e-07,
+      "loss": 17.2466,
+      "step": 11871
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.634762743160662e-07,
+      "loss": 19.0178,
+      "step": 11872
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.61231520733996e-07,
+      "loss": 0.9315,
+      "step": 11873
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.589900466824784e-07,
+      "loss": 18.0539,
+      "step": 11874
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.567518523108052e-07,
+      "loss": 0.9572,
+      "step": 11875
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.545169377680516e-07,
+      "loss": 17.8533,
+      "step": 11876
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.522853032030708e-07,
+      "loss": 0.1934,
+      "step": 11877
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.500569487645049e-07,
+      "loss": 16.8503,
+      "step": 11878
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.478318746007685e-07,
+      "loss": 20.2361,
+      "step": 11879
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.456100808600597e-07,
+      "loss": 0.1991,
+      "step": 11880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.433915676903658e-07,
+      "loss": 19.051,
+      "step": 11881
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.411763352394463e-07,
+      "loss": 19.144,
+      "step": 11882
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.389643836548444e-07,
+      "loss": 0.9075,
+      "step": 11883
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.36755713083892e-07,
+      "loss": 17.7746,
+      "step": 11884
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.345503236736939e-07,
+      "loss": 0.9028,
+      "step": 11885
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.323482155711381e-07,
+      "loss": 0.955,
+      "step": 11886
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.301493889228961e-07,
+      "loss": 0.9435,
+      "step": 11887
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.27953843875423e-07,
+      "loss": 0.9267,
+      "step": 11888
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.257615805749462e-07,
+      "loss": 0.2092,
+      "step": 11889
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.23572599167488e-07,
+      "loss": 0.9684,
+      "step": 11890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.213868997988371e-07,
+      "loss": 0.9426,
+      "step": 11891
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.192044826145771e-07,
+      "loss": 0.9187,
+      "step": 11892
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.170253477600642e-07,
+      "loss": 0.8874,
+      "step": 11893
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.148494953804374e-07,
+      "loss": 0.9278,
+      "step": 11894
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.126769256206256e-07,
+      "loss": 0.9574,
+      "step": 11895
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.10507638625324e-07,
+      "loss": 0.9889,
+      "step": 11896
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.083416345390225e-07,
+      "loss": 0.9299,
+      "step": 11897
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.061789135059782e-07,
+      "loss": 18.9348,
+      "step": 11898
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.040194756702478e-07,
+      "loss": 18.6486,
+      "step": 11899
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.018633211756609e-07,
+      "loss": 20.6071,
+      "step": 11900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.997104501658191e-07,
+      "loss": 19.3882,
+      "step": 11901
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.975608627841191e-07,
+      "loss": 0.1902,
+      "step": 11902
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.954145591737348e-07,
+      "loss": 19.1558,
+      "step": 11903
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.93271539477619e-07,
+      "loss": 0.2079,
+      "step": 11904
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.911318038384962e-07,
+      "loss": 0.9334,
+      "step": 11905
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.889953523988968e-07,
+      "loss": 19.0201,
+      "step": 11906
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.868621853011181e-07,
+      "loss": 18.4191,
+      "step": 11907
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.847323026872299e-07,
+      "loss": 17.4185,
+      "step": 11908
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.826057046990964e-07,
+      "loss": 17.8149,
+      "step": 11909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.804823914783653e-07,
+      "loss": 0.9796,
+      "step": 11910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.783623631664404e-07,
+      "loss": 18.785,
+      "step": 11911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.762456199045475e-07,
+      "loss": 20.5426,
+      "step": 11912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.741321618336627e-07,
+      "loss": 0.1957,
+      "step": 11913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.720219890945511e-07,
+      "loss": 19.139,
+      "step": 11914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.699151018277505e-07,
+      "loss": 0.2111,
+      "step": 11915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.678115001736097e-07,
+      "loss": 21.3575,
+      "step": 11916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.657111842722274e-07,
+      "loss": 20.2455,
+      "step": 11917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.636141542634866e-07,
+      "loss": 20.3603,
+      "step": 11918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.615204102870753e-07,
+      "loss": 0.9443,
+      "step": 11919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.594299524824377e-07,
+      "loss": 21.2391,
+      "step": 11920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.573427809888067e-07,
+      "loss": 20.7462,
+      "step": 11921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.552588959451988e-07,
+      "loss": 0.1884,
+      "step": 11922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.531782974904144e-07,
+      "loss": 17.956,
+      "step": 11923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.511009857630257e-07,
+      "loss": 20.318,
+      "step": 11924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.490269609013888e-07,
+      "loss": 0.8794,
+      "step": 11925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.469562230436599e-07,
+      "loss": 20.4723,
+      "step": 11926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.448887723277452e-07,
+      "loss": 0.192,
+      "step": 11927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.428246088913459e-07,
+      "loss": 18.3816,
+      "step": 11928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.407637328719462e-07,
+      "loss": 20.6151,
+      "step": 11929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.387061444068143e-07,
+      "loss": 18.7358,
+      "step": 11930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.36651843632996e-07,
+      "loss": 18.2718,
+      "step": 11931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.346008306873152e-07,
+      "loss": 0.9559,
+      "step": 11932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.325531057063738e-07,
+      "loss": 19.4002,
+      "step": 11933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.305086688265682e-07,
+      "loss": 17.9744,
+      "step": 11934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.284675201840561e-07,
+      "loss": 19.2457,
+      "step": 11935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.264296599148012e-07,
+      "loss": 0.9335,
+      "step": 11936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.24395088154528e-07,
+      "loss": 0.9695,
+      "step": 11937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.223638050387448e-07,
+      "loss": 19.2971,
+      "step": 11938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.20335810702749e-07,
+      "loss": 19.9452,
+      "step": 11939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.183111052816159e-07,
+      "loss": 0.9319,
+      "step": 11940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.162896889101932e-07,
+      "loss": 17.1687,
+      "step": 11941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.142715617231232e-07,
+      "loss": 17.3678,
+      "step": 11942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.122567238548204e-07,
+      "loss": 0.9638,
+      "step": 11943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.102451754394833e-07,
+      "loss": 0.9212,
+      "step": 11944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.082369166110879e-07,
+      "loss": 0.8967,
+      "step": 11945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.062319475033995e-07,
+      "loss": 19.9029,
+      "step": 11946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.042302682499501e-07,
+      "loss": 17.602,
+      "step": 11947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.022318789840664e-07,
+      "loss": 20.925,
+      "step": 11948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.002367798388475e-07,
+      "loss": 0.9237,
+      "step": 11949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.982449709471816e-07,
+      "loss": 0.9416,
+      "step": 11950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.962564524417291e-07,
+      "loss": 0.9836,
+      "step": 11951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.942712244549232e-07,
+      "loss": 17.3366,
+      "step": 11952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.922892871190077e-07,
+      "loss": 18.4612,
+      "step": 11953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.903106405659831e-07,
+      "loss": 18.2327,
+      "step": 11954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.883352849276269e-07,
+      "loss": 0.917,
+      "step": 11955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.86363220335523e-07,
+      "loss": 0.9983,
+      "step": 11956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.843944469210105e-07,
+      "loss": 0.9526,
+      "step": 11957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.824289648152126e-07,
+      "loss": 19.1601,
+      "step": 11958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.804667741490521e-07,
+      "loss": 18.4575,
+      "step": 11959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.785078750532136e-07,
+      "loss": 0.9353,
+      "step": 11960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.765522676581703e-07,
+      "loss": 0.8522,
+      "step": 11961
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.745999520941737e-07,
+      "loss": 0.176,
+      "step": 11962
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.726509284912585e-07,
+      "loss": 19.4243,
+      "step": 11963
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.707051969792376e-07,
+      "loss": 17.4348,
+      "step": 11964
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.687627576877019e-07,
+      "loss": 0.9666,
+      "step": 11965
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.668236107460368e-07,
+      "loss": 20.9538,
+      "step": 11966
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.648877562833943e-07,
+      "loss": 17.4093,
+      "step": 11967
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.629551944287104e-07,
+      "loss": 19.6373,
+      "step": 11968
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.610259253106931e-07,
+      "loss": 19.5433,
+      "step": 11969
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.590999490578619e-07,
+      "loss": 0.9417,
+      "step": 11970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.571772657984752e-07,
+      "loss": 0.8955,
+      "step": 11971
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.552578756605975e-07,
+      "loss": 0.9285,
+      "step": 11972
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.533417787720819e-07,
+      "loss": 0.9106,
+      "step": 11973
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.514289752605373e-07,
+      "loss": 17.7074,
+      "step": 11974
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.495194652533619e-07,
+      "loss": 19.091,
+      "step": 11975
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.476132488777541e-07,
+      "loss": 21.0966,
+      "step": 11976
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.457103262606622e-07,
+      "loss": 0.9803,
+      "step": 11977
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.438106975288349e-07,
+      "loss": 21.1986,
+      "step": 11978
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.419143628087931e-07,
+      "loss": 19.1325,
+      "step": 11979
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.400213222268413e-07,
+      "loss": 20.7678,
+      "step": 11980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.381315759090733e-07,
+      "loss": 0.938,
+      "step": 11981
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.362451239813493e-07,
+      "loss": 0.9227,
+      "step": 11982
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.343619665693134e-07,
+      "loss": 0.2084,
+      "step": 11983
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.324821037983929e-07,
+      "loss": 19.3013,
+      "step": 11984
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.30605535793799e-07,
+      "loss": 19.0315,
+      "step": 11985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.287322626805202e-07,
+      "loss": 0.9506,
+      "step": 11986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.268622845833182e-07,
+      "loss": 19.3177,
+      "step": 11987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.249956016267543e-07,
+      "loss": 18.6419,
+      "step": 11988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.231322139351457e-07,
+      "loss": 0.1835,
+      "step": 11989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.212721216326099e-07,
+      "loss": 18.4044,
+      "step": 11990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.194153248430367e-07,
+      "loss": 18.0839,
+      "step": 11991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.175618236900937e-07,
+      "loss": 19.7332,
+      "step": 11992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.157116182972377e-07,
+      "loss": 18.9109,
+      "step": 11993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.13864708787698e-07,
+      "loss": 0.9739,
+      "step": 11994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.120210952844872e-07,
+      "loss": 0.1949,
+      "step": 11995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.101807779104018e-07,
+      "loss": 0.2229,
+      "step": 11996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.083437567880101e-07,
+      "loss": 0.9458,
+      "step": 11997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.065100320396698e-07,
+      "loss": 18.0413,
+      "step": 11998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.046796037875167e-07,
+      "loss": 0.948,
+      "step": 11999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.028524721534644e-07,
+      "loss": 0.9664,
+      "step": 12000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.010286372592099e-07,
+      "loss": 0.2034,
+      "step": 12001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.992080992262227e-07,
+      "loss": 0.9403,
+      "step": 12002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.973908581757669e-07,
+      "loss": 0.2253,
+      "step": 12003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.955769142288791e-07,
+      "loss": 19.0391,
+      "step": 12004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.937662675063626e-07,
+      "loss": 0.1993,
+      "step": 12005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.91958918128832e-07,
+      "loss": 1.0008,
+      "step": 12006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.901548662166577e-07,
+      "loss": 19.7401,
+      "step": 12007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.883541118899993e-07,
+      "loss": 0.19,
+      "step": 12008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.865566552687939e-07,
+      "loss": 0.9596,
+      "step": 12009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.847624964727627e-07,
+      "loss": 18.0554,
+      "step": 12010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.829716356214042e-07,
+      "loss": 21.3343,
+      "step": 12011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.811840728339956e-07,
+      "loss": 19.0247,
+      "step": 12012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.793998082295969e-07,
+      "loss": 19.4268,
+      "step": 12013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.776188419270577e-07,
+      "loss": 15.9318,
+      "step": 12014
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7584117404498865e-07,
+      "loss": 19.5638,
+      "step": 12015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.740668047017893e-07,
+      "loss": 0.9002,
+      "step": 12016
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.722957340156486e-07,
+      "loss": 0.9081,
+      "step": 12017
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.705279621045222e-07,
+      "loss": 18.099,
+      "step": 12018
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6876348908615477e-07,
+      "loss": 19.4841,
+      "step": 12019
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6700231507806893e-07,
+      "loss": 1.0374,
+      "step": 12020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.652444401975653e-07,
+      "loss": 0.8909,
+      "step": 12021
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6348986456172807e-07,
+      "loss": 19.1388,
+      "step": 12022
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.617385882874192e-07,
+      "loss": 18.4211,
+      "step": 12023
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5999061149127887e-07,
+      "loss": 18.7931,
+      "step": 12024
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.582459342897361e-07,
+      "loss": 18.5939,
+      "step": 12025
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.565045567989923e-07,
+      "loss": 19.0546,
+      "step": 12026
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5476647913502703e-07,
+      "loss": 20.0476,
+      "step": 12027
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.530317014136143e-07,
+      "loss": 0.2083,
+      "step": 12028
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5130022375028393e-07,
+      "loss": 0.2051,
+      "step": 12029
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4957204626037695e-07,
+      "loss": 20.7946,
+      "step": 12030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.478471690589847e-07,
+      "loss": 21.3228,
+      "step": 12031
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.461255922609986e-07,
+      "loss": 20.4743,
+      "step": 12032
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4440731598107686e-07,
+      "loss": 0.907,
+      "step": 12033
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.42692340333678e-07,
+      "loss": 0.1891,
+      "step": 12034
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4098066543301065e-07,
+      "loss": 18.612,
+      "step": 12035
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.392722913930947e-07,
+      "loss": 0.9042,
+      "step": 12036
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3756721832770576e-07,
+      "loss": 0.9622,
+      "step": 12037
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.358654463504197e-07,
+      "loss": 19.0553,
+      "step": 12038
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3416697557457364e-07,
+      "loss": 0.9131,
+      "step": 12039
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.324718061132937e-07,
+      "loss": 0.9418,
+      "step": 12040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.307799380794897e-07,
+      "loss": 0.9179,
+      "step": 12041
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.290913715858491e-07,
+      "loss": 18.9935,
+      "step": 12042
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2740610674483207e-07,
+      "loss": 20.5948,
+      "step": 12043
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.257241436686932e-07,
+      "loss": 0.9652,
+      "step": 12044
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.24045482469454e-07,
+      "loss": 18.5079,
+      "step": 12045
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.22370123258925e-07,
+      "loss": 0.9408,
+      "step": 12046
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.206980661486837e-07,
+      "loss": 19.8603,
+      "step": 12047
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.190293112501076e-07,
+      "loss": 20.3798,
+      "step": 12048
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.173638586743356e-07,
+      "loss": 18.1856,
+      "step": 12049
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.157017085323012e-07,
+      "loss": 0.2119,
+      "step": 12050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1404286093471025e-07,
+      "loss": 0.2221,
+      "step": 12051
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.123873159920411e-07,
+      "loss": 0.9301,
+      "step": 12052
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.107350738145721e-07,
+      "loss": 0.8965,
+      "step": 12053
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.090861345123431e-07,
+      "loss": 19.7683,
+      "step": 12054
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.074404981951885e-07,
+      "loss": 0.9604,
+      "step": 12055
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0579816497270384e-07,
+      "loss": 0.9658,
+      "step": 12056
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0415913495427946e-07,
+      "loss": 18.5208,
+      "step": 12057
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.025234082490947e-07,
+      "loss": 0.9768,
+      "step": 12058
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0089098496607915e-07,
+      "loss": 19.0349,
+      "step": 12059
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9926186521397347e-07,
+      "loss": 0.9716,
+      "step": 12060
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.976360491012798e-07,
+      "loss": 19.7374,
+      "step": 12061
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.960135367362783e-07,
+      "loss": 0.8935,
+      "step": 12062
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.943943282270435e-07,
+      "loss": 19.8825,
+      "step": 12063
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9277842368142805e-07,
+      "loss": 0.909,
+      "step": 12064
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.911658232070459e-07,
+      "loss": 0.982,
+      "step": 12065
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.895565269113055e-07,
+      "loss": 18.484,
+      "step": 12066
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8795053490139897e-07,
+      "loss": 19.1712,
+      "step": 12067
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8634784728429077e-07,
+      "loss": 18.645,
+      "step": 12068
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8474846416672874e-07,
+      "loss": 21.022,
+      "step": 12069
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.831523856552388e-07,
+      "loss": 20.531,
+      "step": 12070
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.815596118561249e-07,
+      "loss": 17.5298,
+      "step": 12071
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7997014287547427e-07,
+      "loss": 0.9369,
+      "step": 12072
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.783839788191468e-07,
+      "loss": 0.963,
+      "step": 12073
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.768011197928079e-07,
+      "loss": 0.9761,
+      "step": 12074
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.752215659018621e-07,
+      "loss": 21.1516,
+      "step": 12075
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7364531725151973e-07,
+      "loss": 0.9292,
+      "step": 12076
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.720723739467746e-07,
+      "loss": 0.2197,
+      "step": 12077
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.705027360923874e-07,
+      "loss": 18.5727,
+      "step": 12078
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.689364037929022e-07,
+      "loss": 0.2129,
+      "step": 12079
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.673733771526466e-07,
+      "loss": 20.8295,
+      "step": 12080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6581365627572086e-07,
+      "loss": 0.9365,
+      "step": 12081
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.642572412660139e-07,
+      "loss": 19.8873,
+      "step": 12082
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6270413222718736e-07,
+      "loss": 19.2469,
+      "step": 12083
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.611543292626862e-07,
+      "loss": 0.9096,
+      "step": 12084
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5960783247573346e-07,
+      "loss": 20.4983,
+      "step": 12085
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5806464196933563e-07,
+      "loss": 18.9165,
+      "step": 12086
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5652475784627715e-07,
+      "loss": 0.8822,
+      "step": 12087
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.549881802091204e-07,
+      "loss": 21.0205,
+      "step": 12088
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5345490916020575e-07,
+      "loss": 0.1839,
+      "step": 12089
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.519249448016626e-07,
+      "loss": 20.5677,
+      "step": 12090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.503982872353817e-07,
+      "loss": 0.1909,
+      "step": 12091
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4887493656305947e-07,
+      "loss": 18.2421,
+      "step": 12092
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.473548928861481e-07,
+      "loss": 17.8276,
+      "step": 12093
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.458381563058943e-07,
+      "loss": 17.5801,
+      "step": 12094
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.443247269233174e-07,
+      "loss": 20.4925,
+      "step": 12095
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4281460483922554e-07,
+      "loss": 0.9211,
+      "step": 12096
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.413077901541883e-07,
+      "loss": 19.5683,
+      "step": 12097
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.398042829685755e-07,
+      "loss": 19.5155,
+      "step": 12098
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.383040833825235e-07,
+      "loss": 19.9743,
+      "step": 12099
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3680719149595805e-07,
+      "loss": 0.9527,
+      "step": 12100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3531360740857165e-07,
+      "loss": 17.9123,
+      "step": 12101
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.338233312198458e-07,
+      "loss": 0.9517,
+      "step": 12102
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3233636302904546e-07,
+      "loss": 18.647,
+      "step": 12103
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3085270293520265e-07,
+      "loss": 0.182,
+      "step": 12104
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2937235103713826e-07,
+      "loss": 0.1996,
+      "step": 12105
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.278953074334512e-07,
+      "loss": 0.2155,
+      "step": 12106
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.264215722225239e-07,
+      "loss": 19.6468,
+      "step": 12107
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.249511455025056e-07,
+      "loss": 18.3158,
+      "step": 12108
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2348402737133467e-07,
+      "loss": 0.9409,
+      "step": 12109
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.220202179267329e-07,
+      "loss": 18.8049,
+      "step": 12110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2055971726620005e-07,
+      "loss": 20.8173,
+      "step": 12111
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.191025254870028e-07,
+      "loss": 0.9641,
+      "step": 12112
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.176486426861969e-07,
+      "loss": 18.9204,
+      "step": 12113
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1619806896062697e-07,
+      "loss": 19.8268,
+      "step": 12114
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1475080440690477e-07,
+      "loss": 20.3256,
+      "step": 12115
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.133068491214142e-07,
+      "loss": 0.9555,
+      "step": 12116
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.11866203200345e-07,
+      "loss": 18.9884,
+      "step": 12117
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1042886673964264e-07,
+      "loss": 0.987,
+      "step": 12118
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0899483983504176e-07,
+      "loss": 18.405,
+      "step": 12119
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.075641225820491e-07,
+      "loss": 19.7396,
+      "step": 12120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.06136715075972e-07,
+      "loss": 0.8938,
+      "step": 12121
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0471261741186754e-07,
+      "loss": 0.9332,
+      "step": 12122
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0329182968459323e-07,
+      "loss": 19.195,
+      "step": 12123
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.01874351988779e-07,
+      "loss": 20.6261,
+      "step": 12124
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.004601844188437e-07,
+      "loss": 20.6708,
+      "step": 12125
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9904932706896205e-07,
+      "loss": 0.9083,
+      "step": 12126
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.976417800331144e-07,
+      "loss": 0.9098,
+      "step": 12127
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9623754340504797e-07,
+      "loss": 0.9726,
+      "step": 12128
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.948366172782935e-07,
+      "loss": 0.9362,
+      "step": 12129
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.934390017461541e-07,
+      "loss": 19.5638,
+      "step": 12130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9204469690171655e-07,
+      "loss": 19.1326,
+      "step": 12131
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.906537028378564e-07,
+      "loss": 0.9178,
+      "step": 12132
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8926601964721635e-07,
+      "loss": 20.1624,
+      "step": 12133
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.878816474222168e-07,
+      "loss": 0.1964,
+      "step": 12134
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.865005862550674e-07,
+      "loss": 19.2425,
+      "step": 12135
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.851228362377556e-07,
+      "loss": 0.1908,
+      "step": 12136
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8374839746204694e-07,
+      "loss": 0.9064,
+      "step": 12137
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8237727001947934e-07,
+      "loss": 20.4878,
+      "step": 12138
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.810094540013797e-07,
+      "loss": 17.8363,
+      "step": 12139
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.796449494988529e-07,
+      "loss": 18.6632,
+      "step": 12140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7828375660278183e-07,
+      "loss": 0.2171,
+      "step": 12141
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7692587540382174e-07,
+      "loss": 19.1,
+      "step": 12142
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.75571305992417e-07,
+      "loss": 0.9507,
+      "step": 12143
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.742200484587898e-07,
+      "loss": 0.9546,
+      "step": 12144
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7287210289293486e-07,
+      "loss": 18.4344,
+      "step": 12145
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7152746938464145e-07,
+      "loss": 18.9353,
+      "step": 12146
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7018614802346574e-07,
+      "loss": 18.5743,
+      "step": 12147
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6884813889873626e-07,
+      "loss": 18.3982,
+      "step": 12148
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6751344209958173e-07,
+      "loss": 17.5864,
+      "step": 12149
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.661820577148921e-07,
+      "loss": 19.2802,
+      "step": 12150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.648539858333521e-07,
+      "loss": 0.9679,
+      "step": 12151
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.635292265434075e-07,
+      "loss": 20.2191,
+      "step": 12152
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.622077799332989e-07,
+      "loss": 20.0158,
+      "step": 12153
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6088964609103925e-07,
+      "loss": 0.8791,
+      "step": 12154
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.59574825104425e-07,
+      "loss": 19.8485,
+      "step": 12155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.582633170610249e-07,
+      "loss": 0.9451,
+      "step": 12156
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.569551220481914e-07,
+      "loss": 1.0088,
+      "step": 12157
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.556502401530658e-07,
+      "loss": 19.3042,
+      "step": 12158
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.543486714625509e-07,
+      "loss": 0.9508,
+      "step": 12159
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.530504160633329e-07,
+      "loss": 17.7243,
+      "step": 12160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.517554740418926e-07,
+      "loss": 0.1721,
+      "step": 12161
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.504638454844721e-07,
+      "loss": 19.5338,
+      "step": 12162
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4917553047710263e-07,
+      "loss": 19.0956,
+      "step": 12163
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4789052910558773e-07,
+      "loss": 0.2016,
+      "step": 12164
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.466088414555201e-07,
+      "loss": 0.9655,
+      "step": 12165
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4533046761226473e-07,
+      "loss": 0.9713,
+      "step": 12166
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.440554076609647e-07,
+      "loss": 0.9364,
+      "step": 12167
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.427836616865464e-07,
+      "loss": 0.9331,
+      "step": 12168
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.415152297737144e-07,
+      "loss": 18.6102,
+      "step": 12169
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4025011200695116e-07,
+      "loss": 16.9328,
+      "step": 12170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3898830847052246e-07,
+      "loss": 19.4952,
+      "step": 12171
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3772981924846672e-07,
+      "loss": 18.4648,
+      "step": 12172
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3647464442460577e-07,
+      "loss": 0.9887,
+      "step": 12173
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.352227840825394e-07,
+      "loss": 0.2204,
+      "step": 12174
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.339742383056509e-07,
+      "loss": 0.1873,
+      "step": 12175
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.32729007177096e-07,
+      "loss": 0.2096,
+      "step": 12176
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.314870907798139e-07,
+      "loss": 0.2149,
+      "step": 12177
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3024848919652177e-07,
+      "loss": 20.027,
+      "step": 12178
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.290132025097147e-07,
+      "loss": 0.8966,
+      "step": 12179
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2778123080167135e-07,
+      "loss": 0.8929,
+      "step": 12180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.265525741544483e-07,
+      "loss": 0.9102,
+      "step": 12181
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2532723264987455e-07,
+      "loss": 0.1765,
+      "step": 12182
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2410520636956255e-07,
+      "loss": 19.1307,
+      "step": 12183
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2288649539491392e-07,
+      "loss": 19.8436,
+      "step": 12184
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2167109980709144e-07,
+      "loss": 19.0311,
+      "step": 12185
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2045901968705264e-07,
+      "loss": 0.9384,
+      "step": 12186
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1925025511552177e-07,
+      "loss": 18.2537,
+      "step": 12187
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1804480617301225e-07,
+      "loss": 0.2315,
+      "step": 12188
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1684267293980988e-07,
+      "loss": 18.3478,
+      "step": 12189
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1564385549598387e-07,
+      "loss": 0.9729,
+      "step": 12190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1444835392138153e-07,
+      "loss": 20.6882,
+      "step": 12191
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.13256168295628e-07,
+      "loss": 0.1876,
+      "step": 12192
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1206729869813203e-07,
+      "loss": 0.9499,
+      "step": 12193
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1088174520806914e-07,
+      "loss": 17.6937,
+      "step": 12194
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0969950790440952e-07,
+      "loss": 17.4781,
+      "step": 12195
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.085205868658957e-07,
+      "loss": 18.214,
+      "step": 12196
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0734498217104824e-07,
+      "loss": 0.9647,
+      "step": 12197
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0617269389816563e-07,
+      "loss": 0.1808,
+      "step": 12198
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.050037221253298e-07,
+      "loss": 21.3184,
+      "step": 12199
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.038380669303952e-07,
+      "loss": 18.8736,
+      "step": 12200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.026757283910108e-07,
+      "loss": 0.2044,
+      "step": 12201
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0151670658458133e-07,
+      "loss": 19.6456,
+      "step": 12202
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0036100158831172e-07,
+      "loss": 0.9653,
+      "step": 12203
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9920861347917376e-07,
+      "loss": 0.9665,
+      "step": 12204
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9805954233392264e-07,
+      "loss": 18.451,
+      "step": 12205
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9691378822909167e-07,
+      "loss": 19.0338,
+      "step": 12206
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.95771351240992e-07,
+      "loss": 18.6098,
+      "step": 12207
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9463223144571274e-07,
+      "loss": 19.0433,
+      "step": 12208
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9349642891913765e-07,
+      "loss": 0.9139,
+      "step": 12209
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.923639437369007e-07,
+      "loss": 0.2044,
+      "step": 12210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9123477597443595e-07,
+      "loss": 18.3164,
+      "step": 12211
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.901089257069555e-07,
+      "loss": 0.1837,
+      "step": 12212
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8898639300943822e-07,
+      "loss": 20.4474,
+      "step": 12213
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8786717795665765e-07,
+      "loss": 19.8622,
+      "step": 12214
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8675128062315972e-07,
+      "loss": 18.0782,
+      "step": 12215
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8563870108326277e-07,
+      "loss": 19.7462,
+      "step": 12216
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8452943941106859e-07,
+      "loss": 0.9138,
+      "step": 12217
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8342349568046258e-07,
+      "loss": 0.9208,
+      "step": 12218
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.82320869965108e-07,
+      "loss": 18.2676,
+      "step": 12219
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.81221562338435e-07,
+      "loss": 0.9475,
+      "step": 12220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8012557287367392e-07,
+      "loss": 0.9343,
+      "step": 12221
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.790329016438219e-07,
+      "loss": 0.2107,
+      "step": 12222
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7794354872164854e-07,
+      "loss": 0.9451,
+      "step": 12223
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.768575141797124e-07,
+      "loss": 0.1542,
+      "step": 12224
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7577479809035013e-07,
+      "loss": 18.8424,
+      "step": 12225
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.746954005256818e-07,
+      "loss": 17.9897,
+      "step": 12226
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.736193215575832e-07,
+      "loss": 0.8861,
+      "step": 12227
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7254656125774703e-07,
+      "loss": 0.9601,
+      "step": 12228
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7147711969760506e-07,
+      "loss": 18.6433,
+      "step": 12229
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7041099694840023e-07,
+      "loss": 0.9406,
+      "step": 12230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6934819308113692e-07,
+      "loss": 18.5538,
+      "step": 12231
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6828870816660292e-07,
+      "loss": 19.2714,
+      "step": 12232
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.67232542275364e-07,
+      "loss": 0.1998,
+      "step": 12233
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.661796954777639e-07,
+      "loss": 18.7476,
+      "step": 12234
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6513016784392986e-07,
+      "loss": 0.1547,
+      "step": 12235
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6408395944376708e-07,
+      "loss": 0.9085,
+      "step": 12236
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6304107034695314e-07,
+      "loss": 0.2257,
+      "step": 12237
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.620015006229547e-07,
+      "loss": 0.9205,
+      "step": 12238
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6096525034101085e-07,
+      "loss": 0.9054,
+      "step": 12239
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.59932319570133e-07,
+      "loss": 0.2214,
+      "step": 12240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.589027083791328e-07,
+      "loss": 18.4052,
+      "step": 12241
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5787641683657207e-07,
+      "loss": 18.538,
+      "step": 12242
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5685344501081834e-07,
+      "loss": 0.2289,
+      "step": 12243
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5583379297000046e-07,
+      "loss": 0.179,
+      "step": 12244
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.548174607820363e-07,
+      "loss": 0.9262,
+      "step": 12245
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5380444851461063e-07,
+      "loss": 18.0669,
+      "step": 12246
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5279475623520279e-07,
+      "loss": 21.0995,
+      "step": 12247
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.51788384011059e-07,
+      "loss": 0.9401,
+      "step": 12248
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5078533190920897e-07,
+      "loss": 18.1507,
+      "step": 12249
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4978559999646035e-07,
+      "loss": 0.9328,
+      "step": 12250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.487891883393988e-07,
+      "loss": 18.9614,
+      "step": 12251
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4779609700439344e-07,
+      "loss": 19.1589,
+      "step": 12252
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4680632605758582e-07,
+      "loss": 19.995,
+      "step": 12253
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4581987556490095e-07,
+      "loss": 17.481,
+      "step": 12254
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4483674559204185e-07,
+      "loss": 19.1745,
+      "step": 12255
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4385693620448394e-07,
+      "loss": 0.2354,
+      "step": 12256
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4288044746749162e-07,
+      "loss": 19.9102,
+      "step": 12257
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4190727944610738e-07,
+      "loss": 17.65,
+      "step": 12258
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.40937432205146e-07,
+      "loss": 19.3736,
+      "step": 12259
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.399709058091947e-07,
+      "loss": 0.9394,
+      "step": 12260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3900770032264088e-07,
+      "loss": 0.953,
+      "step": 12261
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3804781580963322e-07,
+      "loss": 16.6896,
+      "step": 12262
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3709125233410392e-07,
+      "loss": 0.2053,
+      "step": 12263
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.361380099597631e-07,
+      "loss": 0.1587,
+      "step": 12264
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3518808875011002e-07,
+      "loss": 0.1836,
+      "step": 12265
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3424148876840514e-07,
+      "loss": 19.4308,
+      "step": 12266
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3329821007769804e-07,
+      "loss": 17.6302,
+      "step": 12267
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3235825274081625e-07,
+      "loss": 19.7648,
+      "step": 12268
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3142161682036523e-07,
+      "loss": 0.9539,
+      "step": 12269
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3048830237872845e-07,
+      "loss": 18.0148,
+      "step": 12270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.295583094780728e-07,
+      "loss": 0.9682,
+      "step": 12271
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.286316381803321e-07,
+      "loss": 20.5693,
+      "step": 12272
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2770828854724025e-07,
+      "loss": 0.9464,
+      "step": 12273
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.267882606402815e-07,
+      "loss": 0.905,
+      "step": 12274
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2587155452074005e-07,
+      "loss": 19.0279,
+      "step": 12275
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2495817024967827e-07,
+      "loss": 0.1769,
+      "step": 12276
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.240481078879252e-07,
+      "loss": 18.1469,
+      "step": 12277
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2314136749609906e-07,
+      "loss": 0.9274,
+      "step": 12278
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2223794913458487e-07,
+      "loss": 0.94,
+      "step": 12279
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2133785286356781e-07,
+      "loss": 0.8963,
+      "step": 12280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2044107874298882e-07,
+      "loss": 0.903,
+      "step": 12281
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1954762683257793e-07,
+      "loss": 18.7429,
+      "step": 12282
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.186574971918486e-07,
+      "loss": 0.1904,
+      "step": 12283
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1777068988007566e-07,
+      "loss": 19.0146,
+      "step": 12284
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.168872049563452e-07,
+      "loss": 0.2426,
+      "step": 12285
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1600704247948236e-07,
+      "loss": 0.1735,
+      "step": 12286
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1513020250811801e-07,
+      "loss": 18.1406,
+      "step": 12287
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.142566851006499e-07,
+      "loss": 0.9285,
+      "step": 12288
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1338649031526483e-07,
+      "loss": 0.2133,
+      "step": 12289
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1251961820991641e-07,
+      "loss": 19.023,
+      "step": 12290
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1165606884234181e-07,
+      "loss": 20.9718,
+      "step": 12291
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1079584227006168e-07,
+      "loss": 18.177,
+      "step": 12292
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0993893855036907e-07,
+      "loss": 19.0343,
+      "step": 12293
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0908535774034057e-07,
+      "loss": 0.9249,
+      "step": 12294
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0823509989681956e-07,
+      "loss": 17.6918,
+      "step": 12295
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0738816507644967e-07,
+      "loss": 19.3032,
+      "step": 12296
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0654455333563019e-07,
+      "loss": 17.3773,
+      "step": 12297
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0570426473055506e-07,
+      "loss": 18.6382,
+      "step": 12298
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0486729931718509e-07,
+      "loss": 0.9445,
+      "step": 12299
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0403365715127567e-07,
+      "loss": 0.9474,
+      "step": 12300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0320333828834904e-07,
+      "loss": 20.5288,
+      "step": 12301
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0237634278370544e-07,
+      "loss": 17.4509,
+      "step": 12302
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0155267069242302e-07,
+      "loss": 0.2014,
+      "step": 12303
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.00732322069369e-07,
+      "loss": 17.7515,
+      "step": 12304
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.991529696917745e-08,
+      "loss": 0.2138,
+      "step": 12305
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.910159544627151e-08,
+      "loss": 0.201,
+      "step": 12306
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.829121755484116e-08,
+      "loss": 20.3158,
+      "step": 12307
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.748416334886546e-08,
+      "loss": 0.1858,
+      "step": 12308
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.668043288209583e-08,
+      "loss": 18.7542,
+      "step": 12309
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.588002620806725e-08,
+      "loss": 19.0266,
+      "step": 12310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.508294338008705e-08,
+      "loss": 17.2283,
+      "step": 12311
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.428918445124613e-08,
+      "loss": 18.7917,
+      "step": 12312
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.349874947441328e-08,
+      "loss": 0.9303,
+      "step": 12313
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.27116385022353e-08,
+      "loss": 19.0927,
+      "step": 12314
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.19278515871369e-08,
+      "loss": 0.1918,
+      "step": 12315
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.114738878132079e-08,
+      "loss": 0.942,
+      "step": 12316
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.037025013676759e-08,
+      "loss": 18.332,
+      "step": 12317
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.959643570524701e-08,
+      "loss": 0.9508,
+      "step": 12318
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.882594553829005e-08,
+      "loss": 0.9438,
+      "step": 12319
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.805877968721677e-08,
+      "loss": 18.6599,
+      "step": 12320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.729493820312517e-08,
+      "loss": 0.9777,
+      "step": 12321
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.653442113689125e-08,
+      "loss": 19.71,
+      "step": 12322
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.577722853916337e-08,
+      "loss": 0.1605,
+      "step": 12323
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.502336046038451e-08,
+      "loss": 0.9188,
+      "step": 12324
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.427281695075894e-08,
+      "loss": 18.9699,
+      "step": 12325
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.35255980602745e-08,
+      "loss": 19.2479,
+      "step": 12326
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.278170383870798e-08,
+      "loss": 0.9213,
+      "step": 12327
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.204113433559201e-08,
+      "loss": 0.8814,
+      "step": 12328
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.130388960027047e-08,
+      "loss": 0.2072,
+      "step": 12329
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.056996968183183e-08,
+      "loss": 0.2065,
+      "step": 12330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.983937462915925e-08,
+      "loss": 19.8912,
+      "step": 12331
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.911210449092487e-08,
+      "loss": 17.7202,
+      "step": 12332
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.83881593155622e-08,
+      "loss": 0.8918,
+      "step": 12333
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.766753915128821e-08,
+      "loss": 0.1818,
+      "step": 12334
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.695024404609786e-08,
+      "loss": 0.9324,
+      "step": 12335
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.623627404777512e-08,
+      "loss": 19.9988,
+      "step": 12336
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.552562920385975e-08,
+      "loss": 19.1315,
+      "step": 12337
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.481830956169722e-08,
+      "loss": 20.7066,
+      "step": 12338
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.411431516838874e-08,
+      "loss": 17.7585,
+      "step": 12339
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.341364607083012e-08,
+      "loss": 18.7508,
+      "step": 12340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.27163023156896e-08,
+      "loss": 0.9469,
+      "step": 12341
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.20222839494078e-08,
+      "loss": 19.4582,
+      "step": 12342
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.133159101820886e-08,
+      "loss": 20.7705,
+      "step": 12343
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.064422356810596e-08,
+      "loss": 0.9616,
+      "step": 12344
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.996018164487361e-08,
+      "loss": 0.9233,
+      "step": 12345
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.927946529406982e-08,
+      "loss": 0.893,
+      "step": 12346
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.860207456103607e-08,
+      "loss": 19.4891,
+      "step": 12347
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.792800949089739e-08,
+      "loss": 0.9613,
+      "step": 12348
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.725727012853455e-08,
+      "loss": 20.6125,
+      "step": 12349
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.658985651863403e-08,
+      "loss": 18.128,
+      "step": 12350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.59257687056436e-08,
+      "loss": 0.9206,
+      "step": 12351
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.526500673380009e-08,
+      "loss": 0.1865,
+      "step": 12352
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.460757064710721e-08,
+      "loss": 19.9064,
+      "step": 12353
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.395346048935214e-08,
+      "loss": 0.9574,
+      "step": 12354
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.330267630410558e-08,
+      "loss": 0.2057,
+      "step": 12355
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.265521813471065e-08,
+      "loss": 0.9652,
+      "step": 12356
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.201108602429396e-08,
+      "loss": 20.5439,
+      "step": 12357
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.137028001576006e-08,
+      "loss": 19.6035,
+      "step": 12358
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.07328001517804e-08,
+      "loss": 19.4915,
+      "step": 12359
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.0098646474821e-08,
+      "loss": 0.1821,
+      "step": 12360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.9467819027114734e-08,
+      "loss": 18.8525,
+      "step": 12361
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8840317850683555e-08,
+      "loss": 0.211,
+      "step": 12362
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8216142987316257e-08,
+      "loss": 19.8741,
+      "step": 12363
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.759529447859069e-08,
+      "loss": 18.8849,
+      "step": 12364
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.697777236585711e-08,
+      "loss": 0.9136,
+      "step": 12365
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.6363576690243726e-08,
+      "loss": 0.2332,
+      "step": 12366
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.575270749265671e-08,
+      "loss": 21.0819,
+      "step": 12367
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.5145164813785734e-08,
+      "loss": 0.9561,
+      "step": 12368
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.4540948694098424e-08,
+      "loss": 0.9184,
+      "step": 12369
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.3940059173834826e-08,
+      "loss": 18.4718,
+      "step": 12370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.334249629301291e-08,
+      "loss": 0.944,
+      "step": 12371
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.274826009144529e-08,
+      "loss": 0.1947,
+      "step": 12372
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.21573506087003e-08,
+      "loss": 0.9279,
+      "step": 12373
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1569767884135366e-08,
+      "loss": 17.8145,
+      "step": 12374
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0985511956891384e-08,
+      "loss": 0.2012,
+      "step": 12375
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.040458286588723e-08,
+      "loss": 20.003,
+      "step": 12376
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9826980649803066e-08,
+      "loss": 0.9524,
+      "step": 12377
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9252705347113685e-08,
+      "loss": 0.9741,
+      "step": 12378
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.86817569960718e-08,
+      "loss": 0.2212,
+      "step": 12379
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8114135634702576e-08,
+      "loss": 19.9143,
+      "step": 12380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.754984130081464e-08,
+      "loss": 0.9813,
+      "step": 12381
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.698887403199459e-08,
+      "loss": 0.9325,
+      "step": 12382
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.643123386559589e-08,
+      "loss": 0.9454,
+      "step": 12383
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5876920838772155e-08,
+      "loss": 20.2727,
+      "step": 12384
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.532593498843274e-08,
+      "loss": 19.3189,
+      "step": 12385
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4778276351287176e-08,
+      "loss": 0.185,
+      "step": 12386
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4233944963800733e-08,
+      "loss": 18.8131,
+      "step": 12387
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.36929408622333e-08,
+      "loss": 18.2431,
+      "step": 12388
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.315526408261716e-08,
+      "loss": 19.7817,
+      "step": 12389
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.262091466076812e-08,
+      "loss": 0.9634,
+      "step": 12390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.208989263226881e-08,
+      "loss": 18.6794,
+      "step": 12391
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.15621980324965e-08,
+      "loss": 18.9223,
+      "step": 12392
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.103783089658975e-08,
+      "loss": 0.2092,
+      "step": 12393
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.051679125947616e-08,
+      "loss": 0.2087,
+      "step": 12394
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.999907915586687e-08,
+      "loss": 20.7556,
+      "step": 12395
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.94846946202343e-08,
+      "loss": 19.189,
+      "step": 12396
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.897363768684548e-08,
+      "loss": 0.9387,
+      "step": 12397
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.84659083897343e-08,
+      "loss": 19.3167,
+      "step": 12398
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7961506762723696e-08,
+      "loss": 0.9933,
+      "step": 12399
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.746043283939793e-08,
+      "loss": 20.4781,
+      "step": 12400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.696268665314695e-08,
+      "loss": 17.2185,
+      "step": 12401
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6468268237105366e-08,
+      "loss": 18.2884,
+      "step": 12402
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5977177624219037e-08,
+      "loss": 0.23,
+      "step": 12403
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5489414847189596e-08,
+      "loss": 0.879,
+      "step": 12404
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.500497993850216e-08,
+      "loss": 18.82,
+      "step": 12405
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.452387293043091e-08,
+      "loss": 0.9485,
+      "step": 12406
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.404609385501134e-08,
+      "loss": 19.7509,
+      "step": 12407
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.357164274406799e-08,
+      "loss": 1.0092,
+      "step": 12408
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.310051962920335e-08,
+      "loss": 20.1586,
+      "step": 12409
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.263272454179234e-08,
+      "loss": 17.1658,
+      "step": 12410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.21682575129989e-08,
+      "loss": 17.3968,
+      "step": 12411
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.170711857375386e-08,
+      "loss": 0.2359,
+      "step": 12412
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.124930775477153e-08,
+      "loss": 0.939,
+      "step": 12413
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.079482508654419e-08,
+      "loss": 0.9966,
+      "step": 12414
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.034367059934762e-08,
+      "loss": 0.2023,
+      "step": 12415
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.989584432322445e-08,
+      "loss": 0.9201,
+      "step": 12416
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9451346288000835e-08,
+      "loss": 18.0474,
+      "step": 12417
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9010176523286413e-08,
+      "loss": 0.1593,
+      "step": 12418
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8572335058468792e-08,
+      "loss": 20.9128,
+      "step": 12419
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8137821922702423e-08,
+      "loss": 18.9699,
+      "step": 12420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7706637144936375e-08,
+      "loss": 20.7078,
+      "step": 12421
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7278780753881017e-08,
+      "loss": 0.9284,
+      "step": 12422
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6854252778035772e-08,
+      "loss": 19.7243,
+      "step": 12423
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6433053245678018e-08,
+      "loss": 18.9581,
+      "step": 12424
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6015182184863095e-08,
+      "loss": 19.6532,
+      "step": 12425
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5600639623418743e-08,
+      "loss": 18.4066,
+      "step": 12426
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5189425588956207e-08,
+      "loss": 0.9115,
+      "step": 12427
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.478154010887024e-08,
+      "loss": 21.0064,
+      "step": 12428
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4376983210322445e-08,
+      "loss": 0.9593,
+      "step": 12429
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3975754920257943e-08,
+      "loss": 19.4288,
+      "step": 12430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3577855265399796e-08,
+      "loss": 19.3425,
+      "step": 12431
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.318328427225458e-08,
+      "loss": 19.8893,
+      "step": 12432
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2792041967101273e-08,
+      "loss": 0.9203,
+      "step": 12433
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2404128375996815e-08,
+      "loss": 20.2547,
+      "step": 12434
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2019543524776087e-08,
+      "loss": 0.1914,
+      "step": 12435
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.163828743905749e-08,
+      "loss": 18.5031,
+      "step": 12436
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.126036014423183e-08,
+      "loss": 18.1285,
+      "step": 12437
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0885761665473403e-08,
+      "loss": 21.4049,
+      "step": 12438
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0514492027728926e-08,
+      "loss": 0.9171,
+      "step": 12439
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.014655125573417e-08,
+      "loss": 18.5698,
+      "step": 12440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9781939373986204e-08,
+      "loss": 0.9435,
+      "step": 12441
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9420656406776706e-08,
+      "loss": 18.3401,
+      "step": 12442
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.906270237816421e-08,
+      "loss": 18.5823,
+      "step": 12443
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8708077311990758e-08,
+      "loss": 17.9637,
+      "step": 12444
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8356781231881893e-08,
+      "loss": 19.4662,
+      "step": 12445
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8008814161224463e-08,
+      "loss": 0.1836,
+      "step": 12446
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7664176123205477e-08,
+      "loss": 0.9156,
+      "step": 12447
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7322867140778798e-08,
+      "loss": 0.9473,
+      "step": 12448
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6984887236670687e-08,
+      "loss": 0.1851,
+      "step": 12449
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.665023643339092e-08,
+      "loss": 19.1663,
+      "step": 12450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6318914753238323e-08,
+      "loss": 0.8812,
+      "step": 12451
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.599092221827303e-08,
+      "loss": 0.9852,
+      "step": 12452
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566625885034423e-08,
+      "loss": 0.9579,
+      "step": 12453
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5344924671073514e-08,
+      "loss": 20.0808,
+      "step": 12454
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.502691970186043e-08,
+      "loss": 0.9587,
+      "step": 12455
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.471224396389359e-08,
+      "loss": 0.9542,
+      "step": 12456
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.440089747812845e-08,
+      "loss": 0.2223,
+      "step": 12457
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4092880265303976e-08,
+      "loss": 19.4056,
+      "step": 12458
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3788192345925988e-08,
+      "loss": 18.2207,
+      "step": 12459
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3486833740300465e-08,
+      "loss": 0.9469,
+      "step": 12460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3188804468494686e-08,
+      "loss": 0.9141,
+      "step": 12461
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2894104550353892e-08,
+      "loss": 18.3684,
+      "step": 12462
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2602734005517925e-08,
+      "loss": 0.987,
+      "step": 12463
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.231469285337683e-08,
+      "loss": 19.4891,
+      "step": 12464
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2029981113131916e-08,
+      "loss": 0.9541,
+      "step": 12465
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.174859880373469e-08,
+      "loss": 0.2057,
+      "step": 12466
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1470545943936816e-08,
+      "loss": 0.9855,
+      "step": 12467
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1195822552245716e-08,
+      "loss": 0.9283,
+      "step": 12468
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0924428646968965e-08,
+      "loss": 18.984,
+      "step": 12469
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0656364246180994e-08,
+      "loss": 0.1989,
+      "step": 12470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.039162936773419e-08,
+      "loss": 15.2233,
+      "step": 12471
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0130224029258894e-08,
+      "loss": 0.9029,
+      "step": 12472
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.872148248168955e-09,
+      "loss": 19.3657,
+      "step": 12473
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.617402041656176e-09,
+      "loss": 18.6844,
+      "step": 12474
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.365985426679213e-09,
+      "loss": 18.3664,
+      "step": 12475
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.117898419991333e-09,
+      "loss": 0.1588,
+      "step": 12476
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.873141038118205e-09,
+      "loss": 18.951,
+      "step": 12477
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.63171329735235e-09,
+      "loss": 0.2119,
+      "step": 12478
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.393615213780904e-09,
+      "loss": 0.2129,
+      "step": 12479
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.158846803257847e-09,
+      "loss": 17.8167,
+      "step": 12480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.927408081426225e-09,
+      "loss": 18.489,
+      "step": 12481
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.69929906369593e-09,
+      "loss": 19.7722,
+      "step": 12482
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.474519765260368e-09,
+      "loss": 20.8311,
+      "step": 12483
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.253070201096446e-09,
+      "loss": 19.0006,
+      "step": 12484
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.034950385947925e-09,
+      "loss": 17.9677,
+      "step": 12485
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.820160334342074e-09,
+      "loss": 18.5727,
+      "step": 12486
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.608700060584117e-09,
+      "loss": 21.2286,
+      "step": 12487
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.400569578762783e-09,
+      "loss": 0.2021,
+      "step": 12488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.1957689027392074e-09,
+      "loss": 18.4794,
+      "step": 12489
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.99429804615248e-09,
+      "loss": 0.9122,
+      "step": 12490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.796157022425197e-09,
+      "loss": 0.9172,
+      "step": 12491
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.601345844752359e-09,
+      "loss": 20.0993,
+      "step": 12492
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.409864526106923e-09,
+      "loss": 0.9065,
+      "step": 12493
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.221713079245349e-09,
+      "loss": 17.8322,
+      "step": 12494
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.036891516696507e-09,
+      "loss": 18.1972,
+      "step": 12495
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.855399850772768e-09,
+      "loss": 18.7242,
+      "step": 12496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.677238093558911e-09,
+      "loss": 18.6521,
+      "step": 12497
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.502406256928771e-09,
+      "loss": 0.9561,
+      "step": 12498
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.330904352517484e-09,
+      "loss": 0.9344,
+      "step": 12499
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1627323917603486e-09,
+      "loss": 20.5933,
+      "step": 12500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.997890385842862e-09,
+      "loss": 0.1357,
+      "step": 12501
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.836378345756231e-09,
+      "loss": 18.9479,
+      "step": 12502
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6781962822529657e-09,
+      "loss": 0.2088,
+      "step": 12503
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5233442058690835e-09,
+      "loss": 17.9383,
+      "step": 12504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3718221269185556e-09,
+      "loss": 0.885,
+      "step": 12505
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.223630055498861e-09,
+      "loss": 0.9322,
+      "step": 12506
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.078768001468779e-09,
+      "loss": 0.8894,
+      "step": 12507
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9372359744872515e-09,
+      "loss": 0.1737,
+      "step": 12508
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7990339839745192e-09,
+      "loss": 0.9219,
+      "step": 12509
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6641620391398836e-09,
+      "loss": 0.9363,
+      "step": 12510
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.532620148965048e-09,
+      "loss": 0.901,
+      "step": 12511
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4044083222096723e-09,
+      "loss": 0.1888,
+      "step": 12512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.279526567411372e-09,
+      "loss": 0.9445,
+      "step": 12513
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.157974892896819e-09,
+      "loss": 18.6559,
+      "step": 12514
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0397533067484376e-09,
+      "loss": 19.2467,
+      "step": 12515
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.924861816854362e-09,
+      "loss": 0.9565,
+      "step": 12516
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8133004308584778e-09,
+      "loss": 16.5922,
+      "step": 12517
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7050691561937282e-09,
+      "loss": 18.0536,
+      "step": 12518
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6001680000654606e-09,
+      "loss": 0.9464,
+      "step": 12519
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4985969694680802e-09,
+      "loss": 0.2446,
+      "step": 12520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.400356071157294e-09,
+      "loss": 18.4797,
+      "step": 12521
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3054453116834175e-09,
+      "loss": 20.2328,
+      "step": 12522
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2138646973691713e-09,
+      "loss": 0.9265,
+      "step": 12523
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1256142343041287e-09,
+      "loss": 19.6139,
+      "step": 12524
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0406939283780227e-09,
+      "loss": 18.2732,
+      "step": 12525
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.591037852418882e-10,
+      "loss": 18.2356,
+      "step": 12526
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.808438103247163e-10,
+      "loss": 0.93,
+      "step": 12527
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.059140088501061e-10,
+      "loss": 18.1437,
+      "step": 12528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.343143857974077e-10,
+      "loss": 0.9661,
+      "step": 12529
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.660449459461316e-10,
+      "loss": 18.5354,
+      "step": 12530
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.011056938315385e-10,
+      "loss": 19.1118,
+      "step": 12531
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.394966337890495e-10,
+      "loss": 18.6248,
+      "step": 12532
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.812177699209386e-10,
+      "loss": 0.9509,
+      "step": 12533
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.262691061018842e-10,
+      "loss": 18.1124,
+      "step": 12534
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7465064599562227e-10,
+      "loss": 0.9294,
+      "step": 12535
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.263623930438442e-10,
+      "loss": 21.2582,
+      "step": 12536
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8140435046064563e-10,
+      "loss": 21.3162,
+      "step": 12537
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.397765212380776e-10,
+      "loss": 0.9482,
+      "step": 12538
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.014789081516977e-10,
+      "loss": 0.9547,
+      "step": 12539
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.665115137494677e-10,
+      "loss": 19.5223,
+      "step": 12540
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3487434035730495e-10,
+      "loss": 0.1795,
+      "step": 12541
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0656739009018424e-10,
+      "loss": 18.6314,
+      "step": 12542
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.159066482993361e-11,
+      "loss": 0.9325,
+      "step": 12543
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.99441662418876e-11,
+      "loss": 0.184,
+      "step": 12544
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.162789576378501e-11,
+      "loss": 19.859,
+      "step": 12545
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6641854622422303e-11,
+      "loss": 0.1815,
+      "step": 12546
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4986043805897965e-11,
+      "loss": 20.474,
+      "step": 12547
+    },
+    {
+      "epoch": 1.0,
+      "step": 12547,
+      "total_flos": 3.20651325322268e+20,
+      "train_loss": 120.75352399966613,
+      "train_runtime": 47534.6503,
+      "train_samples_per_second": 101.376,
+      "train_steps_per_second": 0.264
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 12550,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 2000,
+  "total_flos": 3.20651325322268e+20,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}