diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,67471 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 11241,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.917159763313609e-06,
+      "loss": 9.9531,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1834319526627219e-05,
+      "loss": 9.7188,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7751479289940828e-05,
+      "loss": 9.5938,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3668639053254438e-05,
+      "loss": 9.7344,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9585798816568047e-05,
+      "loss": 9.4062,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5502958579881656e-05,
+      "loss": 9.2812,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.142011834319527e-05,
+      "loss": 8.8125,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7337278106508875e-05,
+      "loss": 8.625,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.325443786982249e-05,
+      "loss": 8.2812,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9171597633136094e-05,
+      "loss": 7.9297,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.50887573964497e-05,
+      "loss": 7.8438,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.100591715976331e-05,
+      "loss": 7.4688,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.692307692307693e-05,
+      "loss": 7.3359,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.284023668639054e-05,
+      "loss": 6.8828,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.875739644970414e-05,
+      "loss": 6.7031,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.467455621301775e-05,
+      "loss": 6.375,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010059171597633136,
+      "loss": 6.25,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010650887573964498,
+      "loss": 6.125,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011242603550295859,
+      "loss": 5.9453,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011834319526627219,
+      "loss": 6.0078,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001242603550295858,
+      "loss": 5.6328,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001301775147928994,
+      "loss": 5.4922,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000136094674556213,
+      "loss": 5.1797,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014201183431952663,
+      "loss": 5.2969,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014792899408284024,
+      "loss": 5.0078,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015384615384615385,
+      "loss": 4.9219,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015976331360946746,
+      "loss": 4.6797,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016568047337278108,
+      "loss": 4.6094,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017159763313609466,
+      "loss": 4.5312,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017751479289940828,
+      "loss": 4.2422,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001834319526627219,
+      "loss": 4.1875,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001893491124260355,
+      "loss": 3.8438,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019526627218934911,
+      "loss": 3.8789,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00020118343195266273,
+      "loss": 3.5508,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00020710059171597634,
+      "loss": 3.5117,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00021301775147928995,
+      "loss": 3.3242,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00021893491124260357,
+      "loss": 3.4062,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00022485207100591718,
+      "loss": 3.0391,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002307692307692308,
+      "loss": 3.0469,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00023668639053254438,
+      "loss": 2.8516,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000242603550295858,
+      "loss": 2.7031,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002485207100591716,
+      "loss": 2.8438,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00025443786982248524,
+      "loss": 2.7812,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002603550295857988,
+      "loss": 2.8828,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00026627218934911247,
+      "loss": 2.6211,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000272189349112426,
+      "loss": 2.6367,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00027810650887573964,
+      "loss": 2.543,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00028402366863905325,
+      "loss": 2.5742,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00028994082840236686,
+      "loss": 2.4688,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002958579881656805,
+      "loss": 2.5273,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0003017751479289941,
+      "loss": 2.4531,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0003076923076923077,
+      "loss": 2.6367,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0003136094674556213,
+      "loss": 2.5664,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00031952662721893493,
+      "loss": 2.3535,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00032544378698224854,
+      "loss": 2.5273,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00033136094674556215,
+      "loss": 2.5195,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00033727810650887577,
+      "loss": 2.4609,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003431952662721893,
+      "loss": 2.3906,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000349112426035503,
+      "loss": 2.3672,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00035502958579881655,
+      "loss": 2.3906,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003609467455621302,
+      "loss": 2.3867,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003668639053254438,
+      "loss": 2.2695,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00037278106508875744,
+      "loss": 2.2773,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000378698224852071,
+      "loss": 2.2227,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00038461538461538467,
+      "loss": 2.3008,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00039053254437869823,
+      "loss": 2.1484,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003964497041420119,
+      "loss": 2.3398,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00040236686390532545,
+      "loss": 2.2266,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000408284023668639,
+      "loss": 2.1328,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004142011834319527,
+      "loss": 2.2031,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00042011834319526624,
+      "loss": 2.2344,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004260355029585799,
+      "loss": 2.0977,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00043195266272189346,
+      "loss": 2.2188,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00043786982248520713,
+      "loss": 2.1094,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004437869822485207,
+      "loss": 2.0645,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00044970414201183436,
+      "loss": 2.1992,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004556213017751479,
+      "loss": 2.2539,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004615384615384616,
+      "loss": 2.0488,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00046745562130177514,
+      "loss": 2.0801,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00047337278106508875,
+      "loss": 2.2578,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00047928994082840237,
+      "loss": 2.2266,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000485207100591716,
+      "loss": 2.0859,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004911242603550296,
+      "loss": 1.9297,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004970414201183431,
+      "loss": 2.0586,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005029585798816569,
+      "loss": 2.0098,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005088757396449705,
+      "loss": 2.0508,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000514792899408284,
+      "loss": 2.1016,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005207100591715976,
+      "loss": 1.9707,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005266272189349113,
+      "loss": 2.0078,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005325443786982249,
+      "loss": 2.0801,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005384615384615384,
+      "loss": 1.9902,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000544378698224852,
+      "loss": 2.0371,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005502958579881658,
+      "loss": 1.9785,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005562130177514793,
+      "loss": 2.0664,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005621301775147929,
+      "loss": 1.9941,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005680473372781065,
+      "loss": 2.0625,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005739644970414202,
+      "loss": 2.1504,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005798816568047337,
+      "loss": 1.9473,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005857988165680473,
+      "loss": 2.0,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000591715976331361,
+      "loss": 2.0781,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005976331360946747,
+      "loss": 1.8945,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006035502958579882,
+      "loss": 1.9219,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006094674556213018,
+      "loss": 2.0371,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006153846153846154,
+      "loss": 1.9512,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006213017751479289,
+      "loss": 1.9844,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006272189349112426,
+      "loss": 1.9258,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006331360946745562,
+      "loss": 1.9277,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006390532544378699,
+      "loss": 1.8848,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006449704142011834,
+      "loss": 1.9395,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006508875739644971,
+      "loss": 1.8809,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006568047337278107,
+      "loss": 2.0469,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006627218934911243,
+      "loss": 2.0137,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006686390532544378,
+      "loss": 1.7812,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006745562130177515,
+      "loss": 1.9668,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006804733727810651,
+      "loss": 1.8066,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006863905325443787,
+      "loss": 1.8887,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006923076923076923,
+      "loss": 1.8926,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000698224852071006,
+      "loss": 1.9531,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007041420118343196,
+      "loss": 2.0039,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007100591715976331,
+      "loss": 2.0,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007159763313609467,
+      "loss": 2.0391,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007218934911242604,
+      "loss": 1.875,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000727810650887574,
+      "loss": 1.9707,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007337278106508876,
+      "loss": 1.7715,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007396449704142012,
+      "loss": 1.9121,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007455621301775149,
+      "loss": 1.9688,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007514792899408284,
+      "loss": 2.0977,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000757396449704142,
+      "loss": 2.002,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007633136094674556,
+      "loss": 1.9375,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007692307692307693,
+      "loss": 1.8633,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007751479289940828,
+      "loss": 1.9336,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007810650887573965,
+      "loss": 1.8184,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007869822485207101,
+      "loss": 1.8672,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007928994082840238,
+      "loss": 1.877,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007988165680473373,
+      "loss": 1.8281,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008047337278106509,
+      "loss": 1.9043,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008106508875739645,
+      "loss": 1.9297,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000816568047337278,
+      "loss": 1.832,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008224852071005917,
+      "loss": 1.9395,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008284023668639054,
+      "loss": 1.9434,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000834319526627219,
+      "loss": 1.9043,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008402366863905325,
+      "loss": 1.8027,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008461538461538462,
+      "loss": 1.7734,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008520710059171598,
+      "loss": 1.9688,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008579881656804734,
+      "loss": 1.957,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008639053254437869,
+      "loss": 1.8926,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008698224852071006,
+      "loss": 2.0215,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008757396449704143,
+      "loss": 1.877,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008816568047337278,
+      "loss": 1.832,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008875739644970414,
+      "loss": 1.8672,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008934911242603551,
+      "loss": 1.9238,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008994082840236687,
+      "loss": 1.8848,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009053254437869822,
+      "loss": 1.8848,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009112426035502958,
+      "loss": 1.9551,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009171597633136096,
+      "loss": 1.8457,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009230769230769232,
+      "loss": 1.8652,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009289940828402367,
+      "loss": 1.8223,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009349112426035503,
+      "loss": 1.7852,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000940828402366864,
+      "loss": 1.8828,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009467455621301775,
+      "loss": 1.8301,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009526627218934911,
+      "loss": 1.959,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009585798816568047,
+      "loss": 1.8145,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009644970414201185,
+      "loss": 1.8301,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000970414201183432,
+      "loss": 1.8613,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009763313609467456,
+      "loss": 1.8008,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009822485207100593,
+      "loss": 1.8379,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009881656804733728,
+      "loss": 1.8008,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009940828402366863,
+      "loss": 1.8438,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001,
+      "loss": 1.9199,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0010059171597633137,
+      "loss": 1.7637,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0010118343195266272,
+      "loss": 1.9297,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001017751479289941,
+      "loss": 2.0293,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0010236686390532543,
+      "loss": 1.8398,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001029585798816568,
+      "loss": 1.8438,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0010355029585798817,
+      "loss": 1.7617,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0010414201183431952,
+      "loss": 1.8945,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001047337278106509,
+      "loss": 1.9238,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0010532544378698226,
+      "loss": 1.7637,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0010591715976331361,
+      "loss": 1.9199,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0010650887573964499,
+      "loss": 1.7949,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0010710059171597632,
+      "loss": 1.7852,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0010769230769230769,
+      "loss": 1.8203,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0010828402366863906,
+      "loss": 1.8027,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001088757396449704,
+      "loss": 1.8203,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0010946745562130178,
+      "loss": 1.9082,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0011005917159763315,
+      "loss": 1.9902,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001106508875739645,
+      "loss": 1.8945,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0011124260355029586,
+      "loss": 1.7168,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001118343195266272,
+      "loss": 1.7695,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0011242603550295858,
+      "loss": 1.7793,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0011301775147928995,
+      "loss": 1.9023,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001136094674556213,
+      "loss": 1.793,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0011420118343195267,
+      "loss": 1.8457,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0011479289940828404,
+      "loss": 1.8984,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0011538461538461537,
+      "loss": 1.8301,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0011597633136094675,
+      "loss": 1.8223,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001165680473372781,
+      "loss": 1.8711,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0011715976331360947,
+      "loss": 1.8105,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0011775147928994084,
+      "loss": 1.9648,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001183431952662722,
+      "loss": 1.7695,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0011893491124260356,
+      "loss": 1.8809,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0011952662721893494,
+      "loss": 1.7676,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0012011834319526626,
+      "loss": 1.959,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0012071005917159764,
+      "loss": 1.7734,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0012130177514792899,
+      "loss": 1.7676,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0012189349112426036,
+      "loss": 1.8047,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0012248520710059173,
+      "loss": 1.7715,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0012307692307692308,
+      "loss": 1.875,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0012366863905325445,
+      "loss": 1.8008,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0012426035502958578,
+      "loss": 1.8281,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0012485207100591715,
+      "loss": 1.8613,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0012544378698224853,
+      "loss": 1.8496,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0012603550295857988,
+      "loss": 1.8027,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0012662721893491125,
+      "loss": 1.8828,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0012721893491124262,
+      "loss": 1.9023,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0012781065088757397,
+      "loss": 1.8711,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0012840236686390532,
+      "loss": 1.8984,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0012899408284023667,
+      "loss": 1.8711,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0012958579881656804,
+      "loss": 1.8789,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0013017751479289942,
+      "loss": 1.8262,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0013076923076923077,
+      "loss": 1.8262,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0013136094674556214,
+      "loss": 1.748,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0013195266272189351,
+      "loss": 1.8906,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0013254437869822486,
+      "loss": 1.8477,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0013313609467455621,
+      "loss": 1.918,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0013372781065088756,
+      "loss": 1.8691,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0013431952662721893,
+      "loss": 1.9434,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001349112426035503,
+      "loss": 1.7695,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0013550295857988166,
+      "loss": 1.8477,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0013609467455621303,
+      "loss": 1.8125,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001366863905325444,
+      "loss": 1.9004,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0013727810650887573,
+      "loss": 1.8477,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001378698224852071,
+      "loss": 1.877,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0013846153846153845,
+      "loss": 1.7969,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0013905325443786982,
+      "loss": 1.8496,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001396449704142012,
+      "loss": 1.8887,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0014023668639053255,
+      "loss": 1.9023,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0014082840236686392,
+      "loss": 1.7559,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0014142011834319525,
+      "loss": 1.6758,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0014201183431952662,
+      "loss": 1.8828,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00142603550295858,
+      "loss": 1.7461,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0014319526627218934,
+      "loss": 1.7773,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0014378698224852072,
+      "loss": 1.8086,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0014437869822485209,
+      "loss": 2.002,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0014497041420118344,
+      "loss": 1.7988,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001455621301775148,
+      "loss": 1.8828,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0014615384615384614,
+      "loss": 1.7305,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001467455621301775,
+      "loss": 1.7578,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0014733727810650888,
+      "loss": 1.8789,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0014792899408284023,
+      "loss": 1.7246,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001485207100591716,
+      "loss": 1.8027,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0014911242603550298,
+      "loss": 1.7441,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0014970414201183433,
+      "loss": 1.6816,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0015029585798816568,
+      "loss": 1.8633,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0015088757396449703,
+      "loss": 1.8711,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001514792899408284,
+      "loss": 1.8164,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0015207100591715977,
+      "loss": 1.8242,
+      "step": 257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0015266272189349112,
+      "loss": 1.7559,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001532544378698225,
+      "loss": 1.8984,
+      "step": 259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0015384615384615387,
+      "loss": 1.7012,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001544378698224852,
+      "loss": 1.7344,
+      "step": 261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0015502958579881657,
+      "loss": 1.7363,
+      "step": 262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0015562130177514792,
+      "loss": 1.7109,
+      "step": 263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001562130177514793,
+      "loss": 1.8008,
+      "step": 264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0015680473372781066,
+      "loss": 1.8633,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0015739644970414201,
+      "loss": 1.7773,
+      "step": 266
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0015798816568047339,
+      "loss": 1.9023,
+      "step": 267
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0015857988165680476,
+      "loss": 1.8711,
+      "step": 268
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0015917159763313609,
+      "loss": 1.7285,
+      "step": 269
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0015976331360946746,
+      "loss": 1.8555,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001603550295857988,
+      "loss": 1.8867,
+      "step": 271
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0016094674556213018,
+      "loss": 1.8418,
+      "step": 272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0016153846153846155,
+      "loss": 1.8379,
+      "step": 273
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001621301775147929,
+      "loss": 1.8789,
+      "step": 274
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0016272189349112428,
+      "loss": 1.7383,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001633136094674556,
+      "loss": 1.7598,
+      "step": 276
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0016390532544378698,
+      "loss": 1.7949,
+      "step": 277
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0016449704142011835,
+      "loss": 1.9961,
+      "step": 278
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001650887573964497,
+      "loss": 1.8105,
+      "step": 279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0016568047337278107,
+      "loss": 1.9062,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0016627218934911244,
+      "loss": 1.7773,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001668639053254438,
+      "loss": 1.7188,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0016745562130177514,
+      "loss": 1.8555,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001680473372781065,
+      "loss": 1.8691,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0016863905325443787,
+      "loss": 1.832,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0016923076923076924,
+      "loss": 1.8477,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001698224852071006,
+      "loss": 1.8984,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0017041420118343196,
+      "loss": 1.8555,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0017100591715976333,
+      "loss": 1.832,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0017159763313609468,
+      "loss": 1.8691,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0017218934911242603,
+      "loss": 1.7168,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0017278106508875739,
+      "loss": 1.7441,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0017337278106508876,
+      "loss": 1.709,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0017396449704142013,
+      "loss": 1.7793,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0017455621301775148,
+      "loss": 1.8633,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0017514792899408285,
+      "loss": 1.7305,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0017573964497041422,
+      "loss": 1.8164,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0017633136094674555,
+      "loss": 1.8164,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0017692307692307693,
+      "loss": 1.8359,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0017751479289940828,
+      "loss": 1.8496,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0017810650887573965,
+      "loss": 1.8613,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0017869822485207102,
+      "loss": 1.7598,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0017928994082840237,
+      "loss": 1.8633,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0017988165680473374,
+      "loss": 1.8867,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0018047337278106507,
+      "loss": 1.8047,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0018106508875739644,
+      "loss": 1.8086,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0018165680473372782,
+      "loss": 1.7617,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0018224852071005917,
+      "loss": 1.8008,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0018284023668639054,
+      "loss": 1.9238,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001834319526627219,
+      "loss": 1.8242,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0018402366863905326,
+      "loss": 1.9336,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0018461538461538463,
+      "loss": 1.8535,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0018520710059171596,
+      "loss": 1.7441,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0018579881656804733,
+      "loss": 1.8691,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001863905325443787,
+      "loss": 1.8828,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0018698224852071006,
+      "loss": 1.7598,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0018757396449704143,
+      "loss": 1.7695,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001881656804733728,
+      "loss": 1.8184,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0018875739644970415,
+      "loss": 1.8477,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001893491124260355,
+      "loss": 1.8301,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0018994082840236685,
+      "loss": 1.8125,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019053254437869822,
+      "loss": 1.9004,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001911242603550296,
+      "loss": 1.8672,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019171597633136095,
+      "loss": 1.8047,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019230769230769232,
+      "loss": 1.6836,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001928994082840237,
+      "loss": 1.8789,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019349112426035502,
+      "loss": 1.8125,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001940828402366864,
+      "loss": 1.7109,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019467455621301774,
+      "loss": 1.7441,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019526627218934911,
+      "loss": 1.998,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019585798816568046,
+      "loss": 1.8613,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019644970414201186,
+      "loss": 1.8691,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001970414201183432,
+      "loss": 1.7637,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019763313609467456,
+      "loss": 1.7266,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001982248520710059,
+      "loss": 1.7871,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019881656804733726,
+      "loss": 1.8301,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019940828402366865,
+      "loss": 1.8828,
+      "step": 337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.002,
+      "loss": 1.832,
+      "step": 338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999999584876115,
+      "loss": 1.791,
+      "step": 339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999998339504496,
+      "loss": 1.7656,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999996263885236,
+      "loss": 1.8145,
+      "step": 341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999993358018526,
+      "loss": 1.8633,
+      "step": 342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999998962190459,
+      "loss": 1.8906,
+      "step": 343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999985055543746,
+      "loss": 1.8398,
+      "step": 344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999997965893637,
+      "loss": 1.8926,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999973432082915,
+      "loss": 1.8105,
+      "step": 346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999996637498389,
+      "loss": 1.8242,
+      "step": 347
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999958487639893,
+      "loss": 1.8145,
+      "step": 348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999994977005157,
+      "loss": 1.8926,
+      "step": 349
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999940222219645,
+      "loss": 1.8027,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999929844144912,
+      "loss": 1.7754,
+      "step": 351
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999918635828233,
+      "loss": 1.8809,
+      "step": 352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999990659727054,
+      "loss": 1.6914,
+      "step": 353
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999893728472828,
+      "loss": 1.8145,
+      "step": 354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999988002943617,
+      "loss": 1.8574,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999865500161702,
+      "loss": 1.9062,
+      "step": 356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999985014065063,
+      "loss": 1.7656,
+      "step": 357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999833950904225,
+      "loss": 1.8027,
+      "step": 358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999981693092384,
+      "loss": 1.9082,
+      "step": 359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999979908071088,
+      "loss": 1.834,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999780400266837,
+      "loss": 1.7715,
+      "step": 361
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999760889593252,
+      "loss": 1.9512,
+      "step": 362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999740548691746,
+      "loss": 1.6934,
+      "step": 363
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999719377564013,
+      "loss": 1.9688,
+      "step": 364
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999697376211806,
+      "loss": 1.7852,
+      "step": 365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999674544636955,
+      "loss": 1.7949,
+      "step": 366
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999965088284135,
+      "loss": 1.7578,
+      "step": 367
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999626390826966,
+      "loss": 1.8867,
+      "step": 368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999601068595827,
+      "loss": 1.8262,
+      "step": 369
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999574916150036,
+      "loss": 1.8574,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999547933491767,
+      "loss": 1.8516,
+      "step": 371
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999952012062326,
+      "loss": 1.793,
+      "step": 372
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999491477546826,
+      "loss": 1.8867,
+      "step": 373
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999462004264843,
+      "loss": 1.6699,
+      "step": 374
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999431700779753,
+      "loss": 1.7324,
+      "step": 375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999400567094075,
+      "loss": 1.8242,
+      "step": 376
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999368603210396,
+      "loss": 1.8691,
+      "step": 377
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999933580913137,
+      "loss": 1.8027,
+      "step": 378
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999930218485971,
+      "loss": 1.7188,
+      "step": 379
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999267730398222,
+      "loss": 1.7715,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999923244574976,
+      "loss": 1.8594,
+      "step": 381
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999196330917246,
+      "loss": 1.8203,
+      "step": 382
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999915938590369,
+      "loss": 1.791,
+      "step": 383
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999121610712154,
+      "loss": 1.7246,
+      "step": 384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999083005345776,
+      "loss": 1.8477,
+      "step": 385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999904356980776,
+      "loss": 1.9648,
+      "step": 386
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019999003304101383,
+      "loss": 1.7988,
+      "step": 387
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019998962208229985,
+      "loss": 1.8672,
+      "step": 388
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019998920282196975,
+      "loss": 1.7578,
+      "step": 389
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999887752600584,
+      "loss": 1.8105,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019998833939660126,
+      "loss": 1.8477,
+      "step": 391
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0019998789523163456,
+      "loss": 1.7539,
+      "step": 392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001999874427651951,
+      "loss": 1.8262,
+      "step": 393
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019998698199732053,
+      "loss": 1.8027,
+      "step": 394
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019998651292804904,
+      "loss": 1.7188,
+      "step": 395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019998603555741964,
+      "loss": 1.7109,
+      "step": 396
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019998554988547193,
+      "loss": 1.8789,
+      "step": 397
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999850559122462,
+      "loss": 1.9414,
+      "step": 398
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999845536377835,
+      "loss": 1.8828,
+      "step": 399
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019998404306212554,
+      "loss": 1.7051,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019998352418531466,
+      "loss": 1.7031,
+      "step": 401
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00199982997007394,
+      "loss": 1.7051,
+      "step": 402
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019998246152840727,
+      "loss": 1.7441,
+      "step": 403
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00199981917748399,
+      "loss": 1.7734,
+      "step": 404
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019998136566741426,
+      "loss": 1.9043,
+      "step": 405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019998080528549895,
+      "loss": 1.9492,
+      "step": 406
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019998023660269958,
+      "loss": 1.9219,
+      "step": 407
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999796596190633,
+      "loss": 1.8906,
+      "step": 408
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999790743346381,
+      "loss": 1.8691,
+      "step": 409
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019997848074947257,
+      "loss": 1.7305,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999778788636159,
+      "loss": 1.6387,
+      "step": 411
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999772686771182,
+      "loss": 1.7773,
+      "step": 412
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019997665019002996,
+      "loss": 1.7812,
+      "step": 413
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999760234024027,
+      "loss": 1.9238,
+      "step": 414
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019997538831428834,
+      "loss": 1.748,
+      "step": 415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999747449257397,
+      "loss": 1.8438,
+      "step": 416
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999740932368101,
+      "loss": 1.7949,
+      "step": 417
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999734332475537,
+      "loss": 1.8262,
+      "step": 418
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999727649580253,
+      "loss": 1.7871,
+      "step": 419
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999720883682804,
+      "loss": 1.8379,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019997140347837515,
+      "loss": 1.7617,
+      "step": 421
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019997071028836637,
+      "loss": 1.8281,
+      "step": 422
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019997000879831165,
+      "loss": 1.6836,
+      "step": 423
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999692990082693,
+      "loss": 1.7871,
+      "step": 424
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999685809182981,
+      "loss": 1.6562,
+      "step": 425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019996785452845783,
+      "loss": 1.7305,
+      "step": 426
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999671198388087,
+      "loss": 1.8125,
+      "step": 427
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999663768494117,
+      "loss": 1.6777,
+      "step": 428
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019996562556032858,
+      "loss": 1.8223,
+      "step": 429
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019996486597162163,
+      "loss": 1.7188,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019996409808335403,
+      "loss": 1.6855,
+      "step": 431
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019996332189558944,
+      "loss": 1.8613,
+      "step": 432
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019996253740839232,
+      "loss": 1.7051,
+      "step": 433
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019996174462182784,
+      "loss": 1.8359,
+      "step": 434
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999609435359618,
+      "loss": 1.8438,
+      "step": 435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999601341508607,
+      "loss": 1.8555,
+      "step": 436
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999593164665917,
+      "loss": 1.7734,
+      "step": 437
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019995849048322282,
+      "loss": 1.8477,
+      "step": 438
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999576562008225,
+      "loss": 1.7402,
+      "step": 439
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019995681361946005,
+      "loss": 1.8535,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019995596273920543,
+      "loss": 1.8516,
+      "step": 441
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999551035601293,
+      "loss": 1.8301,
+      "step": 442
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00199954236082303,
+      "loss": 1.7812,
+      "step": 443
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999533603057985,
+      "loss": 1.8672,
+      "step": 444
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019995247623068855,
+      "loss": 1.7949,
+      "step": 445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999515838570465,
+      "loss": 1.7441,
+      "step": 446
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019995068318494655,
+      "loss": 1.7773,
+      "step": 447
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019994977421446338,
+      "loss": 1.707,
+      "step": 448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999488569456725,
+      "loss": 1.7617,
+      "step": 449
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019994793137865004,
+      "loss": 1.7129,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019994699751347284,
+      "loss": 1.7324,
+      "step": 451
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999460553502185,
+      "loss": 1.8203,
+      "step": 452
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999451048889651,
+      "loss": 1.7227,
+      "step": 453
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019994414612979176,
+      "loss": 1.8574,
+      "step": 454
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999431790727779,
+      "loss": 1.8223,
+      "step": 455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999422037180039,
+      "loss": 1.8281,
+      "step": 456
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999412200655507,
+      "loss": 1.8398,
+      "step": 457
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999402281155,
+      "loss": 1.748,
+      "step": 458
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019993922786793413,
+      "loss": 1.7891,
+      "step": 459
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019993821932293614,
+      "loss": 1.8691,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019993720248058975,
+      "loss": 1.8223,
+      "step": 461
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019993617734097944,
+      "loss": 1.7031,
+      "step": 462
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019993514390419026,
+      "loss": 1.7109,
+      "step": 463
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019993410217030807,
+      "loss": 1.7734,
+      "step": 464
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019993305213941927,
+      "loss": 1.8164,
+      "step": 465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019993199381161115,
+      "loss": 1.8574,
+      "step": 466
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019993092718697146,
+      "loss": 1.8027,
+      "step": 467
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019992985226558882,
+      "loss": 1.75,
+      "step": 468
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019992876904755255,
+      "loss": 1.8105,
+      "step": 469
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999276775329524,
+      "loss": 1.7695,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999265777218792,
+      "loss": 1.7246,
+      "step": 471
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999254696144241,
+      "loss": 1.7422,
+      "step": 472
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019992435321067917,
+      "loss": 1.9316,
+      "step": 473
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999232285107371,
+      "loss": 1.6855,
+      "step": 474
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019992209551469124,
+      "loss": 1.748,
+      "step": 475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999209542226357,
+      "loss": 1.8281,
+      "step": 476
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999198046346652,
+      "loss": 1.7578,
+      "step": 477
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019991864675087518,
+      "loss": 1.7988,
+      "step": 478
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999174805713618,
+      "loss": 1.752,
+      "step": 479
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019991630609622188,
+      "loss": 1.7637,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999151233255529,
+      "loss": 1.7305,
+      "step": 481
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019991393225945306,
+      "loss": 1.7383,
+      "step": 482
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999127328980213,
+      "loss": 1.8477,
+      "step": 483
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019991152524135715,
+      "loss": 1.7109,
+      "step": 484
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999103092895609,
+      "loss": 1.623,
+      "step": 485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019990908504273347,
+      "loss": 1.7031,
+      "step": 486
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019990785250097653,
+      "loss": 1.752,
+      "step": 487
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019990661166439244,
+      "loss": 1.7207,
+      "step": 488
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019990536253308416,
+      "loss": 1.6895,
+      "step": 489
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001999041051071554,
+      "loss": 1.7129,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019990283938671064,
+      "loss": 1.8516,
+      "step": 491
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019990156537185487,
+      "loss": 1.7578,
+      "step": 492
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019990028306269393,
+      "loss": 1.6152,
+      "step": 493
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001998989924593343,
+      "loss": 1.7539,
+      "step": 494
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019989769356188298,
+      "loss": 1.752,
+      "step": 495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00199896386370448,
+      "loss": 1.6875,
+      "step": 496
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001998950708851378,
+      "loss": 1.7617,
+      "step": 497
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001998937471060616,
+      "loss": 1.8047,
+      "step": 498
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001998924150333293,
+      "loss": 1.7109,
+      "step": 499
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019989107466705154,
+      "loss": 1.8203,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001998897260073395,
+      "loss": 1.7227,
+      "step": 501
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001998883690543053,
+      "loss": 1.6875,
+      "step": 502
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0019988700380806153,
+      "loss": 1.6836,
+      "step": 503
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001998856302687215,
+      "loss": 1.8027,
+      "step": 504
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.001998842484363993,
+      "loss": 1.7012,
+      "step": 505
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019988285831120966,
+      "loss": 1.8359,
+      "step": 506
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019988145989326793,
+      "loss": 1.8145,
+      "step": 507
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019988005318269026,
+      "loss": 1.8203,
+      "step": 508
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998786381795935,
+      "loss": 1.8008,
+      "step": 509
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00199877214884095,
+      "loss": 1.8203,
+      "step": 510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00199875783296313,
+      "loss": 1.6836,
+      "step": 511
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998743434163664,
+      "loss": 1.9004,
+      "step": 512
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019987289524437467,
+      "loss": 1.7148,
+      "step": 513
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998714387804581,
+      "loss": 1.9492,
+      "step": 514
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998699740247376,
+      "loss": 1.6465,
+      "step": 515
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019986850097733475,
+      "loss": 1.748,
+      "step": 516
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019986701963837184,
+      "loss": 1.7617,
+      "step": 517
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998655300079719,
+      "loss": 1.7227,
+      "step": 518
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019986403208625864,
+      "loss": 1.75,
+      "step": 519
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998625258733563,
+      "loss": 1.6562,
+      "step": 520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019986101136939007,
+      "loss": 1.8242,
+      "step": 521
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998594885744856,
+      "loss": 1.7324,
+      "step": 522
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019985795748876938,
+      "loss": 1.8105,
+      "step": 523
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998564181123685,
+      "loss": 1.7812,
+      "step": 524
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019985487044541076,
+      "loss": 1.7773,
+      "step": 525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019985331448802465,
+      "loss": 1.793,
+      "step": 526
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019985175024033936,
+      "loss": 1.8164,
+      "step": 527
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998501777024848,
+      "loss": 1.8828,
+      "step": 528
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998485968745914,
+      "loss": 1.6973,
+      "step": 529
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998470077567906,
+      "loss": 1.7441,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998454103492142,
+      "loss": 1.7734,
+      "step": 531
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998438046519949,
+      "loss": 1.7402,
+      "step": 532
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019984219066526594,
+      "loss": 1.9395,
+      "step": 533
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019984056838916135,
+      "loss": 1.6875,
+      "step": 534
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019983893782381584,
+      "loss": 1.6621,
+      "step": 535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019983729896936478,
+      "loss": 1.7988,
+      "step": 536
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998356518259442,
+      "loss": 1.8223,
+      "step": 537
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998339963936909,
+      "loss": 1.6914,
+      "step": 538
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998323326727423,
+      "loss": 1.6172,
+      "step": 539
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019983066066323655,
+      "loss": 1.8262,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019982898036531244,
+      "loss": 1.75,
+      "step": 541
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998272917791095,
+      "loss": 1.7891,
+      "step": 542
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998255949047679,
+      "loss": 1.6699,
+      "step": 543
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019982388974242858,
+      "loss": 1.9043,
+      "step": 544
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00199822176292233,
+      "loss": 1.7656,
+      "step": 545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998204545543235,
+      "loss": 1.6621,
+      "step": 546
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019981872452884304,
+      "loss": 1.6875,
+      "step": 547
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019981698621593523,
+      "loss": 1.6875,
+      "step": 548
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019981523961574437,
+      "loss": 1.7305,
+      "step": 549
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019981348472841547,
+      "loss": 1.7832,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998117215540943,
+      "loss": 1.7637,
+      "step": 551
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019980995009292713,
+      "loss": 1.7578,
+      "step": 552
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019980817034506116,
+      "loss": 1.793,
+      "step": 553
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001998063823106441,
+      "loss": 1.6934,
+      "step": 554
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019980458598982437,
+      "loss": 1.6387,
+      "step": 555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019980278138275113,
+      "loss": 1.7129,
+      "step": 556
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019980096848957424,
+      "loss": 1.9668,
+      "step": 557
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019979914731044417,
+      "loss": 1.7012,
+      "step": 558
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019979731784551214,
+      "loss": 1.8125,
+      "step": 559
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019979548009493005,
+      "loss": 1.8457,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997936340588505,
+      "loss": 1.7324,
+      "step": 561
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997917797374267,
+      "loss": 1.8301,
+      "step": 562
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997899171308126,
+      "loss": 1.9395,
+      "step": 563
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019978804623916294,
+      "loss": 1.8047,
+      "step": 564
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019978616706263297,
+      "loss": 1.7461,
+      "step": 565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019978427960137867,
+      "loss": 1.7734,
+      "step": 566
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019978238385555684,
+      "loss": 1.7773,
+      "step": 567
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019978047982532484,
+      "loss": 1.7324,
+      "step": 568
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019977856751084073,
+      "loss": 1.8008,
+      "step": 569
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019977664691226333,
+      "loss": 1.6719,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00199774718029752,
+      "loss": 1.8457,
+      "step": 571
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00199772780863467,
+      "loss": 1.7188,
+      "step": 572
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019977083541356908,
+      "loss": 1.7656,
+      "step": 573
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997688816802198,
+      "loss": 1.7031,
+      "step": 574
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019976691966358136,
+      "loss": 1.7598,
+      "step": 575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997649493638167,
+      "loss": 1.8164,
+      "step": 576
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997629707810893,
+      "loss": 1.7988,
+      "step": 577
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997609839155635,
+      "loss": 1.7148,
+      "step": 578
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019975898876740425,
+      "loss": 1.7324,
+      "step": 579
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997569853367772,
+      "loss": 1.8105,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019975497362384863,
+      "loss": 1.7871,
+      "step": 581
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997529536287857,
+      "loss": 1.7598,
+      "step": 582
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00199750925351756,
+      "loss": 1.8516,
+      "step": 583
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997488887929279,
+      "loss": 1.668,
+      "step": 584
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997468439524706,
+      "loss": 1.8086,
+      "step": 585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019974479083055383,
+      "loss": 1.6914,
+      "step": 586
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019974272942734806,
+      "loss": 1.7305,
+      "step": 587
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019974065974302435,
+      "loss": 1.6855,
+      "step": 588
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997385817777546,
+      "loss": 1.6895,
+      "step": 589
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019973649553171138,
+      "loss": 1.8418,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997344010050678,
+      "loss": 1.8379,
+      "step": 591
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019973229819799784,
+      "loss": 1.7422,
+      "step": 592
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019973018711067604,
+      "loss": 1.8008,
+      "step": 593
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019972806774327767,
+      "loss": 1.7168,
+      "step": 594
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019972594009597874,
+      "loss": 1.6992,
+      "step": 595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997238041689558,
+      "loss": 1.7656,
+      "step": 596
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997216599623863,
+      "loss": 1.8418,
+      "step": 597
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019971950747644826,
+      "loss": 1.7168,
+      "step": 598
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997173467113203,
+      "loss": 1.8262,
+      "step": 599
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019971517766718184,
+      "loss": 1.793,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00199713000344213,
+      "loss": 1.668,
+      "step": 601
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997108147425945,
+      "loss": 1.748,
+      "step": 602
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997086208625079,
+      "loss": 1.8281,
+      "step": 603
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019970641870413522,
+      "loss": 1.6992,
+      "step": 604
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001997042082676594,
+      "loss": 1.7656,
+      "step": 605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019970198955326385,
+      "loss": 1.7715,
+      "step": 606
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019969976256113287,
+      "loss": 1.8105,
+      "step": 607
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019969752729145137,
+      "loss": 1.7559,
+      "step": 608
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019969528374440488,
+      "loss": 1.5664,
+      "step": 609
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019969303192017966,
+      "loss": 1.8008,
+      "step": 610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019969077181896267,
+      "loss": 1.7988,
+      "step": 611
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.001996885034409416,
+      "loss": 1.9199,
+      "step": 612
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019968622678630476,
+      "loss": 1.7207,
+      "step": 613
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019968394185524113,
+      "loss": 1.6094,
+      "step": 614
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019968164864794047,
+      "loss": 1.7148,
+      "step": 615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019967934716459317,
+      "loss": 1.8418,
+      "step": 616
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019967703740539026,
+      "loss": 1.6602,
+      "step": 617
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0019967471937052356,
+      "loss": 1.7812,
+      "step": 618
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001996723930601855,
+      "loss": 1.752,
+      "step": 619
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001996700584745692,
+      "loss": 1.748,
+      "step": 620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019966771561386854,
+      "loss": 1.752,
+      "step": 621
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00199665364478278,
+      "loss": 1.8242,
+      "step": 622
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001996630050679928,
+      "loss": 1.6992,
+      "step": 623
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019966063738320883,
+      "loss": 1.7871,
+      "step": 624
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019965826142412262,
+      "loss": 1.7891,
+      "step": 625
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001996558771909315,
+      "loss": 1.6797,
+      "step": 626
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019965348468383337,
+      "loss": 1.7188,
+      "step": 627
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019965108390302693,
+      "loss": 1.8848,
+      "step": 628
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001996486748487114,
+      "loss": 1.7871,
+      "step": 629
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001996462575210869,
+      "loss": 1.6836,
+      "step": 630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001996438319203541,
+      "loss": 1.7324,
+      "step": 631
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001996413980467143,
+      "loss": 1.6738,
+      "step": 632
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001996389559003697,
+      "loss": 1.8379,
+      "step": 633
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019963650548152296,
+      "loss": 1.7441,
+      "step": 634
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019963404679037757,
+      "loss": 1.8203,
+      "step": 635
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019963157982713768,
+      "loss": 1.7676,
+      "step": 636
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019962910459200805,
+      "loss": 1.7715,
+      "step": 637
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019962662108519427,
+      "loss": 1.752,
+      "step": 638
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019962412930690245,
+      "loss": 1.8535,
+      "step": 639
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001996216292573395,
+      "loss": 1.7773,
+      "step": 640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00199619120936713,
+      "loss": 1.627,
+      "step": 641
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001996166043452312,
+      "loss": 1.6992,
+      "step": 642
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00199614079483103,
+      "loss": 1.7617,
+      "step": 643
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019961154635053805,
+      "loss": 1.8262,
+      "step": 644
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001996090049477467,
+      "loss": 1.7246,
+      "step": 645
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001996064552749399,
+      "loss": 1.7402,
+      "step": 646
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019960389733232934,
+      "loss": 1.7383,
+      "step": 647
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019960133112012744,
+      "loss": 1.9062,
+      "step": 648
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019959875663854722,
+      "loss": 1.7676,
+      "step": 649
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019959617388780244,
+      "loss": 1.7773,
+      "step": 650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001995935828681075,
+      "loss": 1.7383,
+      "step": 651
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019959098357967757,
+      "loss": 1.5938,
+      "step": 652
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019958837602272843,
+      "loss": 1.7559,
+      "step": 653
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001995857601974766,
+      "loss": 1.7441,
+      "step": 654
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019958313610413912,
+      "loss": 1.8594,
+      "step": 655
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019958050374293402,
+      "loss": 1.8242,
+      "step": 656
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001995778631140798,
+      "loss": 1.748,
+      "step": 657
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019957521421779573,
+      "loss": 1.6602,
+      "step": 658
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019957255705430168,
+      "loss": 1.707,
+      "step": 659
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019956989162381827,
+      "loss": 1.8027,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001995672179265668,
+      "loss": 1.7461,
+      "step": 661
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001995645359627692,
+      "loss": 1.7051,
+      "step": 662
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001995618457326483,
+      "loss": 1.7539,
+      "step": 663
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001995591472364273,
+      "loss": 1.7305,
+      "step": 664
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019955644047433027,
+      "loss": 1.707,
+      "step": 665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019955372544658203,
+      "loss": 1.7793,
+      "step": 666
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019955100215340786,
+      "loss": 1.7148,
+      "step": 667
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019954827059503397,
+      "loss": 1.7129,
+      "step": 668
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001995455307716871,
+      "loss": 1.7539,
+      "step": 669
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019954278268359466,
+      "loss": 1.7207,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019954002633098497,
+      "loss": 1.6855,
+      "step": 671
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001995372617140867,
+      "loss": 1.6855,
+      "step": 672
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019953448883312953,
+      "loss": 1.8105,
+      "step": 673
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001995317076883436,
+      "loss": 1.7715,
+      "step": 674
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019952891827995983,
+      "loss": 1.7656,
+      "step": 675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001995261206082098,
+      "loss": 1.7285,
+      "step": 676
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001995233146733258,
+      "loss": 1.7695,
+      "step": 677
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019952050047554078,
+      "loss": 1.8105,
+      "step": 678
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001995176780150884,
+      "loss": 1.75,
+      "step": 679
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019951484729220296,
+      "loss": 1.6777,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001995120083071195,
+      "loss": 1.7188,
+      "step": 681
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019950916106007376,
+      "loss": 1.6406,
+      "step": 682
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001995063055513021,
+      "loss": 1.7812,
+      "step": 683
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001995034417810416,
+      "loss": 1.793,
+      "step": 684
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019950056974953004,
+      "loss": 1.6992,
+      "step": 685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019949768945700585,
+      "loss": 1.8535,
+      "step": 686
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019949480090370817,
+      "loss": 1.8828,
+      "step": 687
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019949190408987683,
+      "loss": 1.7305,
+      "step": 688
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994889990157523,
+      "loss": 1.8379,
+      "step": 689
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019948608568157583,
+      "loss": 1.8613,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994831640875893,
+      "loss": 1.7773,
+      "step": 691
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994802342340352,
+      "loss": 1.7266,
+      "step": 692
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994772961211569,
+      "loss": 1.5684,
+      "step": 693
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019947434974919814,
+      "loss": 1.7324,
+      "step": 694
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019947139511840377,
+      "loss": 1.8711,
+      "step": 695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00199468432229019,
+      "loss": 1.6855,
+      "step": 696
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019946546108128974,
+      "loss": 1.7266,
+      "step": 697
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994624816754628,
+      "loss": 1.752,
+      "step": 698
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994594940117855,
+      "loss": 1.6953,
+      "step": 699
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019945649809050585,
+      "loss": 1.7109,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994534939118726,
+      "loss": 1.873,
+      "step": 701
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019945048147613524,
+      "loss": 1.8125,
+      "step": 702
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994474607835438,
+      "loss": 1.7891,
+      "step": 703
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994444318343491,
+      "loss": 1.7402,
+      "step": 704
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019944139462880262,
+      "loss": 1.7266,
+      "step": 705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994383491671565,
+      "loss": 1.7285,
+      "step": 706
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019943529544966363,
+      "loss": 1.7871,
+      "step": 707
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994322334765775,
+      "loss": 1.8535,
+      "step": 708
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994291632481524,
+      "loss": 1.6973,
+      "step": 709
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994260847646431,
+      "loss": 1.877,
+      "step": 710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994229980263053,
+      "loss": 1.7344,
+      "step": 711
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994199030333953,
+      "loss": 1.6836,
+      "step": 712
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019941679978617,
+      "loss": 1.7012,
+      "step": 713
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019941368828488705,
+      "loss": 1.623,
+      "step": 714
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994105685298048,
+      "loss": 1.834,
+      "step": 715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994074405211822,
+      "loss": 1.6953,
+      "step": 716
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001994043042592791,
+      "loss": 1.7402,
+      "step": 717
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019940115974435573,
+      "loss": 1.7734,
+      "step": 718
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019939800697667326,
+      "loss": 1.7305,
+      "step": 719
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001993948459564934,
+      "loss": 1.8672,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019939167668407863,
+      "loss": 1.5957,
+      "step": 721
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019938849915969205,
+      "loss": 1.6738,
+      "step": 722
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019938531338359747,
+      "loss": 1.7188,
+      "step": 723
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019938211935605944,
+      "loss": 1.6914,
+      "step": 724
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001993789170773431,
+      "loss": 1.7422,
+      "step": 725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001993757065477143,
+      "loss": 1.7637,
+      "step": 726
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001993724877674396,
+      "loss": 1.752,
+      "step": 727
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001993692607367863,
+      "loss": 1.7246,
+      "step": 728
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0019936602545602227,
+      "loss": 1.7461,
+      "step": 729
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001993627819254161,
+      "loss": 1.7441,
+      "step": 730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019935953014523712,
+      "loss": 1.7363,
+      "step": 731
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001993562701157553,
+      "loss": 1.7598,
+      "step": 732
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001993530018372413,
+      "loss": 1.6504,
+      "step": 733
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001993497253099665,
+      "loss": 1.6113,
+      "step": 734
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019934644053420285,
+      "loss": 1.791,
+      "step": 735
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019934314751022314,
+      "loss": 1.7969,
+      "step": 736
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019933984623830077,
+      "loss": 1.8516,
+      "step": 737
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001993365367187098,
+      "loss": 1.6367,
+      "step": 738
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00199333218951725,
+      "loss": 1.7305,
+      "step": 739
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019932989293762185,
+      "loss": 1.8203,
+      "step": 740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001993265586766765,
+      "loss": 1.6445,
+      "step": 741
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001993232161691657,
+      "loss": 1.6895,
+      "step": 742
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019931986541536707,
+      "loss": 1.6543,
+      "step": 743
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001993165064155588,
+      "loss": 1.6738,
+      "step": 744
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001993131391700196,
+      "loss": 1.7715,
+      "step": 745
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019930976367902925,
+      "loss": 1.7344,
+      "step": 746
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019930637994286793,
+      "loss": 1.7617,
+      "step": 747
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001993029879618165,
+      "loss": 1.8223,
+      "step": 748
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019929958773615662,
+      "loss": 1.7422,
+      "step": 749
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019929617926617064,
+      "loss": 1.7754,
+      "step": 750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001992927625521415,
+      "loss": 1.6777,
+      "step": 751
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019928933759435284,
+      "loss": 1.6523,
+      "step": 752
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001992859043930891,
+      "loss": 1.7559,
+      "step": 753
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001992824629486353,
+      "loss": 1.8965,
+      "step": 754
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019927901326127714,
+      "loss": 1.791,
+      "step": 755
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00199275555331301,
+      "loss": 1.75,
+      "step": 756
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019927208915899402,
+      "loss": 1.8516,
+      "step": 757
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00199268614744644,
+      "loss": 1.8086,
+      "step": 758
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019926513208853936,
+      "loss": 1.7363,
+      "step": 759
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001992616411909692,
+      "loss": 1.7031,
+      "step": 760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019925814205222344,
+      "loss": 1.7441,
+      "step": 761
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019925463467259257,
+      "loss": 1.6602,
+      "step": 762
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019925111905236778,
+      "loss": 1.9531,
+      "step": 763
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019924759519184096,
+      "loss": 1.7363,
+      "step": 764
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001992440630913047,
+      "loss": 1.6699,
+      "step": 765
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019924052275105218,
+      "loss": 1.6621,
+      "step": 766
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001992369741713774,
+      "loss": 1.6816,
+      "step": 767
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019923341735257494,
+      "loss": 1.832,
+      "step": 768
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019922985229494016,
+      "loss": 1.6074,
+      "step": 769
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019922627899876897,
+      "loss": 1.7246,
+      "step": 770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019922269746435813,
+      "loss": 1.8887,
+      "step": 771
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019921910769200493,
+      "loss": 1.6016,
+      "step": 772
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019921550968200743,
+      "loss": 1.8828,
+      "step": 773
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019921190343466437,
+      "loss": 1.7051,
+      "step": 774
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019920828895027515,
+      "loss": 1.6387,
+      "step": 775
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019920466622913983,
+      "loss": 1.7441,
+      "step": 776
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019920103527155918,
+      "loss": 1.6758,
+      "step": 777
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019919739607783477,
+      "loss": 1.7402,
+      "step": 778
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001991937486482686,
+      "loss": 1.7285,
+      "step": 779
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019919009298316357,
+      "loss": 1.7285,
+      "step": 780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019918642908282325,
+      "loss": 1.6035,
+      "step": 781
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001991827569475517,
+      "loss": 1.6699,
+      "step": 782
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019917907657765388,
+      "loss": 1.7773,
+      "step": 783
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019917538797343533,
+      "loss": 1.6582,
+      "step": 784
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019917169113520226,
+      "loss": 1.6836,
+      "step": 785
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019916798606326174,
+      "loss": 1.7578,
+      "step": 786
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001991642727579212,
+      "loss": 1.6445,
+      "step": 787
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019916055121948903,
+      "loss": 1.6914,
+      "step": 788
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019915682144827423,
+      "loss": 1.6875,
+      "step": 789
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001991530834445864,
+      "loss": 1.6953,
+      "step": 790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001991493372087359,
+      "loss": 1.7363,
+      "step": 791
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019914558274103386,
+      "loss": 1.791,
+      "step": 792
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001991418200417919,
+      "loss": 1.6445,
+      "step": 793
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001991380491113224,
+      "loss": 1.6465,
+      "step": 794
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019913426994993847,
+      "loss": 1.7148,
+      "step": 795
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001991304825579539,
+      "loss": 1.6836,
+      "step": 796
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019912668693568307,
+      "loss": 1.752,
+      "step": 797
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001991228830834412,
+      "loss": 1.7188,
+      "step": 798
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019911907100154404,
+      "loss": 1.7148,
+      "step": 799
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001991152506903081,
+      "loss": 1.6777,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019911142215005054,
+      "loss": 1.6953,
+      "step": 801
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001991075853810893,
+      "loss": 1.6289,
+      "step": 802
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019910374038374285,
+      "loss": 1.7715,
+      "step": 803
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019909988715833046,
+      "loss": 1.8438,
+      "step": 804
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00199096025705172,
+      "loss": 1.6895,
+      "step": 805
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019909215602458815,
+      "loss": 1.5449,
+      "step": 806
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001990882781169001,
+      "loss": 1.7148,
+      "step": 807
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019908439198242983,
+      "loss": 1.6914,
+      "step": 808
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019908049762150004,
+      "loss": 1.918,
+      "step": 809
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00199076595034434,
+      "loss": 1.8242,
+      "step": 810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019907268422155577,
+      "loss": 1.6836,
+      "step": 811
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019906876518318998,
+      "loss": 1.7656,
+      "step": 812
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019906483791966205,
+      "loss": 1.6738,
+      "step": 813
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001990609024312981,
+      "loss": 1.6855,
+      "step": 814
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001990569587184247,
+      "loss": 1.7441,
+      "step": 815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019905300678136947,
+      "loss": 1.6777,
+      "step": 816
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001990490466204604,
+      "loss": 1.8379,
+      "step": 817
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019904507823602633,
+      "loss": 1.7617,
+      "step": 818
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001990411016283967,
+      "loss": 1.6934,
+      "step": 819
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001990371167979017,
+      "loss": 1.7773,
+      "step": 820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019903312374487214,
+      "loss": 1.7383,
+      "step": 821
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019902912246963955,
+      "loss": 1.707,
+      "step": 822
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019902511297253617,
+      "loss": 1.7402,
+      "step": 823
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019902109525389484,
+      "loss": 1.666,
+      "step": 824
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019901706931404915,
+      "loss": 1.709,
+      "step": 825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019901303515333334,
+      "loss": 1.7109,
+      "step": 826
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019900899277208238,
+      "loss": 1.7129,
+      "step": 827
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019900494217063185,
+      "loss": 1.6934,
+      "step": 828
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001990008833493181,
+      "loss": 1.7266,
+      "step": 829
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019899681630847804,
+      "loss": 1.8066,
+      "step": 830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019899274104844943,
+      "loss": 1.6309,
+      "step": 831
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001989886575695705,
+      "loss": 1.6387,
+      "step": 832
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019898456587218037,
+      "loss": 1.7734,
+      "step": 833
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019898046595661872,
+      "loss": 1.7188,
+      "step": 834
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019897635782322592,
+      "loss": 1.7402,
+      "step": 835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001989722414723431,
+      "loss": 1.707,
+      "step": 836
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019896811690431203,
+      "loss": 1.7734,
+      "step": 837
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019896398411947507,
+      "loss": 1.7148,
+      "step": 838
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001989598431181754,
+      "loss": 1.6758,
+      "step": 839
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019895569390075683,
+      "loss": 1.7988,
+      "step": 840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001989515364675638,
+      "loss": 1.7852,
+      "step": 841
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0019894737081894154,
+      "loss": 1.7266,
+      "step": 842
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.001989431969552359,
+      "loss": 1.8281,
+      "step": 843
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019893901487679335,
+      "loss": 1.7715,
+      "step": 844
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019893482458396114,
+      "loss": 1.6914,
+      "step": 845
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019893062607708716,
+      "loss": 1.6094,
+      "step": 846
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019892641935652005,
+      "loss": 1.7246,
+      "step": 847
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00198922204422609,
+      "loss": 1.6641,
+      "step": 848
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00198917981275704,
+      "loss": 1.7578,
+      "step": 849
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001989137499161556,
+      "loss": 1.6348,
+      "step": 850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019890951034431522,
+      "loss": 1.6914,
+      "step": 851
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019890526256053477,
+      "loss": 1.6699,
+      "step": 852
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019890100656516693,
+      "loss": 1.707,
+      "step": 853
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001988967423585651,
+      "loss": 1.6855,
+      "step": 854
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019889246994108326,
+      "loss": 1.7031,
+      "step": 855
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019888818931307617,
+      "loss": 1.75,
+      "step": 856
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019888390047489916,
+      "loss": 1.6855,
+      "step": 857
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001988796034269084,
+      "loss": 1.7891,
+      "step": 858
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001988752981694606,
+      "loss": 1.7637,
+      "step": 859
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019887098470291317,
+      "loss": 1.8262,
+      "step": 860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001988666630276243,
+      "loss": 1.7305,
+      "step": 861
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019886233314395283,
+      "loss": 1.8301,
+      "step": 862
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001988579950522581,
+      "loss": 1.7715,
+      "step": 863
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001988536487529004,
+      "loss": 1.6914,
+      "step": 864
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001988492942462406,
+      "loss": 1.7422,
+      "step": 865
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001988449315326401,
+      "loss": 1.7188,
+      "step": 866
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019884056061246127,
+      "loss": 1.7051,
+      "step": 867
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019883618148606686,
+      "loss": 1.6816,
+      "step": 868
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019883179415382058,
+      "loss": 1.791,
+      "step": 869
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019882739861608652,
+      "loss": 1.8066,
+      "step": 870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019882299487322982,
+      "loss": 1.7012,
+      "step": 871
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019881858292561594,
+      "loss": 1.75,
+      "step": 872
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001988141627736113,
+      "loss": 1.6699,
+      "step": 873
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019880973441758275,
+      "loss": 1.6348,
+      "step": 874
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019880529785789806,
+      "loss": 1.7852,
+      "step": 875
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001988008530949256,
+      "loss": 1.7441,
+      "step": 876
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019879640012903424,
+      "loss": 1.6738,
+      "step": 877
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019879193896059385,
+      "loss": 1.7129,
+      "step": 878
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019878746958997473,
+      "loss": 1.7227,
+      "step": 879
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019878299201754796,
+      "loss": 1.7949,
+      "step": 880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019877850624368527,
+      "loss": 1.5879,
+      "step": 881
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001987740122687592,
+      "loss": 1.7285,
+      "step": 882
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019876951009314267,
+      "loss": 1.7031,
+      "step": 883
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001987649997172097,
+      "loss": 1.75,
+      "step": 884
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019876048114133454,
+      "loss": 1.8691,
+      "step": 885
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001987559543658925,
+      "loss": 1.6855,
+      "step": 886
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019875141939125936,
+      "loss": 1.6543,
+      "step": 887
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019874687621781162,
+      "loss": 1.7949,
+      "step": 888
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001987423248459265,
+      "loss": 1.8301,
+      "step": 889
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001987377652759819,
+      "loss": 1.793,
+      "step": 890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019873319750835635,
+      "loss": 1.7988,
+      "step": 891
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019872862154342906,
+      "loss": 1.6816,
+      "step": 892
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019872403738158,
+      "loss": 1.8105,
+      "step": 893
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001987194450231897,
+      "loss": 1.6523,
+      "step": 894
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001987148444686395,
+      "loss": 1.7207,
+      "step": 895
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019871023571831135,
+      "loss": 1.6035,
+      "step": 896
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001987056187725879,
+      "loss": 1.6426,
+      "step": 897
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019870099363185245,
+      "loss": 1.7266,
+      "step": 898
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00198696360296489,
+      "loss": 1.6992,
+      "step": 899
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019869171876688223,
+      "loss": 1.791,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001986870690434175,
+      "loss": 1.6406,
+      "step": 901
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001986824111264809,
+      "loss": 1.7422,
+      "step": 902
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001986777450164591,
+      "loss": 1.5859,
+      "step": 903
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001986730707137395,
+      "loss": 1.7363,
+      "step": 904
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001986683882187102,
+      "loss": 1.7012,
+      "step": 905
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019866369753175996,
+      "loss": 1.6621,
+      "step": 906
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019865899865327825,
+      "loss": 1.6816,
+      "step": 907
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019865429158365516,
+      "loss": 1.8027,
+      "step": 908
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001986495763232815,
+      "loss": 1.5918,
+      "step": 909
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019864485287254875,
+      "loss": 1.8086,
+      "step": 910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019864012123184905,
+      "loss": 1.6602,
+      "step": 911
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019863538140157534,
+      "loss": 1.6406,
+      "step": 912
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019863063338212104,
+      "loss": 1.582,
+      "step": 913
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001986258771738804,
+      "loss": 1.6445,
+      "step": 914
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001986211127772483,
+      "loss": 1.6543,
+      "step": 915
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019861634019262024,
+      "loss": 1.7363,
+      "step": 916
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019861155942039257,
+      "loss": 1.7754,
+      "step": 917
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019860677046096217,
+      "loss": 1.7227,
+      "step": 918
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019860197331472663,
+      "loss": 1.5625,
+      "step": 919
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001985971679820842,
+      "loss": 1.8379,
+      "step": 920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001985923544634339,
+      "loss": 1.7852,
+      "step": 921
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019858753275917534,
+      "loss": 1.6719,
+      "step": 922
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001985827028697089,
+      "loss": 1.6074,
+      "step": 923
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019857786479543548,
+      "loss": 1.7891,
+      "step": 924
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019857301853675683,
+      "loss": 1.7578,
+      "step": 925
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019856816409407527,
+      "loss": 1.7852,
+      "step": 926
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019856330146779387,
+      "loss": 1.6328,
+      "step": 927
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019855843065831632,
+      "loss": 1.7402,
+      "step": 928
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019855355166604706,
+      "loss": 1.7188,
+      "step": 929
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001985486644913911,
+      "loss": 1.8281,
+      "step": 930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001985437691347543,
+      "loss": 1.6973,
+      "step": 931
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00198538865596543,
+      "loss": 1.7461,
+      "step": 932
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019853395387716436,
+      "loss": 1.7578,
+      "step": 933
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019852903397702616,
+      "loss": 1.6367,
+      "step": 934
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001985241058965369,
+      "loss": 1.7051,
+      "step": 935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019851916963610564,
+      "loss": 1.7441,
+      "step": 936
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019851422519614236,
+      "loss": 1.8184,
+      "step": 937
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019850927257705746,
+      "loss": 1.7695,
+      "step": 938
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019850431177926217,
+      "loss": 1.6855,
+      "step": 939
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019849934280316834,
+      "loss": 1.7422,
+      "step": 940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019849436564918855,
+      "loss": 1.6641,
+      "step": 941
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00198489380317736,
+      "loss": 1.8047,
+      "step": 942
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001984843868092246,
+      "loss": 1.584,
+      "step": 943
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019847938512406894,
+      "loss": 1.6816,
+      "step": 944
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001984743752626843,
+      "loss": 1.8281,
+      "step": 945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001984693572254866,
+      "loss": 1.7617,
+      "step": 946
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019846433101289246,
+      "loss": 1.6797,
+      "step": 947
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001984592966253192,
+      "loss": 1.7344,
+      "step": 948
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001984542540631847,
+      "loss": 1.5918,
+      "step": 949
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001984492033269078,
+      "loss": 1.7793,
+      "step": 950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019844414441690774,
+      "loss": 1.6211,
+      "step": 951
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001984390773336045,
+      "loss": 1.6367,
+      "step": 952
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001984340020774188,
+      "loss": 1.7266,
+      "step": 953
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019842891864877204,
+      "loss": 1.6777,
+      "step": 954
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0019842382704808627,
+      "loss": 1.7266,
+      "step": 955
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019841872727578416,
+      "loss": 1.7637,
+      "step": 956
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001984136193322892,
+      "loss": 1.6543,
+      "step": 957
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001984085032180254,
+      "loss": 1.6992,
+      "step": 958
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019840337893341753,
+      "loss": 1.7617,
+      "step": 959
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019839824647889114,
+      "loss": 1.7578,
+      "step": 960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019839310585487223,
+      "loss": 1.5879,
+      "step": 961
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019838795706178763,
+      "loss": 1.7129,
+      "step": 962
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019838280010006483,
+      "loss": 1.6152,
+      "step": 963
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00198377634970132,
+      "loss": 1.8027,
+      "step": 964
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001983724616724179,
+      "loss": 1.7363,
+      "step": 965
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001983672802073522,
+      "loss": 1.7832,
+      "step": 966
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019836209057536493,
+      "loss": 1.8223,
+      "step": 967
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00198356892776887,
+      "loss": 1.6777,
+      "step": 968
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019835168681235,
+      "loss": 1.6738,
+      "step": 969
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019834647268218613,
+      "loss": 1.6387,
+      "step": 970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019834125038682825,
+      "loss": 1.7012,
+      "step": 971
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019833601992671,
+      "loss": 1.6934,
+      "step": 972
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019833078130226563,
+      "loss": 1.6836,
+      "step": 973
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019832553451393006,
+      "loss": 1.6699,
+      "step": 974
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019832027956213885,
+      "loss": 1.6504,
+      "step": 975
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019831501644732843,
+      "loss": 1.7012,
+      "step": 976
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019830974516993564,
+      "loss": 1.8672,
+      "step": 977
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001983044657303982,
+      "loss": 1.7031,
+      "step": 978
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001982991781291544,
+      "loss": 1.7012,
+      "step": 979
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019829388236664324,
+      "loss": 1.834,
+      "step": 980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001982885784433044,
+      "loss": 1.5859,
+      "step": 981
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019828326635957824,
+      "loss": 1.7344,
+      "step": 982
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001982779461159058,
+      "loss": 1.6934,
+      "step": 983
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019827261771272883,
+      "loss": 1.7207,
+      "step": 984
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019826728115048963,
+      "loss": 1.8027,
+      "step": 985
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019826193642963133,
+      "loss": 1.7949,
+      "step": 986
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001982565835505977,
+      "loss": 1.6953,
+      "step": 987
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001982512225138331,
+      "loss": 1.6836,
+      "step": 988
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019824585331978265,
+      "loss": 1.6074,
+      "step": 989
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019824047596889213,
+      "loss": 1.7383,
+      "step": 990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00198235090461608,
+      "loss": 1.7031,
+      "step": 991
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019822969679837736,
+      "loss": 1.9316,
+      "step": 992
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019822429497964806,
+      "loss": 1.7578,
+      "step": 993
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019821888500586855,
+      "loss": 1.7168,
+      "step": 994
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00198213466877488,
+      "loss": 1.6992,
+      "step": 995
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001982080405949563,
+      "loss": 1.6914,
+      "step": 996
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001982026061587239,
+      "loss": 1.6211,
+      "step": 997
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019819716356924196,
+      "loss": 1.6348,
+      "step": 998
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001981917128269625,
+      "loss": 1.7832,
+      "step": 999
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001981862539323379,
+      "loss": 1.7773,
+      "step": 1000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001981807868858215,
+      "loss": 1.7734,
+      "step": 1001
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001981753116878671,
+      "loss": 1.7617,
+      "step": 1002
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001981698283389294,
+      "loss": 1.7539,
+      "step": 1003
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019816433683946355,
+      "loss": 1.6484,
+      "step": 1004
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001981588371899255,
+      "loss": 1.6211,
+      "step": 1005
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001981533293907719,
+      "loss": 1.6719,
+      "step": 1006
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019814781344246,
+      "loss": 1.6562,
+      "step": 1007
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001981422893454478,
+      "loss": 1.6211,
+      "step": 1008
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019813675710019384,
+      "loss": 1.7344,
+      "step": 1009
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001981312167071575,
+      "loss": 1.6426,
+      "step": 1010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019812566816679884,
+      "loss": 1.668,
+      "step": 1011
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001981201114795784,
+      "loss": 1.7637,
+      "step": 1012
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001981145466459576,
+      "loss": 1.6816,
+      "step": 1013
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019810897366639843,
+      "loss": 1.7148,
+      "step": 1014
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001981033925413636,
+      "loss": 1.6934,
+      "step": 1015
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019809780327131646,
+      "loss": 1.7773,
+      "step": 1016
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001980922058567211,
+      "loss": 1.7266,
+      "step": 1017
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019808660029804214,
+      "loss": 1.6777,
+      "step": 1018
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001980809865957451,
+      "loss": 1.7109,
+      "step": 1019
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00198075364750296,
+      "loss": 1.709,
+      "step": 1020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019806973476216164,
+      "loss": 1.7539,
+      "step": 1021
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019806409663180938,
+      "loss": 1.6953,
+      "step": 1022
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001980584503597074,
+      "loss": 1.6211,
+      "step": 1023
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001980527959463244,
+      "loss": 1.6797,
+      "step": 1024
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019804713339212987,
+      "loss": 1.6094,
+      "step": 1025
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019804146269759397,
+      "loss": 1.6875,
+      "step": 1026
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019803578386318747,
+      "loss": 1.6504,
+      "step": 1027
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001980300968893819,
+      "loss": 1.7305,
+      "step": 1028
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019802440177664935,
+      "loss": 1.7246,
+      "step": 1029
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001980186985254627,
+      "loss": 1.75,
+      "step": 1030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019801298713629546,
+      "loss": 1.7715,
+      "step": 1031
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001980072676096218,
+      "loss": 1.7344,
+      "step": 1032
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001980015399459166,
+      "loss": 1.7461,
+      "step": 1033
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019799580414565543,
+      "loss": 1.7891,
+      "step": 1034
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001979900602093144,
+      "loss": 1.666,
+      "step": 1035
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001979843081373705,
+      "loss": 1.6758,
+      "step": 1036
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019797854793030126,
+      "loss": 1.7871,
+      "step": 1037
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001979727795885849,
+      "loss": 1.5508,
+      "step": 1038
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001979670031127004,
+      "loss": 1.8125,
+      "step": 1039
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001979612185031273,
+      "loss": 1.7344,
+      "step": 1040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019795542576034582,
+      "loss": 1.6348,
+      "step": 1041
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019794962488483697,
+      "loss": 1.6289,
+      "step": 1042
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019794381587708238,
+      "loss": 1.7598,
+      "step": 1043
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001979379987375643,
+      "loss": 1.8125,
+      "step": 1044
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019793217346676566,
+      "loss": 1.7266,
+      "step": 1045
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001979263400651702,
+      "loss": 1.6348,
+      "step": 1046
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019792049853326215,
+      "loss": 1.6895,
+      "step": 1047
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019791464887152655,
+      "loss": 1.7207,
+      "step": 1048
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001979087910804491,
+      "loss": 1.6836,
+      "step": 1049
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019790292516051605,
+      "loss": 1.709,
+      "step": 1050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019789705111221446,
+      "loss": 1.7461,
+      "step": 1051
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019789116893603204,
+      "loss": 1.6074,
+      "step": 1052
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001978852786324571,
+      "loss": 1.7031,
+      "step": 1053
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019787938020197878,
+      "loss": 1.793,
+      "step": 1054
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001978734736450867,
+      "loss": 1.5547,
+      "step": 1055
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001978675589622713,
+      "loss": 1.7383,
+      "step": 1056
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019786163615402364,
+      "loss": 1.5176,
+      "step": 1057
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019785570522083545,
+      "loss": 1.8633,
+      "step": 1058
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001978497661631992,
+      "loss": 1.6895,
+      "step": 1059
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019784381898160786,
+      "loss": 1.7441,
+      "step": 1060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019783786367655524,
+      "loss": 1.8242,
+      "step": 1061
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019783190024853584,
+      "loss": 1.6328,
+      "step": 1062
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019782592869804472,
+      "loss": 1.7598,
+      "step": 1063
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001978199490255777,
+      "loss": 1.6094,
+      "step": 1064
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001978139612316312,
+      "loss": 1.6953,
+      "step": 1065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019780796531670237,
+      "loss": 1.7461,
+      "step": 1066
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0019780196128128904,
+      "loss": 1.6289,
+      "step": 1067
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019779594912588965,
+      "loss": 1.7109,
+      "step": 1068
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001977899288510034,
+      "loss": 1.7129,
+      "step": 1069
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001977839004571301,
+      "loss": 1.6953,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019777786394477026,
+      "loss": 1.6816,
+      "step": 1071
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001977718193144251,
+      "loss": 1.7012,
+      "step": 1072
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019776576656659636,
+      "loss": 1.5195,
+      "step": 1073
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019775970570178668,
+      "loss": 1.6465,
+      "step": 1074
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019775363672049923,
+      "loss": 1.6973,
+      "step": 1075
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019774755962323792,
+      "loss": 1.7715,
+      "step": 1076
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001977414744105072,
+      "loss": 1.8105,
+      "step": 1077
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001977353810828124,
+      "loss": 1.5996,
+      "step": 1078
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019772927964065933,
+      "loss": 1.6484,
+      "step": 1079
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019772317008455464,
+      "loss": 1.7637,
+      "step": 1080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019771705241500553,
+      "loss": 1.8125,
+      "step": 1081
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001977109266325199,
+      "loss": 1.7773,
+      "step": 1082
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019770479273760644,
+      "loss": 1.7832,
+      "step": 1083
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019769865073077427,
+      "loss": 1.7578,
+      "step": 1084
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019769250061253342,
+      "loss": 1.5469,
+      "step": 1085
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001976863423833945,
+      "loss": 1.7578,
+      "step": 1086
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019768017604386875,
+      "loss": 1.7031,
+      "step": 1087
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001976740015944682,
+      "loss": 1.6465,
+      "step": 1088
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019766781903570543,
+      "loss": 1.7012,
+      "step": 1089
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019766162836809373,
+      "loss": 1.6836,
+      "step": 1090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001976554295921471,
+      "loss": 1.6719,
+      "step": 1091
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019764922270838025,
+      "loss": 1.7676,
+      "step": 1092
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019764300771730843,
+      "loss": 1.7461,
+      "step": 1093
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001976367846194476,
+      "loss": 1.7598,
+      "step": 1094
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019763055341531456,
+      "loss": 1.6797,
+      "step": 1095
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001976243141054266,
+      "loss": 1.707,
+      "step": 1096
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001976180666903017,
+      "loss": 1.7695,
+      "step": 1097
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001976118111704585,
+      "loss": 1.6934,
+      "step": 1098
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019760554754641654,
+      "loss": 1.6973,
+      "step": 1099
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001975992758186957,
+      "loss": 1.5938,
+      "step": 1100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001975929959878168,
+      "loss": 1.6875,
+      "step": 1101
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001975867080543011,
+      "loss": 1.7715,
+      "step": 1102
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001975804120186707,
+      "loss": 1.6758,
+      "step": 1103
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001975741078814484,
+      "loss": 1.7461,
+      "step": 1104
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019756779564315753,
+      "loss": 1.6953,
+      "step": 1105
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019756147530432216,
+      "loss": 1.6211,
+      "step": 1106
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001975551468654671,
+      "loss": 1.6348,
+      "step": 1107
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001975488103271177,
+      "loss": 1.7891,
+      "step": 1108
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019754246568980003,
+      "loss": 1.6699,
+      "step": 1109
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001975361129540409,
+      "loss": 1.6953,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019752975212036774,
+      "loss": 1.6328,
+      "step": 1111
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001975233831893087,
+      "loss": 1.75,
+      "step": 1112
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019751700616139245,
+      "loss": 1.7422,
+      "step": 1113
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001975106210371485,
+      "loss": 1.7207,
+      "step": 1114
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00197504227817107,
+      "loss": 1.6543,
+      "step": 1115
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019749782650179872,
+      "loss": 1.7109,
+      "step": 1116
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019749141709175513,
+      "loss": 1.6406,
+      "step": 1117
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019748499958750837,
+      "loss": 1.6777,
+      "step": 1118
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019747857398959125,
+      "loss": 1.7988,
+      "step": 1119
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019747214029853725,
+      "loss": 1.752,
+      "step": 1120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001974656985148805,
+      "loss": 1.7051,
+      "step": 1121
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001974592486391559,
+      "loss": 1.6602,
+      "step": 1122
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019745279067189887,
+      "loss": 1.6719,
+      "step": 1123
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019744632461364564,
+      "loss": 1.7344,
+      "step": 1124
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00197439850464933,
+      "loss": 1.7207,
+      "step": 1125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019743336822629857,
+      "loss": 1.748,
+      "step": 1126
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019742687789828044,
+      "loss": 1.7578,
+      "step": 1127
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019742037948141744,
+      "loss": 1.6348,
+      "step": 1128
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019741387297624916,
+      "loss": 1.6328,
+      "step": 1129
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019740735838331587,
+      "loss": 1.7578,
+      "step": 1130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001974008357031583,
+      "loss": 1.7461,
+      "step": 1131
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019739430493631803,
+      "loss": 1.7871,
+      "step": 1132
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019738776608333734,
+      "loss": 1.752,
+      "step": 1133
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001973812191447591,
+      "loss": 1.6211,
+      "step": 1134
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019737466412112683,
+      "loss": 1.8281,
+      "step": 1135
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019736810101298476,
+      "loss": 1.6934,
+      "step": 1136
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001973615298208778,
+      "loss": 1.6562,
+      "step": 1137
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019735495054535156,
+      "loss": 1.6562,
+      "step": 1138
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019734836318695227,
+      "loss": 1.6055,
+      "step": 1139
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001973417677462268,
+      "loss": 1.7305,
+      "step": 1140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019733516422372277,
+      "loss": 1.7754,
+      "step": 1141
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019732855261998844,
+      "loss": 1.8477,
+      "step": 1142
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001973219329355727,
+      "loss": 1.666,
+      "step": 1143
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019731530517102516,
+      "loss": 1.5293,
+      "step": 1144
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001973086693268961,
+      "loss": 1.6855,
+      "step": 1145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001973020254037365,
+      "loss": 1.7031,
+      "step": 1146
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019729537340209794,
+      "loss": 1.75,
+      "step": 1147
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019728871332253266,
+      "loss": 1.6797,
+      "step": 1148
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019728204516559367,
+      "loss": 1.6562,
+      "step": 1149
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019727536893183453,
+      "loss": 1.7207,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001972686846218096,
+      "loss": 1.832,
+      "step": 1151
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001972619922360738,
+      "loss": 1.7891,
+      "step": 1152
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001972552917751828,
+      "loss": 1.5117,
+      "step": 1153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019724858323969282,
+      "loss": 1.7246,
+      "step": 1154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019724186663016095,
+      "loss": 1.7559,
+      "step": 1155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019723514194714474,
+      "loss": 1.7734,
+      "step": 1156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019722840919120257,
+      "loss": 1.7637,
+      "step": 1157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001972216683628934,
+      "loss": 1.7715,
+      "step": 1158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019721491946277686,
+      "loss": 1.6953,
+      "step": 1159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019720816249141336,
+      "loss": 1.625,
+      "step": 1160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001972013974493638,
+      "loss": 1.7383,
+      "step": 1161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019719462433718986,
+      "loss": 1.6211,
+      "step": 1162
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019718784315545393,
+      "loss": 1.6367,
+      "step": 1163
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019718105390471898,
+      "loss": 1.8809,
+      "step": 1164
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001971742565855487,
+      "loss": 1.5742,
+      "step": 1165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001971674511985074,
+      "loss": 1.7227,
+      "step": 1166
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001971606377441602,
+      "loss": 1.5957,
+      "step": 1167
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019715381622307264,
+      "loss": 1.6367,
+      "step": 1168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001971469866358112,
+      "loss": 1.75,
+      "step": 1169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001971401489829428,
+      "loss": 1.7285,
+      "step": 1170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001971333032650352,
+      "loss": 1.7129,
+      "step": 1171
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019712644948265677,
+      "loss": 1.7852,
+      "step": 1172
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019711958763637648,
+      "loss": 1.6504,
+      "step": 1173
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001971127177267641,
+      "loss": 1.6426,
+      "step": 1174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019710583975438996,
+      "loss": 1.8242,
+      "step": 1175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019709895371982514,
+      "loss": 1.6348,
+      "step": 1176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019709205962364137,
+      "loss": 1.6738,
+      "step": 1177
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019708515746641093,
+      "loss": 1.7188,
+      "step": 1178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00197078247248707,
+      "loss": 1.7324,
+      "step": 1179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0019707132897110314,
+      "loss": 1.7441,
+      "step": 1180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019706440263417392,
+      "loss": 1.5859,
+      "step": 1181
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019705746823849425,
+      "loss": 1.6895,
+      "step": 1182
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019705052578463996,
+      "loss": 1.7227,
+      "step": 1183
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019704357527318738,
+      "loss": 1.6504,
+      "step": 1184
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001970366167047136,
+      "loss": 1.6445,
+      "step": 1185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019702965007979633,
+      "loss": 1.627,
+      "step": 1186
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00197022675399014,
+      "loss": 1.7188,
+      "step": 1187
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001970156926629457,
+      "loss": 1.8477,
+      "step": 1188
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001970087018721711,
+      "loss": 1.8516,
+      "step": 1189
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019700170302727068,
+      "loss": 1.7715,
+      "step": 1190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001969946961288255,
+      "loss": 1.6973,
+      "step": 1191
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019698768117741724,
+      "loss": 1.6934,
+      "step": 1192
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019698065817362843,
+      "loss": 1.6719,
+      "step": 1193
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019697362711804207,
+      "loss": 1.7168,
+      "step": 1194
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019696658801124194,
+      "loss": 1.7598,
+      "step": 1195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019695954085381245,
+      "loss": 1.5449,
+      "step": 1196
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001969524856463387,
+      "loss": 1.7773,
+      "step": 1197
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019694542238940646,
+      "loss": 1.7168,
+      "step": 1198
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019693835108360214,
+      "loss": 1.6582,
+      "step": 1199
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019693127172951278,
+      "loss": 1.6797,
+      "step": 1200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001969241843277262,
+      "loss": 1.6484,
+      "step": 1201
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019691708887883083,
+      "loss": 1.6797,
+      "step": 1202
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001969099853834158,
+      "loss": 1.5957,
+      "step": 1203
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001969028738420708,
+      "loss": 1.6641,
+      "step": 1204
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019689575425538633,
+      "loss": 1.6758,
+      "step": 1205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019688862662395343,
+      "loss": 1.6934,
+      "step": 1206
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001968814909483639,
+      "loss": 1.6133,
+      "step": 1207
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019687434722921024,
+      "loss": 1.7578,
+      "step": 1208
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001968671954670854,
+      "loss": 1.666,
+      "step": 1209
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001968600356625833,
+      "loss": 1.7539,
+      "step": 1210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019685286781629834,
+      "loss": 1.6309,
+      "step": 1211
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019684569192882566,
+      "loss": 1.6387,
+      "step": 1212
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019683850800076095,
+      "loss": 1.6797,
+      "step": 1213
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019683131603270066,
+      "loss": 1.7461,
+      "step": 1214
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019682411602524198,
+      "loss": 1.6953,
+      "step": 1215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019681690797898267,
+      "loss": 1.6465,
+      "step": 1216
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019680969189452115,
+      "loss": 1.6816,
+      "step": 1217
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019680246777245653,
+      "loss": 1.707,
+      "step": 1218
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001967952356133886,
+      "loss": 1.7578,
+      "step": 1219
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001967879954179178,
+      "loss": 1.8828,
+      "step": 1220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001967807471866453,
+      "loss": 1.6016,
+      "step": 1221
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001967734909201728,
+      "loss": 1.7305,
+      "step": 1222
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001967662266191028,
+      "loss": 1.6797,
+      "step": 1223
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019675895428403842,
+      "loss": 1.7305,
+      "step": 1224
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019675167391558336,
+      "loss": 1.6621,
+      "step": 1225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019674438551434224,
+      "loss": 1.7441,
+      "step": 1226
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019673708908092002,
+      "loss": 1.7754,
+      "step": 1227
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019672978461592254,
+      "loss": 1.7109,
+      "step": 1228
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001967224721199563,
+      "loss": 1.6562,
+      "step": 1229
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001967151515936283,
+      "loss": 1.5645,
+      "step": 1230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019670782303754645,
+      "loss": 1.5996,
+      "step": 1231
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019670048645231914,
+      "loss": 1.6973,
+      "step": 1232
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019669314183855553,
+      "loss": 1.6367,
+      "step": 1233
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019668578919686533,
+      "loss": 1.7969,
+      "step": 1234
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019667842852785907,
+      "loss": 1.6738,
+      "step": 1235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001966710598321478,
+      "loss": 1.6855,
+      "step": 1236
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019666368311034334,
+      "loss": 1.7324,
+      "step": 1237
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019665629836305818,
+      "loss": 1.625,
+      "step": 1238
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019664890559090535,
+      "loss": 1.8184,
+      "step": 1239
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019664150479449865,
+      "loss": 1.5449,
+      "step": 1240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019663409597445265,
+      "loss": 1.5801,
+      "step": 1241
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001966266791313823,
+      "loss": 1.7188,
+      "step": 1242
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019661925426590347,
+      "loss": 1.6523,
+      "step": 1243
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001966118213786326,
+      "loss": 1.6621,
+      "step": 1244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001966043804701868,
+      "loss": 1.627,
+      "step": 1245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019659693154118383,
+      "loss": 1.7129,
+      "step": 1246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019658947459224216,
+      "loss": 1.7988,
+      "step": 1247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001965820096239809,
+      "loss": 1.7344,
+      "step": 1248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001965745366370198,
+      "loss": 1.6426,
+      "step": 1249
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019656705563197937,
+      "loss": 1.7754,
+      "step": 1250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019655956660948066,
+      "loss": 1.7539,
+      "step": 1251
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001965520695701454,
+      "loss": 1.7695,
+      "step": 1252
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019654456451459615,
+      "loss": 1.7695,
+      "step": 1253
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019653705144345593,
+      "loss": 1.7578,
+      "step": 1254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019652953035734855,
+      "loss": 1.8418,
+      "step": 1255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001965220012568984,
+      "loss": 1.7344,
+      "step": 1256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019651446414273065,
+      "loss": 1.7832,
+      "step": 1257
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00196506919015471,
+      "loss": 1.6621,
+      "step": 1258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001964993658757459,
+      "loss": 1.666,
+      "step": 1259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001964918047241825,
+      "loss": 1.8926,
+      "step": 1260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019648423556140847,
+      "loss": 1.7168,
+      "step": 1261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019647665838805235,
+      "loss": 1.7715,
+      "step": 1262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001964690732047431,
+      "loss": 1.6367,
+      "step": 1263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019646148001211065,
+      "loss": 1.7227,
+      "step": 1264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019645387881078525,
+      "loss": 1.6758,
+      "step": 1265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001964462696013981,
+      "loss": 1.6465,
+      "step": 1266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019643865238458086,
+      "loss": 1.5703,
+      "step": 1267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001964310271609661,
+      "loss": 1.7598,
+      "step": 1268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001964233939311867,
+      "loss": 1.7266,
+      "step": 1269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001964157526958766,
+      "loss": 1.6797,
+      "step": 1270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019640810345567007,
+      "loss": 1.6133,
+      "step": 1271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019640044621120224,
+      "loss": 1.6465,
+      "step": 1272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001963927809631089,
+      "loss": 1.625,
+      "step": 1273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019638510771202637,
+      "loss": 1.5566,
+      "step": 1274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019637742645859176,
+      "loss": 1.6777,
+      "step": 1275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019636973720344284,
+      "loss": 1.5898,
+      "step": 1276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019636203994721794,
+      "loss": 1.6797,
+      "step": 1277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019635433469055615,
+      "loss": 1.6875,
+      "step": 1278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001963466214340972,
+      "loss": 1.7266,
+      "step": 1279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019633890017848148,
+      "loss": 1.6758,
+      "step": 1280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001963311709243501,
+      "loss": 1.7949,
+      "step": 1281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019632343367234468,
+      "loss": 1.793,
+      "step": 1282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019631568842310766,
+      "loss": 1.623,
+      "step": 1283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001963079351772821,
+      "loss": 1.5859,
+      "step": 1284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019630017393551164,
+      "loss": 1.7812,
+      "step": 1285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019629240469844073,
+      "loss": 1.6562,
+      "step": 1286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001962846274667144,
+      "loss": 1.6914,
+      "step": 1287
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001962768422409783,
+      "loss": 1.6016,
+      "step": 1288
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019626904902187887,
+      "loss": 1.7266,
+      "step": 1289
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001962612478100631,
+      "loss": 1.6602,
+      "step": 1290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001962534386061787,
+      "loss": 1.7031,
+      "step": 1291
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0019624562141087403,
+      "loss": 1.7031,
+      "step": 1292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019623779622479805,
+      "loss": 1.6836,
+      "step": 1293
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019622996304860052,
+      "loss": 1.7031,
+      "step": 1294
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001962221218829317,
+      "loss": 1.623,
+      "step": 1295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019621427272844276,
+      "loss": 1.7227,
+      "step": 1296
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001962064155857852,
+      "loss": 1.6289,
+      "step": 1297
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019619855045561146,
+      "loss": 1.6074,
+      "step": 1298
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001961906773385745,
+      "loss": 1.6953,
+      "step": 1299
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00196182796235328,
+      "loss": 1.6523,
+      "step": 1300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019617490714652634,
+      "loss": 1.6719,
+      "step": 1301
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001961670100728244,
+      "loss": 1.7344,
+      "step": 1302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019615910501487787,
+      "loss": 1.6719,
+      "step": 1303
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001961511919733431,
+      "loss": 1.7793,
+      "step": 1304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019614327094887706,
+      "loss": 1.6992,
+      "step": 1305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019613534194213734,
+      "loss": 1.6758,
+      "step": 1306
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019612740495378233,
+      "loss": 1.707,
+      "step": 1307
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00196119459984471,
+      "loss": 1.7441,
+      "step": 1308
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019611150703486282,
+      "loss": 1.6074,
+      "step": 1309
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019610354610561827,
+      "loss": 1.6211,
+      "step": 1310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001960955771973982,
+      "loss": 1.7734,
+      "step": 1311
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019608760031086425,
+      "loss": 1.7227,
+      "step": 1312
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019607961544667873,
+      "loss": 1.5996,
+      "step": 1313
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019607162260550453,
+      "loss": 1.6738,
+      "step": 1314
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001960636217880053,
+      "loss": 1.5801,
+      "step": 1315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019605561299484526,
+      "loss": 1.8379,
+      "step": 1316
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019604759622668935,
+      "loss": 1.6914,
+      "step": 1317
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001960395714842032,
+      "loss": 1.6797,
+      "step": 1318
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019603153876805303,
+      "loss": 1.793,
+      "step": 1319
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001960234980789058,
+      "loss": 1.6602,
+      "step": 1320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00196015449417429,
+      "loss": 1.7734,
+      "step": 1321
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001960073927842909,
+      "loss": 1.6523,
+      "step": 1322
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019599932818016048,
+      "loss": 1.5859,
+      "step": 1323
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001959912556057072,
+      "loss": 1.6953,
+      "step": 1324
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019598317506160133,
+      "loss": 1.6289,
+      "step": 1325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019597508654851377,
+      "loss": 1.6543,
+      "step": 1326
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00195966990067116,
+      "loss": 1.5977,
+      "step": 1327
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001959588856180803,
+      "loss": 1.707,
+      "step": 1328
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019595077320207948,
+      "loss": 1.6309,
+      "step": 1329
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001959426528197871,
+      "loss": 1.7656,
+      "step": 1330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019593452447187742,
+      "loss": 1.6055,
+      "step": 1331
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001959263881590252,
+      "loss": 1.748,
+      "step": 1332
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019591824388190595,
+      "loss": 1.666,
+      "step": 1333
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001959100916411959,
+      "loss": 1.5996,
+      "step": 1334
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001959019314375719,
+      "loss": 1.6973,
+      "step": 1335
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001958937632717114,
+      "loss": 1.7109,
+      "step": 1336
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001958855871442926,
+      "loss": 1.709,
+      "step": 1337
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019587740305599425,
+      "loss": 1.6289,
+      "step": 1338
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019586921100749597,
+      "loss": 1.7695,
+      "step": 1339
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019586101099947776,
+      "loss": 1.6738,
+      "step": 1340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001958528030326205,
+      "loss": 1.7129,
+      "step": 1341
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001958445871076056,
+      "loss": 1.6309,
+      "step": 1342
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019583636322511528,
+      "loss": 1.6758,
+      "step": 1343
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001958281313858322,
+      "loss": 1.7129,
+      "step": 1344
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001958198915904399,
+      "loss": 1.623,
+      "step": 1345
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019581164383962245,
+      "loss": 1.6875,
+      "step": 1346
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001958033881340647,
+      "loss": 1.7578,
+      "step": 1347
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019579512447445192,
+      "loss": 1.7969,
+      "step": 1348
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001957868528614703,
+      "loss": 1.7754,
+      "step": 1349
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019577857329580663,
+      "loss": 1.541,
+      "step": 1350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001957702857781482,
+      "loss": 1.7773,
+      "step": 1351
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019576199030918316,
+      "loss": 1.8203,
+      "step": 1352
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019575368688960026,
+      "loss": 1.7734,
+      "step": 1353
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001957453755200888,
+      "loss": 1.6211,
+      "step": 1354
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001957370562013389,
+      "loss": 1.7148,
+      "step": 1355
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019572872893404126,
+      "loss": 1.6465,
+      "step": 1356
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019572039371888723,
+      "loss": 1.7676,
+      "step": 1357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019571205055656884,
+      "loss": 1.6426,
+      "step": 1358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001957036994477788,
+      "loss": 1.5352,
+      "step": 1359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019569534039321043,
+      "loss": 1.7148,
+      "step": 1360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001956869733935578,
+      "loss": 1.7148,
+      "step": 1361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019567859844951546,
+      "loss": 1.6602,
+      "step": 1362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001956702155617789,
+      "loss": 1.6191,
+      "step": 1363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019566182473104393,
+      "loss": 1.8203,
+      "step": 1364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001956534259580074,
+      "loss": 1.709,
+      "step": 1365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001956450192433664,
+      "loss": 1.6523,
+      "step": 1366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019563660458781903,
+      "loss": 1.6074,
+      "step": 1367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001956281819920639,
+      "loss": 1.7793,
+      "step": 1368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019561975145680026,
+      "loss": 1.7109,
+      "step": 1369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019561131298272806,
+      "loss": 1.6113,
+      "step": 1370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019560286657054797,
+      "loss": 1.668,
+      "step": 1371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001955944122209611,
+      "loss": 1.7246,
+      "step": 1372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019558594993466954,
+      "loss": 1.668,
+      "step": 1373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019557747971237576,
+      "loss": 1.7988,
+      "step": 1374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019556900155478308,
+      "loss": 1.6367,
+      "step": 1375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019556051546259527,
+      "loss": 1.5684,
+      "step": 1376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019555202143651707,
+      "loss": 1.6523,
+      "step": 1377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001955435194772535,
+      "loss": 1.6934,
+      "step": 1378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019553500958551053,
+      "loss": 1.7305,
+      "step": 1379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019552649176199474,
+      "loss": 1.7031,
+      "step": 1380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001955179660074132,
+      "loss": 1.7012,
+      "step": 1381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019550943232247387,
+      "loss": 1.7188,
+      "step": 1382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019550089070788524,
+      "loss": 1.6621,
+      "step": 1383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019549234116435638,
+      "loss": 1.7344,
+      "step": 1384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001954837836925972,
+      "loss": 1.7051,
+      "step": 1385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001954752182933182,
+      "loss": 1.666,
+      "step": 1386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001954666449672305,
+      "loss": 1.707,
+      "step": 1387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019545806371504583,
+      "loss": 1.7852,
+      "step": 1388
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001954494745374767,
+      "loss": 1.7324,
+      "step": 1389
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001954408774352362,
+      "loss": 1.8223,
+      "step": 1390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001954322724090382,
+      "loss": 1.7227,
+      "step": 1391
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00195423659459597,
+      "loss": 1.7129,
+      "step": 1392
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001954150385876278,
+      "loss": 1.7344,
+      "step": 1393
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019540640979384626,
+      "loss": 1.5195,
+      "step": 1394
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001953977730789688,
+      "loss": 1.6836,
+      "step": 1395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001953891284437125,
+      "loss": 1.6816,
+      "step": 1396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019538047588879507,
+      "loss": 1.6172,
+      "step": 1397
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019537181541493492,
+      "loss": 1.6914,
+      "step": 1398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00195363147022851,
+      "loss": 1.7363,
+      "step": 1399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019535447071326313,
+      "loss": 1.627,
+      "step": 1400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019534578648689152,
+      "loss": 1.7461,
+      "step": 1401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001953370943444573,
+      "loss": 1.5645,
+      "step": 1402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019532839428668207,
+      "loss": 1.7324,
+      "step": 1403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001953196863142881,
+      "loss": 1.5293,
+      "step": 1404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0019531097042799846,
+      "loss": 1.6992,
+      "step": 1405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019530224662853675,
+      "loss": 1.625,
+      "step": 1406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019529351491662723,
+      "loss": 1.6758,
+      "step": 1407
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019528477529299488,
+      "loss": 1.7031,
+      "step": 1408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019527602775836535,
+      "loss": 1.7871,
+      "step": 1409
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001952672723134648,
+      "loss": 1.832,
+      "step": 1410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019525850895902024,
+      "loss": 1.6113,
+      "step": 1411
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001952497376957592,
+      "loss": 1.8066,
+      "step": 1412
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019524095852440988,
+      "loss": 1.5859,
+      "step": 1413
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019523217144570128,
+      "loss": 1.6895,
+      "step": 1414
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001952233764603628,
+      "loss": 1.6895,
+      "step": 1415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019521457356912477,
+      "loss": 1.7109,
+      "step": 1416
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019520576277271798,
+      "loss": 1.6055,
+      "step": 1417
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019519694407187393,
+      "loss": 1.6973,
+      "step": 1418
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019518811746732485,
+      "loss": 1.7578,
+      "step": 1419
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019517928295980354,
+      "loss": 1.6836,
+      "step": 1420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001951704405500435,
+      "loss": 1.6445,
+      "step": 1421
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001951615902387788,
+      "loss": 1.6387,
+      "step": 1422
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001951527320267443,
+      "loss": 1.7441,
+      "step": 1423
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019514386591467545,
+      "loss": 1.7012,
+      "step": 1424
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019513499190330832,
+      "loss": 1.6953,
+      "step": 1425
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001951261099933797,
+      "loss": 1.6348,
+      "step": 1426
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019511722018562701,
+      "loss": 1.6719,
+      "step": 1427
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019510832248078832,
+      "loss": 1.8047,
+      "step": 1428
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019509941687960235,
+      "loss": 1.6523,
+      "step": 1429
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019509050338280848,
+      "loss": 1.6992,
+      "step": 1430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001950815819911468,
+      "loss": 1.6738,
+      "step": 1431
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019507265270535794,
+      "loss": 1.666,
+      "step": 1432
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001950637155261833,
+      "loss": 1.8105,
+      "step": 1433
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019505477045436486,
+      "loss": 1.5684,
+      "step": 1434
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001950458174906453,
+      "loss": 1.6641,
+      "step": 1435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019503685663576791,
+      "loss": 1.5898,
+      "step": 1436
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019502788789047673,
+      "loss": 1.6602,
+      "step": 1437
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001950189112555163,
+      "loss": 1.6504,
+      "step": 1438
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019500992673163193,
+      "loss": 1.6816,
+      "step": 1439
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001950009343195696,
+      "loss": 1.7871,
+      "step": 1440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019499193402007587,
+      "loss": 1.7168,
+      "step": 1441
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019498292583389801,
+      "loss": 1.6523,
+      "step": 1442
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019497390976178388,
+      "loss": 1.7812,
+      "step": 1443
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019496488580448207,
+      "loss": 1.6016,
+      "step": 1444
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001949558539627418,
+      "loss": 1.627,
+      "step": 1445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019494681423731291,
+      "loss": 1.6699,
+      "step": 1446
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019493776662894593,
+      "loss": 1.6816,
+      "step": 1447
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019492871113839204,
+      "loss": 1.666,
+      "step": 1448
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019491964776640308,
+      "loss": 1.625,
+      "step": 1449
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019491057651373153,
+      "loss": 1.7324,
+      "step": 1450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001949014973811305,
+      "loss": 1.6719,
+      "step": 1451
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001948924103693538,
+      "loss": 1.6816,
+      "step": 1452
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019488331547915592,
+      "loss": 1.7031,
+      "step": 1453
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019487421271129188,
+      "loss": 1.6523,
+      "step": 1454
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001948651020665175,
+      "loss": 1.4961,
+      "step": 1455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019485598354558916,
+      "loss": 1.6113,
+      "step": 1456
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019484685714926395,
+      "loss": 1.5957,
+      "step": 1457
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019483772287829954,
+      "loss": 1.6641,
+      "step": 1458
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019482858073345435,
+      "loss": 1.7422,
+      "step": 1459
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001948194307154874,
+      "loss": 1.6777,
+      "step": 1460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019481027282515831,
+      "loss": 1.7402,
+      "step": 1461
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001948011070632275,
+      "loss": 1.7832,
+      "step": 1462
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019479193343045584,
+      "loss": 1.7266,
+      "step": 1463
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001947827519276051,
+      "loss": 1.7812,
+      "step": 1464
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019477356255543747,
+      "loss": 1.6367,
+      "step": 1465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019476436531471598,
+      "loss": 1.7031,
+      "step": 1466
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019475516020620416,
+      "loss": 1.7656,
+      "step": 1467
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001947459472306663,
+      "loss": 1.7031,
+      "step": 1468
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019473672638886726,
+      "loss": 1.6172,
+      "step": 1469
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019472749768157265,
+      "loss": 1.627,
+      "step": 1470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019471826110954867,
+      "loss": 1.6211,
+      "step": 1471
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019470901667356215,
+      "loss": 1.6387,
+      "step": 1472
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019469976437438063,
+      "loss": 1.6602,
+      "step": 1473
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019469050421277233,
+      "loss": 1.6855,
+      "step": 1474
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00194681236189506,
+      "loss": 1.7344,
+      "step": 1475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019467196030535114,
+      "loss": 1.6445,
+      "step": 1476
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019466267656107787,
+      "loss": 1.6836,
+      "step": 1477
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019465338495745698,
+      "loss": 1.7324,
+      "step": 1478
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019464408549525994,
+      "loss": 1.7012,
+      "step": 1479
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019463477817525873,
+      "loss": 1.7109,
+      "step": 1480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019462546299822622,
+      "loss": 1.7285,
+      "step": 1481
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019461613996493573,
+      "loss": 1.6387,
+      "step": 1482
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019460680907616129,
+      "loss": 1.7227,
+      "step": 1483
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019459747033267762,
+      "loss": 1.6934,
+      "step": 1484
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019458812373526007,
+      "loss": 1.6113,
+      "step": 1485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019457876928468463,
+      "loss": 1.6875,
+      "step": 1486
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019456940698172795,
+      "loss": 1.7363,
+      "step": 1487
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019456003682716734,
+      "loss": 1.7168,
+      "step": 1488
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019455065882178079,
+      "loss": 1.7324,
+      "step": 1489
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001945412729663468,
+      "loss": 1.7129,
+      "step": 1490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019453187926164474,
+      "loss": 1.6836,
+      "step": 1491
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019452247770845447,
+      "loss": 1.5508,
+      "step": 1492
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019451306830755655,
+      "loss": 1.6738,
+      "step": 1493
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001945036510597322,
+      "loss": 1.6035,
+      "step": 1494
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001944942259657633,
+      "loss": 1.8027,
+      "step": 1495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019448479302643237,
+      "loss": 1.5293,
+      "step": 1496
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001944753522425225,
+      "loss": 1.6543,
+      "step": 1497
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001944659036148176,
+      "loss": 1.6758,
+      "step": 1498
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019445644714410216,
+      "loss": 1.6309,
+      "step": 1499
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019444698283116121,
+      "loss": 1.5605,
+      "step": 1500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019443751067678055,
+      "loss": 1.7344,
+      "step": 1501
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019442803068174662,
+      "loss": 1.6309,
+      "step": 1502
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019441854284684653,
+      "loss": 1.6035,
+      "step": 1503
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019440904717286794,
+      "loss": 1.6992,
+      "step": 1504
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019439954366059925,
+      "loss": 1.8281,
+      "step": 1505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001943900323108295,
+      "loss": 1.5645,
+      "step": 1506
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019438051312434834,
+      "loss": 1.5898,
+      "step": 1507
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019437098610194615,
+      "loss": 1.8359,
+      "step": 1508
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019436145124441385,
+      "loss": 1.7617,
+      "step": 1509
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001943519085525431,
+      "loss": 1.7188,
+      "step": 1510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001943423580271262,
+      "loss": 1.6367,
+      "step": 1511
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019433279966895602,
+      "loss": 1.6738,
+      "step": 1512
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019432323347882618,
+      "loss": 1.6465,
+      "step": 1513
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019431365945753094,
+      "loss": 1.6191,
+      "step": 1514
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001943040776058651,
+      "loss": 1.6855,
+      "step": 1515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0019429448792462427,
+      "loss": 1.793,
+      "step": 1516
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001942848904146046,
+      "loss": 1.5898,
+      "step": 1517
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019427528507660293,
+      "loss": 1.668,
+      "step": 1518
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019426567191141672,
+      "loss": 1.6934,
+      "step": 1519
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001942560509198441,
+      "loss": 1.6875,
+      "step": 1520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019424642210268388,
+      "loss": 1.6465,
+      "step": 1521
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019423678546073547,
+      "loss": 1.7344,
+      "step": 1522
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019422714099479894,
+      "loss": 1.7715,
+      "step": 1523
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019421748870567504,
+      "loss": 1.7051,
+      "step": 1524
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019420782859416517,
+      "loss": 1.6797,
+      "step": 1525
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001941981606610713,
+      "loss": 1.5957,
+      "step": 1526
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019418848490719617,
+      "loss": 1.7012,
+      "step": 1527
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019417880133334303,
+      "loss": 1.6875,
+      "step": 1528
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019416910994031593,
+      "loss": 1.6426,
+      "step": 1529
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019415941072891947,
+      "loss": 1.6543,
+      "step": 1530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019414970369995893,
+      "loss": 1.6855,
+      "step": 1531
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019413998885424023,
+      "loss": 1.6738,
+      "step": 1532
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019413026619256994,
+      "loss": 1.793,
+      "step": 1533
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019412053571575525,
+      "loss": 1.6133,
+      "step": 1534
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001941107974246041,
+      "loss": 1.6758,
+      "step": 1535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019410105131992498,
+      "loss": 1.6562,
+      "step": 1536
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019409129740252703,
+      "loss": 1.6465,
+      "step": 1537
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019408153567322007,
+      "loss": 1.6289,
+      "step": 1538
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019407176613281461,
+      "loss": 1.7344,
+      "step": 1539
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019406198878212173,
+      "loss": 1.6523,
+      "step": 1540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001940522036219532,
+      "loss": 1.6953,
+      "step": 1541
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019404241065312142,
+      "loss": 1.623,
+      "step": 1542
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019403260987643947,
+      "loss": 1.6758,
+      "step": 1543
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019402280129272105,
+      "loss": 1.6328,
+      "step": 1544
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019401298490278052,
+      "loss": 1.6152,
+      "step": 1545
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019400316070743286,
+      "loss": 1.584,
+      "step": 1546
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019399332870749374,
+      "loss": 1.6152,
+      "step": 1547
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019398348890377944,
+      "loss": 1.6641,
+      "step": 1548
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019397364129710695,
+      "loss": 1.6289,
+      "step": 1549
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019396378588829384,
+      "loss": 1.5391,
+      "step": 1550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019395392267815832,
+      "loss": 1.502,
+      "step": 1551
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019394405166751937,
+      "loss": 1.6465,
+      "step": 1552
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019393417285719643,
+      "loss": 1.7363,
+      "step": 1553
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019392428624800972,
+      "loss": 1.627,
+      "step": 1554
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001939143918407801,
+      "loss": 1.623,
+      "step": 1555
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019390448963632902,
+      "loss": 1.5645,
+      "step": 1556
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019389457963547864,
+      "loss": 1.6797,
+      "step": 1557
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019388466183905173,
+      "loss": 1.6094,
+      "step": 1558
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019387473624787167,
+      "loss": 1.623,
+      "step": 1559
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019386480286276257,
+      "loss": 1.6445,
+      "step": 1560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019385486168454914,
+      "loss": 1.7637,
+      "step": 1561
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019384491271405672,
+      "loss": 1.7285,
+      "step": 1562
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019383495595211135,
+      "loss": 1.7402,
+      "step": 1563
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019382499139953971,
+      "loss": 1.6797,
+      "step": 1564
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019381501905716904,
+      "loss": 1.707,
+      "step": 1565
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019380503892582732,
+      "loss": 1.6699,
+      "step": 1566
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019379505100634317,
+      "loss": 1.6504,
+      "step": 1567
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019378505529954585,
+      "loss": 1.7637,
+      "step": 1568
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019377505180626518,
+      "loss": 1.6465,
+      "step": 1569
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019376504052733175,
+      "loss": 1.7266,
+      "step": 1570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019375502146357674,
+      "loss": 1.707,
+      "step": 1571
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019374499461583194,
+      "loss": 1.6465,
+      "step": 1572
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001937349599849299,
+      "loss": 1.6113,
+      "step": 1573
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019372491757170367,
+      "loss": 1.707,
+      "step": 1574
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019371486737698707,
+      "loss": 1.6191,
+      "step": 1575
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001937048094016145,
+      "loss": 1.6309,
+      "step": 1576
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00193694743646421,
+      "loss": 1.6543,
+      "step": 1577
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019368467011224228,
+      "loss": 1.709,
+      "step": 1578
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019367458879991477,
+      "loss": 1.6797,
+      "step": 1579
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019366449971027536,
+      "loss": 1.6523,
+      "step": 1580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019365440284416177,
+      "loss": 1.7676,
+      "step": 1581
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001936442982024122,
+      "loss": 1.7051,
+      "step": 1582
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019363418578586572,
+      "loss": 1.7383,
+      "step": 1583
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001936240655953618,
+      "loss": 1.7207,
+      "step": 1584
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019361393763174068,
+      "loss": 1.6816,
+      "step": 1585
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001936038018958433,
+      "loss": 1.5742,
+      "step": 1586
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019359365838851113,
+      "loss": 1.6719,
+      "step": 1587
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019358350711058633,
+      "loss": 1.6777,
+      "step": 1588
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019357334806291168,
+      "loss": 1.7422,
+      "step": 1589
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019356318124633073,
+      "loss": 1.9023,
+      "step": 1590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019355300666168748,
+      "loss": 1.6074,
+      "step": 1591
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001935428243098267,
+      "loss": 1.7305,
+      "step": 1592
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019353263419159379,
+      "loss": 1.7344,
+      "step": 1593
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019352243630783477,
+      "loss": 1.8516,
+      "step": 1594
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019351223065939634,
+      "loss": 1.7344,
+      "step": 1595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001935020172471258,
+      "loss": 1.6406,
+      "step": 1596
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019349179607187112,
+      "loss": 1.666,
+      "step": 1597
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001934815671344809,
+      "loss": 1.707,
+      "step": 1598
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001934713304358044,
+      "loss": 1.5449,
+      "step": 1599
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019346108597669157,
+      "loss": 1.6445,
+      "step": 1600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019345083375799286,
+      "loss": 1.7031,
+      "step": 1601
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019344057378055954,
+      "loss": 1.5938,
+      "step": 1602
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019343030604524341,
+      "loss": 1.6055,
+      "step": 1603
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019342003055289694,
+      "loss": 1.6367,
+      "step": 1604
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019340974730437327,
+      "loss": 1.6621,
+      "step": 1605
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019339945630052614,
+      "loss": 1.6895,
+      "step": 1606
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019338915754220998,
+      "loss": 1.748,
+      "step": 1607
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019337885103027982,
+      "loss": 1.5957,
+      "step": 1608
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019336853676559138,
+      "loss": 1.6602,
+      "step": 1609
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019335821474900097,
+      "loss": 1.7754,
+      "step": 1610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001933478849813656,
+      "loss": 1.6289,
+      "step": 1611
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001933375474635429,
+      "loss": 1.6387,
+      "step": 1612
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019332720219639114,
+      "loss": 1.5645,
+      "step": 1613
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019331684918076919,
+      "loss": 1.6523,
+      "step": 1614
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019330648841753666,
+      "loss": 1.582,
+      "step": 1615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001932961199075537,
+      "loss": 1.7168,
+      "step": 1616
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019328574365168123,
+      "loss": 1.7344,
+      "step": 1617
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019327535965078066,
+      "loss": 1.6816,
+      "step": 1618
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019326496790571415,
+      "loss": 1.7051,
+      "step": 1619
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019325456841734448,
+      "loss": 1.7617,
+      "step": 1620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019324416118653503,
+      "loss": 1.7012,
+      "step": 1621
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001932337462141499,
+      "loss": 1.5938,
+      "step": 1622
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019322332350105377,
+      "loss": 1.7266,
+      "step": 1623
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00193212893048112,
+      "loss": 1.6836,
+      "step": 1624
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019320245485619054,
+      "loss": 1.7266,
+      "step": 1625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019319200892615608,
+      "loss": 1.5859,
+      "step": 1626
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001931815552588758,
+      "loss": 1.6582,
+      "step": 1627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001931710938552177,
+      "loss": 1.6152,
+      "step": 1628
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019316062471605032,
+      "loss": 1.623,
+      "step": 1629
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019315014784224282,
+      "loss": 1.5391,
+      "step": 1630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019313966323466505,
+      "loss": 1.6641,
+      "step": 1631
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019312917089418752,
+      "loss": 1.623,
+      "step": 1632
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001931186708216813,
+      "loss": 1.7266,
+      "step": 1633
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019310816301801824,
+      "loss": 1.6719,
+      "step": 1634
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019309764748407067,
+      "loss": 1.5762,
+      "step": 1635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019308712422071168,
+      "loss": 1.6992,
+      "step": 1636
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019307659322881498,
+      "loss": 1.623,
+      "step": 1637
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019306605450925482,
+      "loss": 1.7422,
+      "step": 1638
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019305550806290627,
+      "loss": 1.6582,
+      "step": 1639
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001930449538906449,
+      "loss": 1.6055,
+      "step": 1640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019303439199334696,
+      "loss": 1.7031,
+      "step": 1641
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019302382237188938,
+      "loss": 1.5547,
+      "step": 1642
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001930132450271497,
+      "loss": 1.75,
+      "step": 1643
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019300265996000606,
+      "loss": 1.7402,
+      "step": 1644
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001929920671713373,
+      "loss": 1.6562,
+      "step": 1645
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019298146666202293,
+      "loss": 1.6641,
+      "step": 1646
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019297085843294297,
+      "loss": 1.6602,
+      "step": 1647
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019296024248497826,
+      "loss": 1.7422,
+      "step": 1648
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019294961881901013,
+      "loss": 1.7402,
+      "step": 1649
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019293898743592058,
+      "loss": 1.7793,
+      "step": 1650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019292834833659234,
+      "loss": 1.5938,
+      "step": 1651
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001929177015219087,
+      "loss": 1.752,
+      "step": 1652
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019290704699275359,
+      "loss": 1.7363,
+      "step": 1653
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019289638475001164,
+      "loss": 1.7363,
+      "step": 1654
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019288571479456804,
+      "loss": 1.8223,
+      "step": 1655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019287503712730864,
+      "loss": 1.6895,
+      "step": 1656
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019286435174912006,
+      "loss": 1.75,
+      "step": 1657
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019285365866088934,
+      "loss": 1.7383,
+      "step": 1658
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001928429578635043,
+      "loss": 1.6504,
+      "step": 1659
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019283224935785338,
+      "loss": 1.623,
+      "step": 1660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019282153314482565,
+      "loss": 1.6953,
+      "step": 1661
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019281080922531081,
+      "loss": 1.6094,
+      "step": 1662
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019280007760019925,
+      "loss": 1.6328,
+      "step": 1663
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019278933827038193,
+      "loss": 1.6504,
+      "step": 1664
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019277859123675047,
+      "loss": 1.6875,
+      "step": 1665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019276783650019717,
+      "loss": 1.5781,
+      "step": 1666
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019275707406161491,
+      "loss": 1.6914,
+      "step": 1667
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019274630392189728,
+      "loss": 1.6582,
+      "step": 1668
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019273552608193843,
+      "loss": 1.6699,
+      "step": 1669
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019272474054263318,
+      "loss": 1.7012,
+      "step": 1670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019271394730487703,
+      "loss": 1.7383,
+      "step": 1671
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019270314636956607,
+      "loss": 1.7305,
+      "step": 1672
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019269233773759704,
+      "loss": 1.6641,
+      "step": 1673
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019268152140986734,
+      "loss": 1.6953,
+      "step": 1674
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00192670697387275,
+      "loss": 1.8652,
+      "step": 1675
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019265986567071867,
+      "loss": 1.5918,
+      "step": 1676
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019264902626109763,
+      "loss": 1.5918,
+      "step": 1677
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019263817915931183,
+      "loss": 1.6973,
+      "step": 1678
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019262732436626188,
+      "loss": 1.6055,
+      "step": 1679
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019261646188284893,
+      "loss": 1.6738,
+      "step": 1680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019260559170997493,
+      "loss": 1.5273,
+      "step": 1681
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019259471384854231,
+      "loss": 1.6445,
+      "step": 1682
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019258382829945417,
+      "loss": 1.625,
+      "step": 1683
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019257293506361438,
+      "loss": 1.627,
+      "step": 1684
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019256203414192725,
+      "loss": 1.7363,
+      "step": 1685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001925511255352979,
+      "loss": 1.707,
+      "step": 1686
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019254020924463196,
+      "loss": 1.8086,
+      "step": 1687
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001925292852708358,
+      "loss": 1.5938,
+      "step": 1688
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019251835361481634,
+      "loss": 1.8379,
+      "step": 1689
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019250741427748119,
+      "loss": 1.6172,
+      "step": 1690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019249646725973861,
+      "loss": 1.6895,
+      "step": 1691
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019248551256249745,
+      "loss": 1.6855,
+      "step": 1692
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019247455018666722,
+      "loss": 1.668,
+      "step": 1693
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019246358013315808,
+      "loss": 1.6602,
+      "step": 1694
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001924526024028808,
+      "loss": 1.627,
+      "step": 1695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019244161699674685,
+      "loss": 1.7383,
+      "step": 1696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001924306239156682,
+      "loss": 1.6934,
+      "step": 1697
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019241962316055763,
+      "loss": 1.625,
+      "step": 1698
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019240861473232846,
+      "loss": 1.5723,
+      "step": 1699
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019239759863189465,
+      "loss": 1.6152,
+      "step": 1700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019238657486017078,
+      "loss": 1.4922,
+      "step": 1701
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019237554341807218,
+      "loss": 1.7031,
+      "step": 1702
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019236450430651464,
+      "loss": 1.6309,
+      "step": 1703
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019235345752641473,
+      "loss": 1.6504,
+      "step": 1704
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001923424030786896,
+      "loss": 1.6777,
+      "step": 1705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019233134096425704,
+      "loss": 1.7539,
+      "step": 1706
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019232027118403546,
+      "loss": 1.7168,
+      "step": 1707
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019230919373894398,
+      "loss": 1.6289,
+      "step": 1708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019229810862990222,
+      "loss": 1.5977,
+      "step": 1709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019228701585783062,
+      "loss": 1.625,
+      "step": 1710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019227591542365005,
+      "loss": 1.668,
+      "step": 1711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001922648073282822,
+      "loss": 1.6562,
+      "step": 1712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001922536915726493,
+      "loss": 1.6348,
+      "step": 1713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019224256815767418,
+      "loss": 1.668,
+      "step": 1714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019223143708428044,
+      "loss": 1.6836,
+      "step": 1715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019222029835339219,
+      "loss": 1.6289,
+      "step": 1716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019220915196593422,
+      "loss": 1.6816,
+      "step": 1717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019219799792283196,
+      "loss": 1.6953,
+      "step": 1718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019218683622501146,
+      "loss": 1.6426,
+      "step": 1719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019217566687339948,
+      "loss": 1.7422,
+      "step": 1720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019216448986892325,
+      "loss": 1.6895,
+      "step": 1721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019215330521251085,
+      "loss": 1.7109,
+      "step": 1722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019214211290509077,
+      "loss": 1.6934,
+      "step": 1723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019213091294759232,
+      "loss": 1.498,
+      "step": 1724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001921197053409454,
+      "loss": 1.6777,
+      "step": 1725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019210849008608046,
+      "loss": 1.7051,
+      "step": 1726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019209726718392864,
+      "loss": 1.6562,
+      "step": 1727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019208603663542175,
+      "loss": 1.7617,
+      "step": 1728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001920747984414922,
+      "loss": 1.6504,
+      "step": 1729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019206355260307304,
+      "loss": 1.6094,
+      "step": 1730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019205229912109796,
+      "loss": 1.6387,
+      "step": 1731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019204103799650124,
+      "loss": 1.5449,
+      "step": 1732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001920297692302179,
+      "loss": 1.6836,
+      "step": 1733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019201849282318342,
+      "loss": 1.6504,
+      "step": 1734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019200720877633414,
+      "loss": 1.6113,
+      "step": 1735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001919959170906068,
+      "loss": 1.6406,
+      "step": 1736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00191984617766939,
+      "loss": 1.6934,
+      "step": 1737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019197331080626882,
+      "loss": 1.6895,
+      "step": 1738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00191961996209535,
+      "loss": 1.6953,
+      "step": 1739
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019195067397767694,
+      "loss": 1.6992,
+      "step": 1740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001919393441116347,
+      "loss": 1.8359,
+      "step": 1741
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019192800661234886,
+      "loss": 1.6348,
+      "step": 1742
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019191666148076082,
+      "loss": 1.5527,
+      "step": 1743
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001919053087178124,
+      "loss": 1.6816,
+      "step": 1744
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019189394832444622,
+      "loss": 1.8086,
+      "step": 1745
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019188258030160548,
+      "loss": 1.7812,
+      "step": 1746
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019187120465023399,
+      "loss": 1.7754,
+      "step": 1747
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001918598213712762,
+      "loss": 1.6445,
+      "step": 1748
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019184843046567723,
+      "loss": 1.6953,
+      "step": 1749
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001918370319343828,
+      "loss": 1.5996,
+      "step": 1750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019182562577833926,
+      "loss": 1.6113,
+      "step": 1751
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019181421199849363,
+      "loss": 1.6855,
+      "step": 1752
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019180279059579349,
+      "loss": 1.6582,
+      "step": 1753
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019179136157118713,
+      "loss": 1.6719,
+      "step": 1754
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019177992492562343,
+      "loss": 1.6289,
+      "step": 1755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019176848066005192,
+      "loss": 1.5586,
+      "step": 1756
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019175702877542277,
+      "loss": 1.5176,
+      "step": 1757
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019174556927268678,
+      "loss": 1.5566,
+      "step": 1758
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019173410215279532,
+      "loss": 1.5762,
+      "step": 1759
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001917226274167005,
+      "loss": 1.6777,
+      "step": 1760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019171114506535496,
+      "loss": 1.623,
+      "step": 1761
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019169965509971206,
+      "loss": 1.582,
+      "step": 1762
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019168815752072575,
+      "loss": 1.5156,
+      "step": 1763
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019167665232935058,
+      "loss": 1.7031,
+      "step": 1764
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019166513952654182,
+      "loss": 1.668,
+      "step": 1765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019165361911325522,
+      "loss": 1.666,
+      "step": 1766
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019164209109044739,
+      "loss": 1.6797,
+      "step": 1767
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019163055545907534,
+      "loss": 1.7031,
+      "step": 1768
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019161901222009689,
+      "loss": 1.6953,
+      "step": 1769
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001916074613744703,
+      "loss": 1.748,
+      "step": 1770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019159590292315472,
+      "loss": 1.6797,
+      "step": 1771
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019158433686710969,
+      "loss": 1.6172,
+      "step": 1772
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001915727632072955,
+      "loss": 1.791,
+      "step": 1773
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019156118194467307,
+      "loss": 1.6426,
+      "step": 1774
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019154959308020393,
+      "loss": 1.5879,
+      "step": 1775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019153799661485022,
+      "loss": 1.6875,
+      "step": 1776
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019152639254957473,
+      "loss": 1.6211,
+      "step": 1777
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019151478088534092,
+      "loss": 1.6895,
+      "step": 1778
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019150316162311284,
+      "loss": 1.627,
+      "step": 1779
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019149153476385515,
+      "loss": 1.6738,
+      "step": 1780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019147990030853315,
+      "loss": 1.6758,
+      "step": 1781
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019146825825811288,
+      "loss": 1.6719,
+      "step": 1782
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019145660861356083,
+      "loss": 1.7051,
+      "step": 1783
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019144495137584424,
+      "loss": 1.6582,
+      "step": 1784
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001914332865459309,
+      "loss": 1.7188,
+      "step": 1785
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001914216141247894,
+      "loss": 1.4785,
+      "step": 1786
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019140993411338872,
+      "loss": 1.627,
+      "step": 1787
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019139824651269867,
+      "loss": 1.5645,
+      "step": 1788
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019138655132368953,
+      "loss": 1.6309,
+      "step": 1789
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019137484854733237,
+      "loss": 1.625,
+      "step": 1790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019136313818459878,
+      "loss": 1.7637,
+      "step": 1791
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019135142023646102,
+      "loss": 1.5918,
+      "step": 1792
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019133969470389194,
+      "loss": 1.6973,
+      "step": 1793
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019132796158786505,
+      "loss": 1.6035,
+      "step": 1794
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019131622088935454,
+      "loss": 1.6641,
+      "step": 1795
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019130447260933515,
+      "loss": 1.6797,
+      "step": 1796
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019129271674878227,
+      "loss": 1.627,
+      "step": 1797
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001912809533086719,
+      "loss": 1.7109,
+      "step": 1798
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019126918228998075,
+      "loss": 1.6602,
+      "step": 1799
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001912574036936861,
+      "loss": 1.6445,
+      "step": 1800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019124561752076584,
+      "loss": 1.7363,
+      "step": 1801
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019123382377219853,
+      "loss": 1.6836,
+      "step": 1802
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019122202244896331,
+      "loss": 1.6875,
+      "step": 1803
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019121021355204004,
+      "loss": 1.6309,
+      "step": 1804
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019119839708240912,
+      "loss": 1.6289,
+      "step": 1805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019118657304105164,
+      "loss": 1.8438,
+      "step": 1806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019117474142894921,
+      "loss": 1.6426,
+      "step": 1807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019116290224708422,
+      "loss": 1.5605,
+      "step": 1808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019115105549643958,
+      "loss": 1.7363,
+      "step": 1809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019113920117799887,
+      "loss": 1.7363,
+      "step": 1810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019112733929274636,
+      "loss": 1.7266,
+      "step": 1811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019111546984166674,
+      "loss": 1.6328,
+      "step": 1812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019110359282574559,
+      "loss": 1.6738,
+      "step": 1813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019109170824596895,
+      "loss": 1.6289,
+      "step": 1814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019107981610332354,
+      "loss": 1.6934,
+      "step": 1815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019106791639879668,
+      "loss": 1.6562,
+      "step": 1816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001910560091333764,
+      "loss": 1.7656,
+      "step": 1817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019104409430805123,
+      "loss": 1.7461,
+      "step": 1818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019103217192381044,
+      "loss": 1.6777,
+      "step": 1819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019102024198164385,
+      "loss": 1.7031,
+      "step": 1820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019100830448254196,
+      "loss": 1.6445,
+      "step": 1821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019099635942749589,
+      "loss": 1.6777,
+      "step": 1822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019098440681749739,
+      "loss": 1.668,
+      "step": 1823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019097244665353874,
+      "loss": 1.6133,
+      "step": 1824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019096047893661303,
+      "loss": 1.7676,
+      "step": 1825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019094850366771384,
+      "loss": 1.7344,
+      "step": 1826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019093652084783534,
+      "loss": 1.582,
+      "step": 1827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019092453047797253,
+      "loss": 1.6816,
+      "step": 1828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019091253255912083,
+      "loss": 1.6797,
+      "step": 1829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001909005270922764,
+      "loss": 1.6289,
+      "step": 1830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019088851407843596,
+      "loss": 1.584,
+      "step": 1831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019087649351859687,
+      "loss": 1.8027,
+      "step": 1832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001908644654137572,
+      "loss": 1.7148,
+      "step": 1833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001908524297649155,
+      "loss": 1.6934,
+      "step": 1834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001908403865730711,
+      "loss": 1.7324,
+      "step": 1835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019082833583922386,
+      "loss": 1.6211,
+      "step": 1836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019081627756437427,
+      "loss": 1.709,
+      "step": 1837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019080421174952349,
+      "loss": 1.666,
+      "step": 1838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019079213839567326,
+      "loss": 1.6953,
+      "step": 1839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019078005750382601,
+      "loss": 1.6191,
+      "step": 1840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019076796907498469,
+      "loss": 1.7305,
+      "step": 1841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00190755873110153,
+      "loss": 1.7461,
+      "step": 1842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019074376961033513,
+      "loss": 1.6738,
+      "step": 1843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019073165857653605,
+      "loss": 1.5918,
+      "step": 1844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019071954000976124,
+      "loss": 1.6562,
+      "step": 1845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019070741391101684,
+      "loss": 1.752,
+      "step": 1846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019069528028130961,
+      "loss": 1.623,
+      "step": 1847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019068313912164697,
+      "loss": 1.6309,
+      "step": 1848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001906709904330369,
+      "loss": 1.7441,
+      "step": 1849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001906588342164881,
+      "loss": 1.5938,
+      "step": 1850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019064667047300975,
+      "loss": 1.6328,
+      "step": 1851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019063449920361183,
+      "loss": 1.6562,
+      "step": 1852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019062232040930479,
+      "loss": 1.5957,
+      "step": 1853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001906101340910998,
+      "loss": 1.6504,
+      "step": 1854
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019059794025000866,
+      "loss": 1.6699,
+      "step": 1855
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019058573888704369,
+      "loss": 1.6875,
+      "step": 1856
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019057353000321796,
+      "loss": 1.6758,
+      "step": 1857
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019056131359954508,
+      "loss": 1.7246,
+      "step": 1858
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019054908967703933,
+      "loss": 1.752,
+      "step": 1859
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001905368582367156,
+      "loss": 1.6719,
+      "step": 1860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001905246192795894,
+      "loss": 1.8105,
+      "step": 1861
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019051237280667684,
+      "loss": 1.6562,
+      "step": 1862
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019050011881899473,
+      "loss": 1.7422,
+      "step": 1863
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001904878573175604,
+      "loss": 1.6777,
+      "step": 1864
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019047558830339192,
+      "loss": 1.6016,
+      "step": 1865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019046331177750787,
+      "loss": 1.5312,
+      "step": 1866
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019045102774092753,
+      "loss": 1.6699,
+      "step": 1867
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001904387361946708,
+      "loss": 1.6484,
+      "step": 1868
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019042643713975812,
+      "loss": 1.7246,
+      "step": 1869
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019041413057721069,
+      "loss": 1.666,
+      "step": 1870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001904018165080502,
+      "loss": 1.7324,
+      "step": 1871
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019038949493329904,
+      "loss": 1.709,
+      "step": 1872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019037716585398022,
+      "loss": 1.7227,
+      "step": 1873
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019036482927111736,
+      "loss": 1.7012,
+      "step": 1874
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001903524851857347,
+      "loss": 1.6094,
+      "step": 1875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019034013359885708,
+      "loss": 1.6855,
+      "step": 1876
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019032777451151,
+      "loss": 1.623,
+      "step": 1877
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019031540792471963,
+      "loss": 1.6504,
+      "step": 1878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001903030338395126,
+      "loss": 1.666,
+      "step": 1879
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001902906522569163,
+      "loss": 1.5625,
+      "step": 1880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019027826317795876,
+      "loss": 1.6719,
+      "step": 1881
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019026586660366854,
+      "loss": 1.543,
+      "step": 1882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019025346253507486,
+      "loss": 1.7148,
+      "step": 1883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019024105097320756,
+      "loss": 1.5801,
+      "step": 1884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019022863191909714,
+      "loss": 1.6914,
+      "step": 1885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019021620537377467,
+      "loss": 1.6836,
+      "step": 1886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019020377133827184,
+      "loss": 1.6406,
+      "step": 1887
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019019132981362102,
+      "loss": 1.623,
+      "step": 1888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019017888080085515,
+      "loss": 1.7441,
+      "step": 1889
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019016642430100777,
+      "loss": 1.6699,
+      "step": 1890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019015396031511313,
+      "loss": 1.7598,
+      "step": 1891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019014148884420605,
+      "loss": 1.5488,
+      "step": 1892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019012900988932193,
+      "loss": 1.6523,
+      "step": 1893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019011652345149689,
+      "loss": 1.6602,
+      "step": 1894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019010402953176756,
+      "loss": 1.7266,
+      "step": 1895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019009152813117126,
+      "loss": 1.7266,
+      "step": 1896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001900790192507459,
+      "loss": 1.5664,
+      "step": 1897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019006650289153009,
+      "loss": 1.6914,
+      "step": 1898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001900539790545629,
+      "loss": 1.7266,
+      "step": 1899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019004144774088423,
+      "loss": 1.6406,
+      "step": 1900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001900289089515344,
+      "loss": 1.6445,
+      "step": 1901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019001636268755448,
+      "loss": 1.6445,
+      "step": 1902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019000380894998612,
+      "loss": 1.8223,
+      "step": 1903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018999124773987157,
+      "loss": 1.5645,
+      "step": 1904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018997867905825375,
+      "loss": 1.6875,
+      "step": 1905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018996610290617614,
+      "loss": 1.5742,
+      "step": 1906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018995351928468294,
+      "loss": 1.6523,
+      "step": 1907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001899409281948188,
+      "loss": 1.5684,
+      "step": 1908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018992832963762917,
+      "loss": 1.5762,
+      "step": 1909
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018991572361416003,
+      "loss": 1.5898,
+      "step": 1910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018990311012545796,
+      "loss": 1.5781,
+      "step": 1911
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001898904891725702,
+      "loss": 1.7285,
+      "step": 1912
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018987786075654466,
+      "loss": 1.7285,
+      "step": 1913
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018986522487842974,
+      "loss": 1.7246,
+      "step": 1914
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018985258153927456,
+      "loss": 1.6816,
+      "step": 1915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018983993074012883,
+      "loss": 1.7617,
+      "step": 1916
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018982727248204286,
+      "loss": 1.5879,
+      "step": 1917
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018981460676606764,
+      "loss": 1.666,
+      "step": 1918
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001898019335932547,
+      "loss": 1.8477,
+      "step": 1919
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018978925296465623,
+      "loss": 1.5781,
+      "step": 1920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018977656488132506,
+      "loss": 1.6855,
+      "step": 1921
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001897638693443146,
+      "loss": 1.6328,
+      "step": 1922
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018975116635467893,
+      "loss": 1.6621,
+      "step": 1923
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018973845591347264,
+      "loss": 1.6504,
+      "step": 1924
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018972573802175105,
+      "loss": 1.7109,
+      "step": 1925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018971301268057008,
+      "loss": 1.6816,
+      "step": 1926
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018970027989098624,
+      "loss": 1.6348,
+      "step": 1927
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018968753965405666,
+      "loss": 1.6699,
+      "step": 1928
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018967479197083909,
+      "loss": 1.6562,
+      "step": 1929
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018966203684239189,
+      "loss": 1.5332,
+      "step": 1930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018964927426977406,
+      "loss": 1.7363,
+      "step": 1931
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018963650425404524,
+      "loss": 1.75,
+      "step": 1932
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018962372679626562,
+      "loss": 1.5781,
+      "step": 1933
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018961094189749608,
+      "loss": 1.8086,
+      "step": 1934
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018959814955879805,
+      "loss": 1.6836,
+      "step": 1935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018958534978123366,
+      "loss": 1.6328,
+      "step": 1936
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001895725425658655,
+      "loss": 1.6309,
+      "step": 1937
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018955972791375704,
+      "loss": 1.6367,
+      "step": 1938
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001895469058259721,
+      "loss": 1.666,
+      "step": 1939
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018953407630357526,
+      "loss": 1.6191,
+      "step": 1940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001895212393476317,
+      "loss": 1.666,
+      "step": 1941
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018950839495920722,
+      "loss": 1.7461,
+      "step": 1942
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018949554313936817,
+      "loss": 1.5508,
+      "step": 1943
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001894826838891816,
+      "loss": 1.6973,
+      "step": 1944
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018946981720971516,
+      "loss": 1.7266,
+      "step": 1945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001894569431020371,
+      "loss": 1.7031,
+      "step": 1946
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018944406156721626,
+      "loss": 1.5527,
+      "step": 1947
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018943117260632216,
+      "loss": 1.7227,
+      "step": 1948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018941827622042488,
+      "loss": 1.6445,
+      "step": 1949
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018940537241059518,
+      "loss": 1.6426,
+      "step": 1950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018939246117790432,
+      "loss": 1.6543,
+      "step": 1951
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018937954252342432,
+      "loss": 1.7461,
+      "step": 1952
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018936661644822772,
+      "loss": 1.6992,
+      "step": 1953
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018935368295338773,
+      "loss": 1.6074,
+      "step": 1954
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018934074203997813,
+      "loss": 1.6289,
+      "step": 1955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018932779370907334,
+      "loss": 1.5898,
+      "step": 1956
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018931483796174838,
+      "loss": 1.6582,
+      "step": 1957
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001893018747990789,
+      "loss": 1.6016,
+      "step": 1958
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018928890422214115,
+      "loss": 1.8301,
+      "step": 1959
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018927592623201206,
+      "loss": 1.6914,
+      "step": 1960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001892629408297691,
+      "loss": 1.6992,
+      "step": 1961
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001892499480164904,
+      "loss": 1.7188,
+      "step": 1962
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001892369477932546,
+      "loss": 1.748,
+      "step": 1963
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018922394016114112,
+      "loss": 1.7012,
+      "step": 1964
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018921092512122993,
+      "loss": 1.6582,
+      "step": 1965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018919790267460154,
+      "loss": 1.709,
+      "step": 1966
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0018918487282233718,
+      "loss": 1.7012,
+      "step": 1967
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018917183556551862,
+      "loss": 1.7266,
+      "step": 1968
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018915879090522832,
+      "loss": 1.6113,
+      "step": 1969
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018914573884254923,
+      "loss": 1.584,
+      "step": 1970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001891326793785651,
+      "loss": 1.5547,
+      "step": 1971
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001891196125143601,
+      "loss": 1.7285,
+      "step": 1972
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018910653825101914,
+      "loss": 1.623,
+      "step": 1973
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001890934565896277,
+      "loss": 1.6387,
+      "step": 1974
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001890803675312719,
+      "loss": 1.7344,
+      "step": 1975
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018906727107703843,
+      "loss": 1.6172,
+      "step": 1976
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018905416722801464,
+      "loss": 1.4727,
+      "step": 1977
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001890410559852885,
+      "loss": 1.6875,
+      "step": 1978
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001890279373499485,
+      "loss": 1.6309,
+      "step": 1979
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018901481132308385,
+      "loss": 1.7812,
+      "step": 1980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018900167790578432,
+      "loss": 1.6816,
+      "step": 1981
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018898853709914034,
+      "loss": 1.7383,
+      "step": 1982
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018897538890424294,
+      "loss": 1.7344,
+      "step": 1983
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018896223332218365,
+      "loss": 1.5703,
+      "step": 1984
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001889490703540548,
+      "loss": 1.7305,
+      "step": 1985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001889359000009492,
+      "loss": 1.7324,
+      "step": 1986
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018892272226396032,
+      "loss": 1.6973,
+      "step": 1987
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018890953714418225,
+      "loss": 1.7266,
+      "step": 1988
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018889634464270971,
+      "loss": 1.7129,
+      "step": 1989
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018888314476063796,
+      "loss": 1.627,
+      "step": 1990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018886993749906292,
+      "loss": 1.8145,
+      "step": 1991
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018885672285908113,
+      "loss": 1.6621,
+      "step": 1992
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018884350084178972,
+      "loss": 1.7773,
+      "step": 1993
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001888302714482865,
+      "loss": 1.7188,
+      "step": 1994
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018881703467966973,
+      "loss": 1.6328,
+      "step": 1995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018880379053703852,
+      "loss": 1.6055,
+      "step": 1996
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018879053902149236,
+      "loss": 1.7344,
+      "step": 1997
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018877728013413147,
+      "loss": 1.6367,
+      "step": 1998
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018876401387605674,
+      "loss": 1.7188,
+      "step": 1999
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001887507402483695,
+      "loss": 1.6055,
+      "step": 2000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018873745925217186,
+      "loss": 1.6172,
+      "step": 2001
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018872417088856644,
+      "loss": 1.6523,
+      "step": 2002
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001887108751586565,
+      "loss": 1.7227,
+      "step": 2003
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018869757206354594,
+      "loss": 1.5996,
+      "step": 2004
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018868426160433922,
+      "loss": 1.7246,
+      "step": 2005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018867094378214144,
+      "loss": 1.6777,
+      "step": 2006
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018865761859805836,
+      "loss": 1.584,
+      "step": 2007
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018864428605319624,
+      "loss": 1.6934,
+      "step": 2008
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018863094614866203,
+      "loss": 1.6406,
+      "step": 2009
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018861759888556327,
+      "loss": 1.748,
+      "step": 2010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001886042442650081,
+      "loss": 1.7363,
+      "step": 2011
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018859088228810532,
+      "loss": 1.6465,
+      "step": 2012
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018857751295596429,
+      "loss": 1.6914,
+      "step": 2013
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018856413626969497,
+      "loss": 1.6445,
+      "step": 2014
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00188550752230408,
+      "loss": 1.7168,
+      "step": 2015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018853736083921456,
+      "loss": 1.627,
+      "step": 2016
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018852396209722646,
+      "loss": 1.5645,
+      "step": 2017
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018851055600555618,
+      "loss": 1.7266,
+      "step": 2018
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018849714256531666,
+      "loss": 1.7578,
+      "step": 2019
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018848372177762166,
+      "loss": 1.6562,
+      "step": 2020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018847029364358536,
+      "loss": 1.6953,
+      "step": 2021
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018845685816432267,
+      "loss": 1.748,
+      "step": 2022
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018844341534094901,
+      "loss": 1.7715,
+      "step": 2023
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018842996517458054,
+      "loss": 1.707,
+      "step": 2024
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018841650766633392,
+      "loss": 1.7051,
+      "step": 2025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018840304281732645,
+      "loss": 1.7441,
+      "step": 2026
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018838957062867606,
+      "loss": 1.7754,
+      "step": 2027
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001883760911015013,
+      "loss": 1.6465,
+      "step": 2028
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001883626042369212,
+      "loss": 1.6602,
+      "step": 2029
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018834911003605566,
+      "loss": 1.7812,
+      "step": 2030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018833560850002491,
+      "loss": 1.6309,
+      "step": 2031
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018832209962995,
+      "loss": 1.5566,
+      "step": 2032
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001883085834269524,
+      "loss": 1.6855,
+      "step": 2033
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018829505989215439,
+      "loss": 1.6621,
+      "step": 2034
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001882815290266787,
+      "loss": 1.6504,
+      "step": 2035
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018826799083164873,
+      "loss": 1.7051,
+      "step": 2036
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018825444530818853,
+      "loss": 1.7031,
+      "step": 2037
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018824089245742263,
+      "loss": 1.6504,
+      "step": 2038
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018822733228047635,
+      "loss": 1.6562,
+      "step": 2039
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018821376477847545,
+      "loss": 1.6699,
+      "step": 2040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018820018995254642,
+      "loss": 1.5723,
+      "step": 2041
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018818660780381626,
+      "loss": 1.6738,
+      "step": 2042
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018817301833341264,
+      "loss": 1.6289,
+      "step": 2043
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018815942154246383,
+      "loss": 1.668,
+      "step": 2044
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018814581743209872,
+      "loss": 1.7852,
+      "step": 2045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018813220600344676,
+      "loss": 1.6074,
+      "step": 2046
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018811858725763804,
+      "loss": 1.5938,
+      "step": 2047
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018810496119580325,
+      "loss": 1.6152,
+      "step": 2048
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001880913278190737,
+      "loss": 1.6816,
+      "step": 2049
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018807768712858128,
+      "loss": 1.75,
+      "step": 2050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018806403912545852,
+      "loss": 1.6602,
+      "step": 2051
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018805038381083858,
+      "loss": 1.7246,
+      "step": 2052
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018803672118585513,
+      "loss": 1.6426,
+      "step": 2053
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018802305125164254,
+      "loss": 1.6074,
+      "step": 2054
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001880093740093357,
+      "loss": 1.666,
+      "step": 2055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018799568946007024,
+      "loss": 1.7852,
+      "step": 2056
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018798199760498226,
+      "loss": 1.6641,
+      "step": 2057
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018796829844520855,
+      "loss": 1.5957,
+      "step": 2058
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001879545919818865,
+      "loss": 1.7246,
+      "step": 2059
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018794087821615401,
+      "loss": 1.6387,
+      "step": 2060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018792715714914974,
+      "loss": 1.8047,
+      "step": 2061
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018791342878201284,
+      "loss": 1.6543,
+      "step": 2062
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001878996931158831,
+      "loss": 1.5527,
+      "step": 2063
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018788595015190093,
+      "loss": 1.6367,
+      "step": 2064
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018787219989120736,
+      "loss": 1.6914,
+      "step": 2065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00187858442334944,
+      "loss": 1.7129,
+      "step": 2066
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00187844677484253,
+      "loss": 1.6348,
+      "step": 2067
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018783090534027728,
+      "loss": 1.7383,
+      "step": 2068
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001878171259041602,
+      "loss": 1.5352,
+      "step": 2069
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018780333917704582,
+      "loss": 1.6211,
+      "step": 2070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018778954516007877,
+      "loss": 1.7227,
+      "step": 2071
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018777574385440432,
+      "loss": 1.666,
+      "step": 2072
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001877619352611683,
+      "loss": 1.5449,
+      "step": 2073
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018774811938151714,
+      "loss": 1.625,
+      "step": 2074
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018773429621659796,
+      "loss": 1.6348,
+      "step": 2075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018772046576755839,
+      "loss": 1.8379,
+      "step": 2076
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001877066280355467,
+      "loss": 1.627,
+      "step": 2077
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018769278302171177,
+      "loss": 1.6465,
+      "step": 2078
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0018767893072720308,
+      "loss": 1.7285,
+      "step": 2079
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018766507115317071,
+      "loss": 1.6875,
+      "step": 2080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018765120430076535,
+      "loss": 1.6133,
+      "step": 2081
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018763733017113828,
+      "loss": 1.6562,
+      "step": 2082
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018762344876544143,
+      "loss": 1.6348,
+      "step": 2083
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018760956008482725,
+      "loss": 1.5781,
+      "step": 2084
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018759566413044891,
+      "loss": 1.7285,
+      "step": 2085
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018758176090346006,
+      "loss": 1.6875,
+      "step": 2086
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018756785040501503,
+      "loss": 1.6973,
+      "step": 2087
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018755393263626874,
+      "loss": 1.668,
+      "step": 2088
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018754000759837672,
+      "loss": 1.7578,
+      "step": 2089
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001875260752924951,
+      "loss": 1.6992,
+      "step": 2090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018751213571978056,
+      "loss": 1.5996,
+      "step": 2091
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018749818888139048,
+      "loss": 1.7344,
+      "step": 2092
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018748423477848277,
+      "loss": 1.627,
+      "step": 2093
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018747027341221596,
+      "loss": 1.6445,
+      "step": 2094
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018745630478374919,
+      "loss": 1.6582,
+      "step": 2095
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018744232889424224,
+      "loss": 1.6074,
+      "step": 2096
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001874283457448554,
+      "loss": 1.6562,
+      "step": 2097
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018741435533674966,
+      "loss": 1.5078,
+      "step": 2098
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018740035767108654,
+      "loss": 1.7305,
+      "step": 2099
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018738635274902823,
+      "loss": 1.6719,
+      "step": 2100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018737234057173745,
+      "loss": 1.6543,
+      "step": 2101
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018735832114037756,
+      "loss": 1.6289,
+      "step": 2102
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018734429445611254,
+      "loss": 1.6875,
+      "step": 2103
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018733026052010695,
+      "loss": 1.6328,
+      "step": 2104
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018731621933352593,
+      "loss": 1.7422,
+      "step": 2105
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018730217089753527,
+      "loss": 1.7246,
+      "step": 2106
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018728811521330135,
+      "loss": 1.6328,
+      "step": 2107
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018727405228199112,
+      "loss": 1.6543,
+      "step": 2108
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018725998210477215,
+      "loss": 1.7754,
+      "step": 2109
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001872459046828126,
+      "loss": 1.6348,
+      "step": 2110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018723182001728128,
+      "loss": 1.6855,
+      "step": 2111
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018721772810934754,
+      "loss": 1.6152,
+      "step": 2112
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018720362896018138,
+      "loss": 1.752,
+      "step": 2113
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018718952257095335,
+      "loss": 1.623,
+      "step": 2114
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018717540894283465,
+      "loss": 1.7422,
+      "step": 2115
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018716128807699706,
+      "loss": 1.6641,
+      "step": 2116
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018714715997461293,
+      "loss": 1.6367,
+      "step": 2117
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001871330246368553,
+      "loss": 1.7344,
+      "step": 2118
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018711888206489774,
+      "loss": 1.7676,
+      "step": 2119
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018710473225991438,
+      "loss": 1.6445,
+      "step": 2120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018709057522308004,
+      "loss": 1.709,
+      "step": 2121
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018707641095557013,
+      "loss": 1.6797,
+      "step": 2122
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018706223945856061,
+      "loss": 1.7773,
+      "step": 2123
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018704806073322803,
+      "loss": 1.6133,
+      "step": 2124
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018703387478074965,
+      "loss": 1.6543,
+      "step": 2125
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018701968160230319,
+      "loss": 1.6992,
+      "step": 2126
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018700548119906708,
+      "loss": 1.627,
+      "step": 2127
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001869912735722203,
+      "loss": 1.6738,
+      "step": 2128
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018697705872294241,
+      "loss": 1.6543,
+      "step": 2129
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001869628366524136,
+      "loss": 1.6836,
+      "step": 2130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001869486073618147,
+      "loss": 1.625,
+      "step": 2131
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00186934370852327,
+      "loss": 1.666,
+      "step": 2132
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018692012712513257,
+      "loss": 1.625,
+      "step": 2133
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018690587618141396,
+      "loss": 1.7559,
+      "step": 2134
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018689161802235438,
+      "loss": 1.7891,
+      "step": 2135
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018687735264913755,
+      "loss": 1.5488,
+      "step": 2136
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018686308006294788,
+      "loss": 1.7168,
+      "step": 2137
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018684880026497036,
+      "loss": 1.6777,
+      "step": 2138
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018683451325639054,
+      "loss": 1.6191,
+      "step": 2139
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018682021903839465,
+      "loss": 1.6406,
+      "step": 2140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018680591761216939,
+      "loss": 1.7246,
+      "step": 2141
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018679160897890222,
+      "loss": 1.668,
+      "step": 2142
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00186777293139781,
+      "loss": 1.75,
+      "step": 2143
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001867629700959944,
+      "loss": 1.7246,
+      "step": 2144
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018674863984873156,
+      "loss": 1.6348,
+      "step": 2145
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018673430239918218,
+      "loss": 1.7695,
+      "step": 2146
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018671995774853674,
+      "loss": 1.6426,
+      "step": 2147
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018670560589798613,
+      "loss": 1.6816,
+      "step": 2148
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001866912468487219,
+      "loss": 1.6055,
+      "step": 2149
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018667688060193621,
+      "loss": 1.6328,
+      "step": 2150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018666250715882184,
+      "loss": 1.7402,
+      "step": 2151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018664812652057214,
+      "loss": 1.6387,
+      "step": 2152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018663373868838104,
+      "loss": 1.6895,
+      "step": 2153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018661934366344312,
+      "loss": 1.6426,
+      "step": 2154
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001866049414469535,
+      "loss": 1.6055,
+      "step": 2155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001865905320401079,
+      "loss": 1.6348,
+      "step": 2156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001865761154441027,
+      "loss": 1.7344,
+      "step": 2157
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018656169166013478,
+      "loss": 1.6387,
+      "step": 2158
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018654726068940176,
+      "loss": 1.7012,
+      "step": 2159
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001865328225331017,
+      "loss": 1.752,
+      "step": 2160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001865183771924333,
+      "loss": 1.7402,
+      "step": 2161
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018650392466859596,
+      "loss": 1.6738,
+      "step": 2162
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018648946496278953,
+      "loss": 1.6562,
+      "step": 2163
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018647499807621455,
+      "loss": 1.707,
+      "step": 2164
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018646052401007217,
+      "loss": 1.5898,
+      "step": 2165
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018644604276556404,
+      "loss": 1.8594,
+      "step": 2166
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018643155434389246,
+      "loss": 1.7441,
+      "step": 2167
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018641705874626036,
+      "loss": 1.7812,
+      "step": 2168
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001864025559738712,
+      "loss": 1.7148,
+      "step": 2169
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018638804602792913,
+      "loss": 1.6074,
+      "step": 2170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001863735289096388,
+      "loss": 1.6602,
+      "step": 2171
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018635900462020546,
+      "loss": 1.6699,
+      "step": 2172
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00186344473160835,
+      "loss": 1.668,
+      "step": 2173
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018632993453273393,
+      "loss": 1.7324,
+      "step": 2174
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018631538873710926,
+      "loss": 1.6992,
+      "step": 2175
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018630083577516872,
+      "loss": 1.6855,
+      "step": 2176
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001862862756481205,
+      "loss": 1.7754,
+      "step": 2177
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018627170835717349,
+      "loss": 1.6602,
+      "step": 2178
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018625713390353713,
+      "loss": 1.6797,
+      "step": 2179
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018624255228842147,
+      "loss": 1.6309,
+      "step": 2180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001862279635130371,
+      "loss": 1.6973,
+      "step": 2181
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018621336757859528,
+      "loss": 1.8047,
+      "step": 2182
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018619876448630786,
+      "loss": 1.6348,
+      "step": 2183
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001861841542373872,
+      "loss": 1.7051,
+      "step": 2184
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018616953683304638,
+      "loss": 1.8496,
+      "step": 2185
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018615491227449896,
+      "loss": 1.6387,
+      "step": 2186
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018614028056295915,
+      "loss": 1.6484,
+      "step": 2187
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018612564169964176,
+      "loss": 1.8359,
+      "step": 2188
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018611099568576215,
+      "loss": 1.6992,
+      "step": 2189
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018609634252253634,
+      "loss": 1.6621,
+      "step": 2190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0018608168221118087,
+      "loss": 1.6309,
+      "step": 2191
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018606701475291291,
+      "loss": 1.6797,
+      "step": 2192
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018605234014895027,
+      "loss": 1.6836,
+      "step": 2193
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018603765840051129,
+      "loss": 1.6172,
+      "step": 2194
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018602296950881484,
+      "loss": 1.6133,
+      "step": 2195
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018600827347508054,
+      "loss": 1.7461,
+      "step": 2196
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018599357030052853,
+      "loss": 1.7285,
+      "step": 2197
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018597885998637952,
+      "loss": 1.666,
+      "step": 2198
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018596414253385478,
+      "loss": 1.6797,
+      "step": 2199
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001859494179441763,
+      "loss": 1.7285,
+      "step": 2200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018593468621856655,
+      "loss": 1.6602,
+      "step": 2201
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018591994735824863,
+      "loss": 1.6641,
+      "step": 2202
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018590520136444624,
+      "loss": 1.8125,
+      "step": 2203
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018589044823838364,
+      "loss": 1.7656,
+      "step": 2204
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018587568798128574,
+      "loss": 1.4941,
+      "step": 2205
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018586092059437798,
+      "loss": 1.5078,
+      "step": 2206
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018584614607888646,
+      "loss": 1.6582,
+      "step": 2207
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018583136443603774,
+      "loss": 1.6621,
+      "step": 2208
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018581657566705916,
+      "loss": 1.7031,
+      "step": 2209
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001858017797731785,
+      "loss": 1.7344,
+      "step": 2210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018578697675562421,
+      "loss": 1.7676,
+      "step": 2211
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001857721666156253,
+      "loss": 1.6543,
+      "step": 2212
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018575734935441137,
+      "loss": 1.6133,
+      "step": 2213
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018574252497321263,
+      "loss": 1.666,
+      "step": 2214
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018572769347325986,
+      "loss": 1.7383,
+      "step": 2215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018571285485578445,
+      "loss": 1.7148,
+      "step": 2216
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001856980091220184,
+      "loss": 1.5918,
+      "step": 2217
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018568315627319422,
+      "loss": 1.6777,
+      "step": 2218
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001856682963105451,
+      "loss": 1.6191,
+      "step": 2219
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018565342923530475,
+      "loss": 1.623,
+      "step": 2220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018563855504870757,
+      "loss": 1.6621,
+      "step": 2221
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018562367375198844,
+      "loss": 1.7559,
+      "step": 2222
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018560878534638287,
+      "loss": 1.7656,
+      "step": 2223
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018559388983312697,
+      "loss": 1.8164,
+      "step": 2224
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018557898721345745,
+      "loss": 1.7051,
+      "step": 2225
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018556407748861162,
+      "loss": 1.6406,
+      "step": 2226
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018554916065982733,
+      "loss": 1.7734,
+      "step": 2227
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018553423672834303,
+      "loss": 1.7031,
+      "step": 2228
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001855193056953978,
+      "loss": 1.6367,
+      "step": 2229
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018550436756223127,
+      "loss": 1.666,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018548942233008367,
+      "loss": 1.6133,
+      "step": 2231
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018547447000019587,
+      "loss": 1.6641,
+      "step": 2232
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018545951057380922,
+      "loss": 1.7246,
+      "step": 2233
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018544454405216577,
+      "loss": 1.7168,
+      "step": 2234
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001854295704365081,
+      "loss": 1.5449,
+      "step": 2235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001854145897280794,
+      "loss": 1.6484,
+      "step": 2236
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001853996019281234,
+      "loss": 1.6992,
+      "step": 2237
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001853846070378845,
+      "loss": 1.5977,
+      "step": 2238
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018536960505860765,
+      "loss": 1.668,
+      "step": 2239
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018535459599153834,
+      "loss": 1.6934,
+      "step": 2240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018533957983792276,
+      "loss": 1.6543,
+      "step": 2241
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018532455659900754,
+      "loss": 1.6367,
+      "step": 2242
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018530952627604008,
+      "loss": 1.582,
+      "step": 2243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001852944888702682,
+      "loss": 1.5156,
+      "step": 2244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018527944438294038,
+      "loss": 1.6641,
+      "step": 2245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018526439281530572,
+      "loss": 1.7422,
+      "step": 2246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018524933416861386,
+      "loss": 1.6113,
+      "step": 2247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018523426844411501,
+      "loss": 1.793,
+      "step": 2248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018521919564306003,
+      "loss": 1.7246,
+      "step": 2249
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018520411576670034,
+      "loss": 1.6641,
+      "step": 2250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001851890288162879,
+      "loss": 1.7031,
+      "step": 2251
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018517393479307537,
+      "loss": 1.6465,
+      "step": 2252
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018515883369831588,
+      "loss": 1.8379,
+      "step": 2253
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018514372553326318,
+      "loss": 1.6621,
+      "step": 2254
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018512861029917167,
+      "loss": 1.7383,
+      "step": 2255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018511348799729627,
+      "loss": 1.6484,
+      "step": 2256
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001850983586288925,
+      "loss": 1.5273,
+      "step": 2257
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018508322219521647,
+      "loss": 1.7109,
+      "step": 2258
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018506807869752488,
+      "loss": 1.8125,
+      "step": 2259
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018505292813707503,
+      "loss": 1.6309,
+      "step": 2260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018503777051512478,
+      "loss": 1.5996,
+      "step": 2261
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001850226058329326,
+      "loss": 1.7363,
+      "step": 2262
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001850074340917575,
+      "loss": 1.6367,
+      "step": 2263
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018499225529285915,
+      "loss": 1.6289,
+      "step": 2264
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018497706943749774,
+      "loss": 1.5879,
+      "step": 2265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018496187652693409,
+      "loss": 1.7559,
+      "step": 2266
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018494667656242959,
+      "loss": 1.6504,
+      "step": 2267
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001849314695452462,
+      "loss": 1.668,
+      "step": 2268
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018491625547664647,
+      "loss": 1.7227,
+      "step": 2269
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018490103435789356,
+      "loss": 1.6738,
+      "step": 2270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018488580619025119,
+      "loss": 1.7598,
+      "step": 2271
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001848705709749837,
+      "loss": 1.5664,
+      "step": 2272
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018485532871335597,
+      "loss": 1.7832,
+      "step": 2273
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001848400794066335,
+      "loss": 1.6348,
+      "step": 2274
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018482482305608233,
+      "loss": 1.6504,
+      "step": 2275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018480955966296915,
+      "loss": 1.627,
+      "step": 2276
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001847942892285612,
+      "loss": 1.6758,
+      "step": 2277
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018477901175412623,
+      "loss": 1.6367,
+      "step": 2278
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018476372724093278,
+      "loss": 1.5547,
+      "step": 2279
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018474843569024973,
+      "loss": 1.7031,
+      "step": 2280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018473313710334672,
+      "loss": 1.6309,
+      "step": 2281
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018471783148149385,
+      "loss": 1.6895,
+      "step": 2282
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018470251882596195,
+      "loss": 1.709,
+      "step": 2283
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018468719913802229,
+      "loss": 1.748,
+      "step": 2284
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018467187241894677,
+      "loss": 1.7402,
+      "step": 2285
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018465653867000796,
+      "loss": 1.9023,
+      "step": 2286
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018464119789247892,
+      "loss": 1.5898,
+      "step": 2287
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001846258500876332,
+      "loss": 1.6445,
+      "step": 2288
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018461049525674523,
+      "loss": 1.5957,
+      "step": 2289
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001845951334010897,
+      "loss": 1.5801,
+      "step": 2290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018457976452194212,
+      "loss": 1.5898,
+      "step": 2291
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018456438862057842,
+      "loss": 1.75,
+      "step": 2292
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018454900569827523,
+      "loss": 1.5996,
+      "step": 2293
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018453361575630964,
+      "loss": 1.6895,
+      "step": 2294
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018451821879595948,
+      "loss": 1.7305,
+      "step": 2295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018450281481850303,
+      "loss": 1.7012,
+      "step": 2296
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018448740382521925,
+      "loss": 1.6172,
+      "step": 2297
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018447198581738756,
+      "loss": 1.6211,
+      "step": 2298
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001844565607962881,
+      "loss": 1.6523,
+      "step": 2299
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001844411287632015,
+      "loss": 1.7148,
+      "step": 2300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018442568971940901,
+      "loss": 1.7051,
+      "step": 2301
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018441024366619243,
+      "loss": 1.5781,
+      "step": 2302
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018439479060483421,
+      "loss": 1.7246,
+      "step": 2303
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0018437933053661731,
+      "loss": 1.5039,
+      "step": 2304
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001843638634628253,
+      "loss": 1.7012,
+      "step": 2305
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018434838938474235,
+      "loss": 1.7754,
+      "step": 2306
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018433290830365317,
+      "loss": 1.6738,
+      "step": 2307
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018431742022084305,
+      "loss": 1.6895,
+      "step": 2308
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001843019251375979,
+      "loss": 1.6504,
+      "step": 2309
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018428642305520426,
+      "loss": 1.7266,
+      "step": 2310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018427091397494914,
+      "loss": 1.666,
+      "step": 2311
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001842553978981201,
+      "loss": 1.6387,
+      "step": 2312
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018423987482600551,
+      "loss": 1.6738,
+      "step": 2313
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018422434475989405,
+      "loss": 1.5332,
+      "step": 2314
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018420880770107513,
+      "loss": 1.6855,
+      "step": 2315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018419326365083873,
+      "loss": 1.6934,
+      "step": 2316
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001841777126104754,
+      "loss": 1.6816,
+      "step": 2317
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018416215458127625,
+      "loss": 1.7578,
+      "step": 2318
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018414658956453296,
+      "loss": 1.7207,
+      "step": 2319
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018413101756153782,
+      "loss": 1.6602,
+      "step": 2320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018411543857358372,
+      "loss": 1.6016,
+      "step": 2321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018409985260196408,
+      "loss": 1.5352,
+      "step": 2322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018408425964797292,
+      "loss": 1.7734,
+      "step": 2323
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018406865971290485,
+      "loss": 1.668,
+      "step": 2324
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018405305279805507,
+      "loss": 1.6191,
+      "step": 2325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001840374389047193,
+      "loss": 1.6191,
+      "step": 2326
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018402181803419387,
+      "loss": 1.5977,
+      "step": 2327
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018400619018777576,
+      "loss": 1.6895,
+      "step": 2328
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001839905553667624,
+      "loss": 1.5352,
+      "step": 2329
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018397491357245194,
+      "loss": 1.707,
+      "step": 2330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00183959264806143,
+      "loss": 1.6797,
+      "step": 2331
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018394360906913482,
+      "loss": 1.6602,
+      "step": 2332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018392794636272719,
+      "loss": 1.7461,
+      "step": 2333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018391227668822052,
+      "loss": 1.7324,
+      "step": 2334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018389660004691582,
+      "loss": 1.6621,
+      "step": 2335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018388091644011455,
+      "loss": 1.6328,
+      "step": 2336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018386522586911892,
+      "loss": 1.668,
+      "step": 2337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001838495283352316,
+      "loss": 1.6328,
+      "step": 2338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018383382383975588,
+      "loss": 1.5293,
+      "step": 2339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018381811238399564,
+      "loss": 1.5879,
+      "step": 2340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001838023939692553,
+      "loss": 1.6836,
+      "step": 2341
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018378666859683985,
+      "loss": 1.6094,
+      "step": 2342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018377093626805495,
+      "loss": 1.6562,
+      "step": 2343
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018375519698420675,
+      "loss": 1.6133,
+      "step": 2344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018373945074660197,
+      "loss": 1.6875,
+      "step": 2345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018372369755654796,
+      "loss": 1.7754,
+      "step": 2346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018370793741535262,
+      "loss": 1.625,
+      "step": 2347
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018369217032432443,
+      "loss": 1.6621,
+      "step": 2348
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018367639628477249,
+      "loss": 1.6055,
+      "step": 2349
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018366061529800637,
+      "loss": 1.5605,
+      "step": 2350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018364482736533635,
+      "loss": 1.5859,
+      "step": 2351
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018362903248807312,
+      "loss": 1.6719,
+      "step": 2352
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018361323066752818,
+      "loss": 1.6191,
+      "step": 2353
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018359742190501335,
+      "loss": 1.6035,
+      "step": 2354
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018358160620184122,
+      "loss": 1.7188,
+      "step": 2355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018356578355932488,
+      "loss": 1.623,
+      "step": 2356
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018354995397877798,
+      "loss": 1.6797,
+      "step": 2357
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018353411746151475,
+      "loss": 1.543,
+      "step": 2358
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018351827400885008,
+      "loss": 1.6328,
+      "step": 2359
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001835024236220993,
+      "loss": 1.7188,
+      "step": 2360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018348656630257842,
+      "loss": 1.6777,
+      "step": 2361
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00183470702051604,
+      "loss": 1.6836,
+      "step": 2362
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018345483087049313,
+      "loss": 1.6348,
+      "step": 2363
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018343895276056353,
+      "loss": 1.707,
+      "step": 2364
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018342306772313349,
+      "loss": 1.6758,
+      "step": 2365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018340717575952184,
+      "loss": 1.6719,
+      "step": 2366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00183391276871048,
+      "loss": 1.4297,
+      "step": 2367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018337537105903201,
+      "loss": 1.6777,
+      "step": 2368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018335945832479443,
+      "loss": 1.5977,
+      "step": 2369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018334353866965639,
+      "loss": 1.6895,
+      "step": 2370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018332761209493963,
+      "loss": 1.7891,
+      "step": 2371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018331167860196644,
+      "loss": 1.8027,
+      "step": 2372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018329573819205974,
+      "loss": 1.5195,
+      "step": 2373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001832797908665429,
+      "loss": 1.6699,
+      "step": 2374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018326383662674002,
+      "loss": 1.627,
+      "step": 2375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018324787547397566,
+      "loss": 1.7188,
+      "step": 2376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018323190740957502,
+      "loss": 1.5234,
+      "step": 2377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018321593243486377,
+      "loss": 1.5625,
+      "step": 2378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018319995055116832,
+      "loss": 1.7598,
+      "step": 2379
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001831839617598155,
+      "loss": 1.6055,
+      "step": 2380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018316796606213282,
+      "loss": 1.709,
+      "step": 2381
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018315196345944826,
+      "loss": 1.6797,
+      "step": 2382
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018313595395309048,
+      "loss": 1.6621,
+      "step": 2383
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018311993754438869,
+      "loss": 1.6543,
+      "step": 2384
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018310391423467258,
+      "loss": 1.5918,
+      "step": 2385
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018308788402527254,
+      "loss": 1.6191,
+      "step": 2386
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018307184691751944,
+      "loss": 1.7207,
+      "step": 2387
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018305580291274476,
+      "loss": 1.6074,
+      "step": 2388
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018303975201228058,
+      "loss": 1.8145,
+      "step": 2389
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018302369421745945,
+      "loss": 1.625,
+      "step": 2390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018300762952961466,
+      "loss": 1.6797,
+      "step": 2391
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018299155795007992,
+      "loss": 1.5957,
+      "step": 2392
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018297547948018958,
+      "loss": 1.6875,
+      "step": 2393
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018295939412127855,
+      "loss": 1.6914,
+      "step": 2394
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018294330187468234,
+      "loss": 1.6582,
+      "step": 2395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018292720274173696,
+      "loss": 1.6484,
+      "step": 2396
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018291109672377908,
+      "loss": 1.7402,
+      "step": 2397
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018289498382214585,
+      "loss": 1.6797,
+      "step": 2398
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018287886403817507,
+      "loss": 1.5996,
+      "step": 2399
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018286273737320507,
+      "loss": 1.5312,
+      "step": 2400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001828466038285748,
+      "loss": 1.6797,
+      "step": 2401
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001828304634056237,
+      "loss": 1.6973,
+      "step": 2402
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018281431610569184,
+      "loss": 1.627,
+      "step": 2403
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018279816193011984,
+      "loss": 1.7227,
+      "step": 2404
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001827820008802489,
+      "loss": 1.6094,
+      "step": 2405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018276583295742082,
+      "loss": 1.7676,
+      "step": 2406
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001827496581629779,
+      "loss": 1.6152,
+      "step": 2407
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00182733476498263,
+      "loss": 1.6699,
+      "step": 2408
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018271728796461972,
+      "loss": 1.666,
+      "step": 2409
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018270109256339202,
+      "loss": 1.6836,
+      "step": 2410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018268489029592456,
+      "loss": 1.7188,
+      "step": 2411
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001826686811635625,
+      "loss": 1.5898,
+      "step": 2412
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018265246516765163,
+      "loss": 1.6914,
+      "step": 2413
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018263624230953827,
+      "loss": 1.6328,
+      "step": 2414
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001826200125905693,
+      "loss": 1.7285,
+      "step": 2415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0018260377601209221,
+      "loss": 1.6445,
+      "step": 2416
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00182587532575455,
+      "loss": 1.7051,
+      "step": 2417
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018257128228200636,
+      "loss": 1.7539,
+      "step": 2418
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018255502513309537,
+      "loss": 1.7773,
+      "step": 2419
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018253876113007183,
+      "loss": 1.6895,
+      "step": 2420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018252249027428608,
+      "loss": 1.6367,
+      "step": 2421
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018250621256708892,
+      "loss": 1.6504,
+      "step": 2422
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001824899280098319,
+      "loss": 1.707,
+      "step": 2423
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018247363660386698,
+      "loss": 1.666,
+      "step": 2424
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018245733835054675,
+      "loss": 1.5137,
+      "step": 2425
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001824410332512244,
+      "loss": 1.6016,
+      "step": 2426
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018242472130725364,
+      "loss": 1.6855,
+      "step": 2427
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018240840251998877,
+      "loss": 1.7207,
+      "step": 2428
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018239207689078462,
+      "loss": 1.6211,
+      "step": 2429
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018237574442099666,
+      "loss": 1.5449,
+      "step": 2430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018235940511198093,
+      "loss": 1.7031,
+      "step": 2431
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018234305896509392,
+      "loss": 1.6289,
+      "step": 2432
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018232670598169278,
+      "loss": 1.5918,
+      "step": 2433
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018231034616313525,
+      "loss": 1.7324,
+      "step": 2434
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018229397951077958,
+      "loss": 1.6504,
+      "step": 2435
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018227760602598459,
+      "loss": 1.5898,
+      "step": 2436
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018226122571010973,
+      "loss": 1.6172,
+      "step": 2437
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001822448385645149,
+      "loss": 1.709,
+      "step": 2438
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001822284445905607,
+      "loss": 1.7266,
+      "step": 2439
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001822120437896082,
+      "loss": 1.5918,
+      "step": 2440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018219563616301913,
+      "loss": 1.6738,
+      "step": 2441
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018217922171215567,
+      "loss": 1.7129,
+      "step": 2442
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018216280043838065,
+      "loss": 1.7285,
+      "step": 2443
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018214637234305745,
+      "loss": 1.6875,
+      "step": 2444
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018212993742754997,
+      "loss": 1.623,
+      "step": 2445
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018211349569322278,
+      "loss": 1.7266,
+      "step": 2446
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001820970471414409,
+      "loss": 1.6035,
+      "step": 2447
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018208059177356995,
+      "loss": 1.7109,
+      "step": 2448
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001820641295909762,
+      "loss": 1.5098,
+      "step": 2449
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018204766059502638,
+      "loss": 1.7324,
+      "step": 2450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018203118478708781,
+      "loss": 1.623,
+      "step": 2451
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018201470216852843,
+      "loss": 1.7012,
+      "step": 2452
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018199821274071668,
+      "loss": 1.75,
+      "step": 2453
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001819817165050216,
+      "loss": 1.7324,
+      "step": 2454
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018196521346281278,
+      "loss": 1.5996,
+      "step": 2455
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018194870361546039,
+      "loss": 1.7246,
+      "step": 2456
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018193218696433513,
+      "loss": 1.6367,
+      "step": 2457
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018191566351080833,
+      "loss": 1.623,
+      "step": 2458
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001818991332562518,
+      "loss": 1.6836,
+      "step": 2459
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018188259620203803,
+      "loss": 1.6621,
+      "step": 2460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018186605234953995,
+      "loss": 1.627,
+      "step": 2461
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001818495017001311,
+      "loss": 1.7129,
+      "step": 2462
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018183294425518563,
+      "loss": 1.7148,
+      "step": 2463
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018181638001607818,
+      "loss": 1.7188,
+      "step": 2464
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018179980898418407,
+      "loss": 1.7344,
+      "step": 2465
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018178323116087903,
+      "loss": 1.6816,
+      "step": 2466
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018176664654753945,
+      "loss": 1.7598,
+      "step": 2467
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018175005514554227,
+      "loss": 1.6094,
+      "step": 2468
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018173345695626497,
+      "loss": 1.7285,
+      "step": 2469
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018171685198108565,
+      "loss": 1.582,
+      "step": 2470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001817002402213829,
+      "loss": 1.6133,
+      "step": 2471
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018168362167853593,
+      "loss": 1.6035,
+      "step": 2472
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018166699635392446,
+      "loss": 1.6289,
+      "step": 2473
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018165036424892882,
+      "loss": 1.7188,
+      "step": 2474
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018163372536492992,
+      "loss": 1.5957,
+      "step": 2475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018161707970330914,
+      "loss": 1.6367,
+      "step": 2476
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018160042726544851,
+      "loss": 1.6719,
+      "step": 2477
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018158376805273062,
+      "loss": 1.7246,
+      "step": 2478
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018156710206653855,
+      "loss": 1.6387,
+      "step": 2479
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018155042930825603,
+      "loss": 1.709,
+      "step": 2480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001815337497792673,
+      "loss": 1.7754,
+      "step": 2481
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018151706348095715,
+      "loss": 1.6816,
+      "step": 2482
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00181500370414711,
+      "loss": 1.6504,
+      "step": 2483
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018148367058191473,
+      "loss": 1.5742,
+      "step": 2484
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001814669639839549,
+      "loss": 1.5996,
+      "step": 2485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018145025062221857,
+      "loss": 1.6289,
+      "step": 2486
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001814335304980933,
+      "loss": 1.6895,
+      "step": 2487
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018141680361296732,
+      "loss": 1.541,
+      "step": 2488
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018140006996822935,
+      "loss": 1.6289,
+      "step": 2489
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018138332956526876,
+      "loss": 1.6543,
+      "step": 2490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018136658240547531,
+      "loss": 1.6758,
+      "step": 2491
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018134982849023955,
+      "loss": 1.5488,
+      "step": 2492
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018133306782095243,
+      "loss": 1.7227,
+      "step": 2493
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018131630039900544,
+      "loss": 1.6367,
+      "step": 2494
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018129952622579074,
+      "loss": 1.6602,
+      "step": 2495
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018128274530270101,
+      "loss": 1.6289,
+      "step": 2496
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018126595763112948,
+      "loss": 1.6328,
+      "step": 2497
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018124916321246994,
+      "loss": 1.5957,
+      "step": 2498
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018123236204811671,
+      "loss": 1.7891,
+      "step": 2499
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018121555413946472,
+      "loss": 1.748,
+      "step": 2500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018119873948790948,
+      "loss": 1.627,
+      "step": 2501
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018118191809484699,
+      "loss": 1.6797,
+      "step": 2502
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018116508996167385,
+      "loss": 1.6992,
+      "step": 2503
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018114825508978721,
+      "loss": 1.5156,
+      "step": 2504
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018113141348058475,
+      "loss": 1.6992,
+      "step": 2505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018111456513546479,
+      "loss": 1.6738,
+      "step": 2506
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018109771005582617,
+      "loss": 1.6914,
+      "step": 2507
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018108084824306821,
+      "loss": 1.6152,
+      "step": 2508
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018106397969859089,
+      "loss": 1.6934,
+      "step": 2509
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018104710442379476,
+      "loss": 1.6055,
+      "step": 2510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018103022242008085,
+      "loss": 1.6758,
+      "step": 2511
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018101333368885078,
+      "loss": 1.5879,
+      "step": 2512
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018099643823150673,
+      "loss": 1.5938,
+      "step": 2513
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018097953604945148,
+      "loss": 1.6016,
+      "step": 2514
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001809626271440883,
+      "loss": 1.707,
+      "step": 2515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018094571151682102,
+      "loss": 1.6836,
+      "step": 2516
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018092878916905408,
+      "loss": 1.6465,
+      "step": 2517
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001809118601021925,
+      "loss": 1.6074,
+      "step": 2518
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018089492431764174,
+      "loss": 1.7031,
+      "step": 2519
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001808779818168079,
+      "loss": 1.6211,
+      "step": 2520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018086103260109767,
+      "loss": 1.6699,
+      "step": 2521
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018084407667191826,
+      "loss": 1.623,
+      "step": 2522
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018082711403067738,
+      "loss": 1.4902,
+      "step": 2523
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001808101446787834,
+      "loss": 1.6074,
+      "step": 2524
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018079316861764516,
+      "loss": 1.6445,
+      "step": 2525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018077618584867211,
+      "loss": 1.7383,
+      "step": 2526
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001807591963732742,
+      "loss": 1.7559,
+      "step": 2527
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018074220019286204,
+      "loss": 1.6602,
+      "step": 2528
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0018072519730884672,
+      "loss": 1.6172,
+      "step": 2529
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018070818772263988,
+      "loss": 1.5977,
+      "step": 2530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018069117143565375,
+      "loss": 1.7305,
+      "step": 2531
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018067414844930108,
+      "loss": 1.5508,
+      "step": 2532
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018065711876499525,
+      "loss": 1.707,
+      "step": 2533
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018064008238415008,
+      "loss": 1.6484,
+      "step": 2534
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018062303930818006,
+      "loss": 1.6934,
+      "step": 2535
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018060598953850017,
+      "loss": 1.6016,
+      "step": 2536
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018058893307652598,
+      "loss": 1.6992,
+      "step": 2537
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018057186992367358,
+      "loss": 1.5508,
+      "step": 2538
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018055480008135963,
+      "loss": 1.541,
+      "step": 2539
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018053772355100138,
+      "loss": 1.6699,
+      "step": 2540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018052064033401657,
+      "loss": 1.7891,
+      "step": 2541
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018050355043182354,
+      "loss": 1.5801,
+      "step": 2542
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001804864538458412,
+      "loss": 1.6426,
+      "step": 2543
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018046935057748896,
+      "loss": 1.5879,
+      "step": 2544
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018045224062818682,
+      "loss": 1.6035,
+      "step": 2545
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018043512399935535,
+      "loss": 2.0059,
+      "step": 2546
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018041800069241564,
+      "loss": 1.6738,
+      "step": 2547
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018040087070878934,
+      "loss": 1.6445,
+      "step": 2548
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018038373404989868,
+      "loss": 1.7402,
+      "step": 2549
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001803665907171664,
+      "loss": 1.6973,
+      "step": 2550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018034944071201587,
+      "loss": 1.6035,
+      "step": 2551
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018033228403587093,
+      "loss": 1.5957,
+      "step": 2552
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018031512069015601,
+      "loss": 1.5957,
+      "step": 2553
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018029795067629611,
+      "loss": 1.6797,
+      "step": 2554
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018028077399571672,
+      "loss": 1.6582,
+      "step": 2555
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018026359064984398,
+      "loss": 1.6934,
+      "step": 2556
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018024640064010454,
+      "loss": 1.6934,
+      "step": 2557
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018022920396792556,
+      "loss": 1.7031,
+      "step": 2558
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001802120006347348,
+      "loss": 1.584,
+      "step": 2559
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018019479064196057,
+      "loss": 1.7285,
+      "step": 2560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001801775739910317,
+      "loss": 1.5508,
+      "step": 2561
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001801603506833777,
+      "loss": 1.5918,
+      "step": 2562
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018014312072042837,
+      "loss": 1.6035,
+      "step": 2563
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018012588410361437,
+      "loss": 1.6738,
+      "step": 2564
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018010864083436667,
+      "loss": 1.6445,
+      "step": 2565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001800913909141169,
+      "loss": 1.7031,
+      "step": 2566
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018007413434429729,
+      "loss": 1.5684,
+      "step": 2567
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018005687112634052,
+      "loss": 1.6211,
+      "step": 2568
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018003960126167986,
+      "loss": 1.7812,
+      "step": 2569
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018002232475174916,
+      "loss": 1.5781,
+      "step": 2570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0018000504159798277,
+      "loss": 1.6641,
+      "step": 2571
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017998775180181567,
+      "loss": 1.6484,
+      "step": 2572
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017997045536468327,
+      "loss": 1.6758,
+      "step": 2573
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017995315228802164,
+      "loss": 1.668,
+      "step": 2574
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017993584257326738,
+      "loss": 1.6738,
+      "step": 2575
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017991852622185763,
+      "loss": 1.5605,
+      "step": 2576
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017990120323523003,
+      "loss": 1.6406,
+      "step": 2577
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017988387361482284,
+      "loss": 1.6172,
+      "step": 2578
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017986653736207483,
+      "loss": 1.502,
+      "step": 2579
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001798491944784254,
+      "loss": 1.7324,
+      "step": 2580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017983184496531436,
+      "loss": 1.6758,
+      "step": 2581
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017981448882418223,
+      "loss": 1.6719,
+      "step": 2582
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017979712605646996,
+      "loss": 1.6621,
+      "step": 2583
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017977975666361905,
+      "loss": 1.6348,
+      "step": 2584
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017976238064707165,
+      "loss": 1.6211,
+      "step": 2585
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017974499800827038,
+      "loss": 1.6758,
+      "step": 2586
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001797276087486584,
+      "loss": 1.6465,
+      "step": 2587
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001797102128696795,
+      "loss": 1.6289,
+      "step": 2588
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017969281037277793,
+      "loss": 1.5625,
+      "step": 2589
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017967540125939855,
+      "loss": 1.6211,
+      "step": 2590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017965798553098676,
+      "loss": 1.5332,
+      "step": 2591
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017964056318898844,
+      "loss": 1.6445,
+      "step": 2592
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017962313423485017,
+      "loss": 1.6113,
+      "step": 2593
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017960569867001889,
+      "loss": 1.6816,
+      "step": 2594
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001795882564959422,
+      "loss": 1.6992,
+      "step": 2595
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001795708077140683,
+      "loss": 1.6699,
+      "step": 2596
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017955335232584582,
+      "loss": 1.7168,
+      "step": 2597
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00179535890332724,
+      "loss": 1.6797,
+      "step": 2598
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017951842173615261,
+      "loss": 1.623,
+      "step": 2599
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017950094653758194,
+      "loss": 1.6914,
+      "step": 2600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017948346473846298,
+      "loss": 1.6719,
+      "step": 2601
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017946597634024705,
+      "loss": 1.7539,
+      "step": 2602
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017944848134438614,
+      "loss": 1.5859,
+      "step": 2603
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017943097975233276,
+      "loss": 1.5977,
+      "step": 2604
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017941347156554,
+      "loss": 1.6699,
+      "step": 2605
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017939595678546148,
+      "loss": 1.5664,
+      "step": 2606
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017937843541355135,
+      "loss": 1.707,
+      "step": 2607
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001793609074512643,
+      "loss": 1.6934,
+      "step": 2608
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001793433729000556,
+      "loss": 1.6758,
+      "step": 2609
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017932583176138107,
+      "loss": 1.7188,
+      "step": 2610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017930828403669703,
+      "loss": 1.666,
+      "step": 2611
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017929072972746036,
+      "loss": 1.6719,
+      "step": 2612
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017927316883512858,
+      "loss": 1.7695,
+      "step": 2613
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001792556013611596,
+      "loss": 1.6445,
+      "step": 2614
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00179238027307012,
+      "loss": 1.5781,
+      "step": 2615
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017922044667414484,
+      "loss": 1.5547,
+      "step": 2616
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017920285946401774,
+      "loss": 1.6465,
+      "step": 2617
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001791852656780909,
+      "loss": 1.6055,
+      "step": 2618
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017916766531782503,
+      "loss": 1.627,
+      "step": 2619
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001791500583846814,
+      "loss": 1.6074,
+      "step": 2620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001791324448801218,
+      "loss": 1.6797,
+      "step": 2621
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001791148248056086,
+      "loss": 1.6035,
+      "step": 2622
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017909719816260475,
+      "loss": 1.6914,
+      "step": 2623
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017907956495257361,
+      "loss": 1.6055,
+      "step": 2624
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017906192517697923,
+      "loss": 1.752,
+      "step": 2625
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017904427883728613,
+      "loss": 1.625,
+      "step": 2626
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001790266259349594,
+      "loss": 1.6152,
+      "step": 2627
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001790089664714647,
+      "loss": 1.543,
+      "step": 2628
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017899130044826816,
+      "loss": 1.6777,
+      "step": 2629
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001789736278668365,
+      "loss": 1.6289,
+      "step": 2630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017895594872863697,
+      "loss": 1.6426,
+      "step": 2631
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017893826303513741,
+      "loss": 1.6699,
+      "step": 2632
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017892057078780618,
+      "loss": 1.5957,
+      "step": 2633
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017890287198811215,
+      "loss": 1.6855,
+      "step": 2634
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017888516663752472,
+      "loss": 1.7266,
+      "step": 2635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017886745473751397,
+      "loss": 1.7012,
+      "step": 2636
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017884973628955031,
+      "loss": 1.709,
+      "step": 2637
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001788320112951049,
+      "loss": 1.6992,
+      "step": 2638
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017881427975564934,
+      "loss": 1.6426,
+      "step": 2639
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0017879654167265579,
+      "loss": 1.7246,
+      "step": 2640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001787787970475969,
+      "loss": 1.6992,
+      "step": 2641
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017876104588194594,
+      "loss": 1.5723,
+      "step": 2642
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017874328817717674,
+      "loss": 1.6836,
+      "step": 2643
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017872552393476356,
+      "loss": 1.6758,
+      "step": 2644
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017870775315618132,
+      "loss": 1.6738,
+      "step": 2645
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017868997584290541,
+      "loss": 1.6836,
+      "step": 2646
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001786721919964118,
+      "loss": 1.7305,
+      "step": 2647
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017865440161817698,
+      "loss": 1.6992,
+      "step": 2648
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00178636604709678,
+      "loss": 1.5273,
+      "step": 2649
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017861880127239247,
+      "loss": 1.6543,
+      "step": 2650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017860099130779846,
+      "loss": 1.6973,
+      "step": 2651
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017858317481737468,
+      "loss": 1.6895,
+      "step": 2652
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017856535180260033,
+      "loss": 1.6406,
+      "step": 2653
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017854752226495517,
+      "loss": 1.5508,
+      "step": 2654
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017852968620591944,
+      "loss": 1.6699,
+      "step": 2655
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017851184362697406,
+      "loss": 1.6797,
+      "step": 2656
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017849399452960033,
+      "loss": 1.5508,
+      "step": 2657
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017847613891528026,
+      "loss": 1.5645,
+      "step": 2658
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001784582767854962,
+      "loss": 1.6621,
+      "step": 2659
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017844040814173122,
+      "loss": 1.6113,
+      "step": 2660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017842253298546882,
+      "loss": 1.6562,
+      "step": 2661
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017840465131819313,
+      "loss": 1.6406,
+      "step": 2662
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017838676314138873,
+      "loss": 1.6543,
+      "step": 2663
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017836886845654077,
+      "loss": 1.6191,
+      "step": 2664
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00178350967265135,
+      "loss": 1.6191,
+      "step": 2665
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017833305956865763,
+      "loss": 1.6289,
+      "step": 2666
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017831514536859549,
+      "loss": 1.6621,
+      "step": 2667
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001782972246664358,
+      "loss": 1.6602,
+      "step": 2668
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017827929746366653,
+      "loss": 1.6406,
+      "step": 2669
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017826136376177602,
+      "loss": 1.6113,
+      "step": 2670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017824342356225322,
+      "loss": 1.6543,
+      "step": 2671
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017822547686658762,
+      "loss": 1.5977,
+      "step": 2672
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001782075236762693,
+      "loss": 1.6367,
+      "step": 2673
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017818956399278869,
+      "loss": 1.6777,
+      "step": 2674
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017817159781763699,
+      "loss": 1.6543,
+      "step": 2675
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001781536251523058,
+      "loss": 1.6562,
+      "step": 2676
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017813564599828732,
+      "loss": 1.543,
+      "step": 2677
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017811766035707422,
+      "loss": 1.6797,
+      "step": 2678
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001780996682301598,
+      "loss": 1.6504,
+      "step": 2679
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017808166961903785,
+      "loss": 1.5879,
+      "step": 2680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017806366452520264,
+      "loss": 1.7617,
+      "step": 2681
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017804565295014912,
+      "loss": 1.6875,
+      "step": 2682
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017802763489537265,
+      "loss": 1.6582,
+      "step": 2683
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001780096103623692,
+      "loss": 1.6777,
+      "step": 2684
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017799157935263525,
+      "loss": 1.6816,
+      "step": 2685
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017797354186766777,
+      "loss": 1.6289,
+      "step": 2686
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017795549790896436,
+      "loss": 1.6172,
+      "step": 2687
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017793744747802314,
+      "loss": 1.7305,
+      "step": 2688
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017791939057634272,
+      "loss": 1.6445,
+      "step": 2689
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017790132720542225,
+      "loss": 1.6523,
+      "step": 2690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017788325736676147,
+      "loss": 1.6211,
+      "step": 2691
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017786518106186061,
+      "loss": 1.7227,
+      "step": 2692
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001778470982922204,
+      "loss": 1.6035,
+      "step": 2693
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017782900905934228,
+      "loss": 1.625,
+      "step": 2694
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017781091336472798,
+      "loss": 1.6641,
+      "step": 2695
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017779281120987999,
+      "loss": 1.7598,
+      "step": 2696
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017777470259630116,
+      "loss": 1.7266,
+      "step": 2697
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017775658752549497,
+      "loss": 1.709,
+      "step": 2698
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017773846599896546,
+      "loss": 1.7051,
+      "step": 2699
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001777203380182171,
+      "loss": 1.7461,
+      "step": 2700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017770220358475505,
+      "loss": 1.8125,
+      "step": 2701
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017768406270008486,
+      "loss": 1.6953,
+      "step": 2702
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017766591536571266,
+      "loss": 1.623,
+      "step": 2703
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017764776158314516,
+      "loss": 1.75,
+      "step": 2704
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017762960135388956,
+      "loss": 1.6914,
+      "step": 2705
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017761143467945365,
+      "loss": 1.5898,
+      "step": 2706
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017759326156134563,
+      "loss": 1.7598,
+      "step": 2707
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017757508200107438,
+      "loss": 1.7148,
+      "step": 2708
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017755689600014923,
+      "loss": 1.6172,
+      "step": 2709
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017753870356008008,
+      "loss": 1.5391,
+      "step": 2710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017752050468237738,
+      "loss": 1.6445,
+      "step": 2711
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017750229936855204,
+      "loss": 1.5547,
+      "step": 2712
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017748408762011557,
+      "loss": 1.6387,
+      "step": 2713
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017746586943858,
+      "loss": 1.6738,
+      "step": 2714
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017744764482545787,
+      "loss": 1.7207,
+      "step": 2715
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017742941378226234,
+      "loss": 1.6113,
+      "step": 2716
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017741117631050695,
+      "loss": 1.6133,
+      "step": 2717
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017739293241170592,
+      "loss": 1.5352,
+      "step": 2718
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001773746820873739,
+      "loss": 1.6797,
+      "step": 2719
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001773564253390262,
+      "loss": 1.7168,
+      "step": 2720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001773381621681785,
+      "loss": 1.6348,
+      "step": 2721
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001773198925763471,
+      "loss": 1.6895,
+      "step": 2722
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001773016165650489,
+      "loss": 1.6816,
+      "step": 2723
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001772833341358012,
+      "loss": 1.625,
+      "step": 2724
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001772650452901219,
+      "loss": 1.6641,
+      "step": 2725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017724675002952946,
+      "loss": 1.6875,
+      "step": 2726
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017722844835554282,
+      "loss": 1.6504,
+      "step": 2727
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017721014026968144,
+      "loss": 1.6074,
+      "step": 2728
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001771918257734654,
+      "loss": 1.5918,
+      "step": 2729
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017717350486841522,
+      "loss": 1.5312,
+      "step": 2730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00177155177556052,
+      "loss": 1.5703,
+      "step": 2731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001771368438378974,
+      "loss": 1.5488,
+      "step": 2732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017711850371547348,
+      "loss": 1.6836,
+      "step": 2733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00177100157190303,
+      "loss": 1.6348,
+      "step": 2734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017708180426390917,
+      "loss": 1.6621,
+      "step": 2735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017706344493781572,
+      "loss": 1.6836,
+      "step": 2736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017704507921354692,
+      "loss": 1.5723,
+      "step": 2737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001770267070926276,
+      "loss": 1.6094,
+      "step": 2738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017700832857658311,
+      "loss": 1.5332,
+      "step": 2739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017698994366693928,
+      "loss": 1.6523,
+      "step": 2740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017697155236522256,
+      "loss": 1.6387,
+      "step": 2741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017695315467295984,
+      "loss": 1.6582,
+      "step": 2742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017693475059167864,
+      "loss": 1.6152,
+      "step": 2743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017691634012290689,
+      "loss": 1.5879,
+      "step": 2744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017689792326817317,
+      "loss": 1.5957,
+      "step": 2745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017687950002900652,
+      "loss": 1.5508,
+      "step": 2746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017686107040693648,
+      "loss": 1.6582,
+      "step": 2747
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017684263440349325,
+      "loss": 1.5527,
+      "step": 2748
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017682419202020739,
+      "loss": 1.6973,
+      "step": 2749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017680574325861012,
+      "loss": 1.7129,
+      "step": 2750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017678728812023316,
+      "loss": 1.5645,
+      "step": 2751
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017676882660660868,
+      "loss": 1.7383,
+      "step": 2752
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017675035871926952,
+      "loss": 1.6719,
+      "step": 2753
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0017673188445974893,
+      "loss": 1.5762,
+      "step": 2754
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017671340382958073,
+      "loss": 1.6113,
+      "step": 2755
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017669491683029927,
+      "loss": 1.5977,
+      "step": 2756
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017667642346343947,
+      "loss": 1.5918,
+      "step": 2757
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017665792373053667,
+      "loss": 1.7129,
+      "step": 2758
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017663941763312685,
+      "loss": 1.6895,
+      "step": 2759
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017662090517274647,
+      "loss": 1.5801,
+      "step": 2760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017660238635093253,
+      "loss": 1.5781,
+      "step": 2761
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017658386116922253,
+      "loss": 1.5352,
+      "step": 2762
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001765653296291545,
+      "loss": 1.584,
+      "step": 2763
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001765467917322671,
+      "loss": 1.5508,
+      "step": 2764
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017652824748009932,
+      "loss": 1.7383,
+      "step": 2765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001765096968741909,
+      "loss": 1.5898,
+      "step": 2766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017649113991608193,
+      "loss": 1.6426,
+      "step": 2767
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017647257660731311,
+      "loss": 1.584,
+      "step": 2768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001764540069494257,
+      "loss": 1.7051,
+      "step": 2769
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017643543094396138,
+      "loss": 1.6387,
+      "step": 2770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017641684859246245,
+      "loss": 1.7891,
+      "step": 2771
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001763982598964717,
+      "loss": 1.8652,
+      "step": 2772
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017637966485753245,
+      "loss": 1.5801,
+      "step": 2773
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017636106347718858,
+      "loss": 1.7227,
+      "step": 2774
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017634245575698438,
+      "loss": 1.707,
+      "step": 2775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017632384169846485,
+      "loss": 1.6113,
+      "step": 2776
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017630522130317537,
+      "loss": 1.5137,
+      "step": 2777
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001762865945726619,
+      "loss": 1.7949,
+      "step": 2778
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017626796150847091,
+      "loss": 1.6445,
+      "step": 2779
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017624932211214941,
+      "loss": 1.6582,
+      "step": 2780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017623067638524496,
+      "loss": 1.7148,
+      "step": 2781
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017621202432930559,
+      "loss": 1.7031,
+      "step": 2782
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017619336594587992,
+      "loss": 1.6328,
+      "step": 2783
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017617470123651698,
+      "loss": 1.5703,
+      "step": 2784
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017615603020276645,
+      "loss": 1.6367,
+      "step": 2785
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017613735284617853,
+      "loss": 1.7539,
+      "step": 2786
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017611866916830384,
+      "loss": 1.6094,
+      "step": 2787
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001760999791706936,
+      "loss": 1.6953,
+      "step": 2788
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017608128285489957,
+      "loss": 1.5605,
+      "step": 2789
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017606258022247396,
+      "loss": 1.5703,
+      "step": 2790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017604387127496963,
+      "loss": 1.6113,
+      "step": 2791
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017602515601393982,
+      "loss": 1.6719,
+      "step": 2792
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017600643444093836,
+      "loss": 1.6875,
+      "step": 2793
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017598770655751963,
+      "loss": 1.7344,
+      "step": 2794
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017596897236523852,
+      "loss": 1.6113,
+      "step": 2795
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017595023186565042,
+      "loss": 1.6328,
+      "step": 2796
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017593148506031124,
+      "loss": 1.6855,
+      "step": 2797
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001759127319507774,
+      "loss": 1.6113,
+      "step": 2798
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017589397253860595,
+      "loss": 1.7441,
+      "step": 2799
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017587520682535436,
+      "loss": 1.6309,
+      "step": 2800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017585643481258062,
+      "loss": 1.7559,
+      "step": 2801
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001758376565018433,
+      "loss": 1.7617,
+      "step": 2802
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017581887189470145,
+      "loss": 1.8145,
+      "step": 2803
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017580008099271462,
+      "loss": 1.5996,
+      "step": 2804
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017578128379744302,
+      "loss": 1.6777,
+      "step": 2805
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017576248031044722,
+      "loss": 1.6934,
+      "step": 2806
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017574367053328835,
+      "loss": 1.5938,
+      "step": 2807
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017572485446752815,
+      "loss": 1.6094,
+      "step": 2808
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017570603211472875,
+      "loss": 1.6777,
+      "step": 2809
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017568720347645294,
+      "loss": 1.6758,
+      "step": 2810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017566836855426394,
+      "loss": 1.7031,
+      "step": 2811
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001756495273497255,
+      "loss": 1.6816,
+      "step": 2812
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017563067986440192,
+      "loss": 1.6445,
+      "step": 2813
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00175611826099858,
+      "loss": 1.4785,
+      "step": 2814
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017559296605765904,
+      "loss": 1.7383,
+      "step": 2815
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017557409973937097,
+      "loss": 1.666,
+      "step": 2816
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001755552271465601,
+      "loss": 1.5703,
+      "step": 2817
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017553634828079332,
+      "loss": 1.6602,
+      "step": 2818
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017551746314363808,
+      "loss": 1.7539,
+      "step": 2819
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001754985717366623,
+      "loss": 1.6309,
+      "step": 2820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017547967406143442,
+      "loss": 1.5566,
+      "step": 2821
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017546077011952343,
+      "loss": 1.5801,
+      "step": 2822
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017544185991249885,
+      "loss": 1.6953,
+      "step": 2823
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017542294344193065,
+      "loss": 1.5332,
+      "step": 2824
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017540402070938936,
+      "loss": 1.7031,
+      "step": 2825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017538509171644609,
+      "loss": 1.7031,
+      "step": 2826
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017536615646467238,
+      "loss": 1.5957,
+      "step": 2827
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017534721495564031,
+      "loss": 1.6445,
+      "step": 2828
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017532826719092252,
+      "loss": 1.6035,
+      "step": 2829
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017530931317209216,
+      "loss": 1.7402,
+      "step": 2830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017529035290072283,
+      "loss": 1.7129,
+      "step": 2831
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017527138637838874,
+      "loss": 1.7305,
+      "step": 2832
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017525241360666461,
+      "loss": 1.625,
+      "step": 2833
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001752334345871256,
+      "loss": 1.6641,
+      "step": 2834
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017521444932134742,
+      "loss": 1.7012,
+      "step": 2835
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017519545781090639,
+      "loss": 1.6582,
+      "step": 2836
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017517646005737921,
+      "loss": 1.6172,
+      "step": 2837
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017515745606234323,
+      "loss": 1.5059,
+      "step": 2838
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001751384458273762,
+      "loss": 1.6621,
+      "step": 2839
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017511942935405642,
+      "loss": 1.5723,
+      "step": 2840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017510040664396281,
+      "loss": 1.6387,
+      "step": 2841
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017508137769867466,
+      "loss": 1.7129,
+      "step": 2842
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017506234251977184,
+      "loss": 1.7129,
+      "step": 2843
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017504330110883481,
+      "loss": 1.627,
+      "step": 2844
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001750242534674444,
+      "loss": 1.6152,
+      "step": 2845
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001750051995971821,
+      "loss": 1.6445,
+      "step": 2846
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001749861394996298,
+      "loss": 1.6328,
+      "step": 2847
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017496707317637,
+      "loss": 1.6836,
+      "step": 2848
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017494800062898563,
+      "loss": 1.7109,
+      "step": 2849
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017492892185906022,
+      "loss": 1.7051,
+      "step": 2850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017490983686817782,
+      "loss": 1.5859,
+      "step": 2851
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017489074565792287,
+      "loss": 1.6172,
+      "step": 2852
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017487164822988048,
+      "loss": 1.7656,
+      "step": 2853
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017485254458563617,
+      "loss": 1.6777,
+      "step": 2854
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017483343472677604,
+      "loss": 1.6133,
+      "step": 2855
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017481431865488668,
+      "loss": 1.7188,
+      "step": 2856
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017479519637155516,
+      "loss": 1.5098,
+      "step": 2857
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017477606787836917,
+      "loss": 1.8203,
+      "step": 2858
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017475693317691682,
+      "loss": 1.5723,
+      "step": 2859
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017473779226878671,
+      "loss": 1.6074,
+      "step": 2860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001747186451555681,
+      "loss": 1.707,
+      "step": 2861
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001746994918388506,
+      "loss": 1.5703,
+      "step": 2862
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017468033232022446,
+      "loss": 1.6934,
+      "step": 2863
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017466116660128038,
+      "loss": 1.627,
+      "step": 2864
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017464199468360958,
+      "loss": 1.6172,
+      "step": 2865
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0017462281656880384,
+      "loss": 1.6211,
+      "step": 2866
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017460363225845535,
+      "loss": 1.7012,
+      "step": 2867
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017458444175415693,
+      "loss": 1.6387,
+      "step": 2868
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017456524505750184,
+      "loss": 1.6094,
+      "step": 2869
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017454604217008395,
+      "loss": 1.7891,
+      "step": 2870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017452683309349749,
+      "loss": 1.7402,
+      "step": 2871
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017450761782933735,
+      "loss": 1.7617,
+      "step": 2872
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017448839637919883,
+      "loss": 1.5996,
+      "step": 2873
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017446916874467781,
+      "loss": 1.668,
+      "step": 2874
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017444993492737064,
+      "loss": 1.7344,
+      "step": 2875
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017443069492887424,
+      "loss": 1.5703,
+      "step": 2876
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017441144875078598,
+      "loss": 1.5918,
+      "step": 2877
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017439219639470377,
+      "loss": 1.7715,
+      "step": 2878
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017437293786222606,
+      "loss": 1.6523,
+      "step": 2879
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001743536731549517,
+      "loss": 1.6953,
+      "step": 2880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001743344022744803,
+      "loss": 1.541,
+      "step": 2881
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017431512522241165,
+      "loss": 1.6426,
+      "step": 2882
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017429584200034632,
+      "loss": 1.6855,
+      "step": 2883
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017427655260988527,
+      "loss": 1.5508,
+      "step": 2884
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017425725705262998,
+      "loss": 1.6289,
+      "step": 2885
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001742379553301825,
+      "loss": 1.6699,
+      "step": 2886
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001742186474441453,
+      "loss": 1.6348,
+      "step": 2887
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017419933339612146,
+      "loss": 1.6484,
+      "step": 2888
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017418001318771448,
+      "loss": 1.623,
+      "step": 2889
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017416068682052847,
+      "loss": 1.7656,
+      "step": 2890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017414135429616798,
+      "loss": 1.6719,
+      "step": 2891
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017412201561623807,
+      "loss": 1.6523,
+      "step": 2892
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001741026707823443,
+      "loss": 1.7168,
+      "step": 2893
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017408331979609285,
+      "loss": 1.6465,
+      "step": 2894
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017406396265909026,
+      "loss": 1.627,
+      "step": 2895
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017404459937294368,
+      "loss": 1.5898,
+      "step": 2896
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017402522993926076,
+      "loss": 1.5938,
+      "step": 2897
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017400585435964962,
+      "loss": 1.6406,
+      "step": 2898
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001739864726357189,
+      "loss": 1.7148,
+      "step": 2899
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017396708476907778,
+      "loss": 1.6953,
+      "step": 2900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00173947690761336,
+      "loss": 1.7715,
+      "step": 2901
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017392829061410363,
+      "loss": 1.8301,
+      "step": 2902
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017390888432899142,
+      "loss": 1.6992,
+      "step": 2903
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017388947190761052,
+      "loss": 1.5059,
+      "step": 2904
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017387005335157274,
+      "loss": 1.7168,
+      "step": 2905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001738506286624902,
+      "loss": 1.7129,
+      "step": 2906
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017383119784197575,
+      "loss": 1.6113,
+      "step": 2907
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017381176089164253,
+      "loss": 1.7109,
+      "step": 2908
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017379231781310426,
+      "loss": 1.6016,
+      "step": 2909
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017377286860797531,
+      "loss": 1.7578,
+      "step": 2910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017375341327787038,
+      "loss": 1.5195,
+      "step": 2911
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017373395182440476,
+      "loss": 1.5547,
+      "step": 2912
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017371448424919422,
+      "loss": 1.5918,
+      "step": 2913
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017369501055385507,
+      "loss": 1.6074,
+      "step": 2914
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017367553074000405,
+      "loss": 1.5859,
+      "step": 2915
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017365604480925857,
+      "loss": 1.7129,
+      "step": 2916
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017363655276323633,
+      "loss": 1.6289,
+      "step": 2917
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017361705460355577,
+      "loss": 1.6699,
+      "step": 2918
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017359755033183562,
+      "loss": 1.6836,
+      "step": 2919
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001735780399496953,
+      "loss": 1.7266,
+      "step": 2920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017355852345875458,
+      "loss": 1.7383,
+      "step": 2921
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017353900086063386,
+      "loss": 1.7363,
+      "step": 2922
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017351947215695396,
+      "loss": 1.7188,
+      "step": 2923
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001734999373493363,
+      "loss": 1.6797,
+      "step": 2924
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017348039643940273,
+      "loss": 1.7422,
+      "step": 2925
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017346084942877562,
+      "loss": 1.748,
+      "step": 2926
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017344129631907785,
+      "loss": 1.6367,
+      "step": 2927
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017342173711193285,
+      "loss": 1.7363,
+      "step": 2928
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001734021718089645,
+      "loss": 1.752,
+      "step": 2929
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001733826004117972,
+      "loss": 1.5625,
+      "step": 2930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017336302292205582,
+      "loss": 1.5684,
+      "step": 2931
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017334343934136583,
+      "loss": 1.5762,
+      "step": 2932
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017332384967135312,
+      "loss": 1.6738,
+      "step": 2933
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017330425391364416,
+      "loss": 1.8047,
+      "step": 2934
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017328465206986588,
+      "loss": 1.584,
+      "step": 2935
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017326504414164568,
+      "loss": 1.6855,
+      "step": 2936
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017324543013061148,
+      "loss": 1.6328,
+      "step": 2937
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017322581003839184,
+      "loss": 1.5742,
+      "step": 2938
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001732061838666156,
+      "loss": 1.5957,
+      "step": 2939
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017318655161691228,
+      "loss": 1.7305,
+      "step": 2940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017316691329091183,
+      "loss": 1.7324,
+      "step": 2941
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001731472688902447,
+      "loss": 1.5898,
+      "step": 2942
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017312761841654187,
+      "loss": 1.6816,
+      "step": 2943
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017310796187143486,
+      "loss": 1.4805,
+      "step": 2944
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017308829925655558,
+      "loss": 1.6836,
+      "step": 2945
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017306863057353654,
+      "loss": 1.627,
+      "step": 2946
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017304895582401078,
+      "loss": 1.7051,
+      "step": 2947
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017302927500961172,
+      "loss": 1.5469,
+      "step": 2948
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017300958813197336,
+      "loss": 1.6113,
+      "step": 2949
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017298989519273025,
+      "loss": 1.7344,
+      "step": 2950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017297019619351738,
+      "loss": 1.5566,
+      "step": 2951
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017295049113597019,
+      "loss": 1.6055,
+      "step": 2952
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017293078002172475,
+      "loss": 1.6094,
+      "step": 2953
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017291106285241755,
+      "loss": 1.666,
+      "step": 2954
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017289133962968561,
+      "loss": 1.5723,
+      "step": 2955
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017287161035516648,
+      "loss": 1.5801,
+      "step": 2956
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001728518750304981,
+      "loss": 1.7559,
+      "step": 2957
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017283213365731909,
+      "loss": 1.6855,
+      "step": 2958
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017281238623726835,
+      "loss": 1.6074,
+      "step": 2959
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017279263277198554,
+      "loss": 1.6367,
+      "step": 2960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001727728732631106,
+      "loss": 1.5195,
+      "step": 2961
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017275310771228405,
+      "loss": 1.6172,
+      "step": 2962
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00172733336121147,
+      "loss": 1.5957,
+      "step": 2963
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001727135584913409,
+      "loss": 1.6914,
+      "step": 2964
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017269377482450779,
+      "loss": 1.6973,
+      "step": 2965
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017267398512229026,
+      "loss": 1.7363,
+      "step": 2966
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017265418938633133,
+      "loss": 1.6445,
+      "step": 2967
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001726343876182745,
+      "loss": 1.6973,
+      "step": 2968
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017261457981976383,
+      "loss": 1.7168,
+      "step": 2969
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001725947659924439,
+      "loss": 1.541,
+      "step": 2970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017257494613795964,
+      "loss": 1.6367,
+      "step": 2971
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017255512025795667,
+      "loss": 1.5879,
+      "step": 2972
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017253528835408103,
+      "loss": 1.5879,
+      "step": 2973
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017251545042797924,
+      "loss": 1.7031,
+      "step": 2974
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0017249560648129837,
+      "loss": 1.6602,
+      "step": 2975
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001724757565156859,
+      "loss": 1.6133,
+      "step": 2976
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001724559005327899,
+      "loss": 1.6055,
+      "step": 2977
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001724360385342589,
+      "loss": 1.5195,
+      "step": 2978
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017241617052174197,
+      "loss": 1.5117,
+      "step": 2979
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017239629649688864,
+      "loss": 1.5312,
+      "step": 2980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001723764164613489,
+      "loss": 1.6035,
+      "step": 2981
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017235653041677336,
+      "loss": 1.6113,
+      "step": 2982
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017233663836481298,
+      "loss": 1.752,
+      "step": 2983
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001723167403071193,
+      "loss": 1.5996,
+      "step": 2984
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001722968362453444,
+      "loss": 1.5312,
+      "step": 2985
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017227692618114081,
+      "loss": 1.6348,
+      "step": 2986
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017225701011616152,
+      "loss": 1.5664,
+      "step": 2987
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017223708805206006,
+      "loss": 1.5996,
+      "step": 2988
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017221715999049048,
+      "loss": 1.7207,
+      "step": 2989
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017219722593310727,
+      "loss": 1.5801,
+      "step": 2990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017217728588156552,
+      "loss": 1.6094,
+      "step": 2991
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017215733983752064,
+      "loss": 1.6074,
+      "step": 2992
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017213738780262875,
+      "loss": 1.6211,
+      "step": 2993
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001721174297785463,
+      "loss": 1.6289,
+      "step": 2994
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001720974657669303,
+      "loss": 1.6035,
+      "step": 2995
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017207749576943832,
+      "loss": 1.6113,
+      "step": 2996
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017205751978772827,
+      "loss": 1.5918,
+      "step": 2997
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017203753782345874,
+      "loss": 1.6738,
+      "step": 2998
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017201754987828867,
+      "loss": 1.6152,
+      "step": 2999
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001719975559538776,
+      "loss": 1.5059,
+      "step": 3000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017197755605188548,
+      "loss": 1.6836,
+      "step": 3001
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017195755017397284,
+      "loss": 1.7324,
+      "step": 3002
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017193753832180061,
+      "loss": 1.5469,
+      "step": 3003
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001719175204970303,
+      "loss": 1.6504,
+      "step": 3004
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017189749670132386,
+      "loss": 1.6777,
+      "step": 3005
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017187746693634383,
+      "loss": 1.6953,
+      "step": 3006
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017185743120375309,
+      "loss": 1.6895,
+      "step": 3007
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017183738950521518,
+      "loss": 1.668,
+      "step": 3008
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00171817341842394,
+      "loss": 1.668,
+      "step": 3009
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017179728821695402,
+      "loss": 1.6895,
+      "step": 3010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017177722863056018,
+      "loss": 1.5938,
+      "step": 3011
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017175716308487794,
+      "loss": 1.5957,
+      "step": 3012
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017173709158157325,
+      "loss": 1.5957,
+      "step": 3013
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001717170141223125,
+      "loss": 1.6289,
+      "step": 3014
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017169693070876266,
+      "loss": 1.5527,
+      "step": 3015
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017167684134259112,
+      "loss": 1.4941,
+      "step": 3016
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001716567460254658,
+      "loss": 1.668,
+      "step": 3017
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017163664475905512,
+      "loss": 1.7129,
+      "step": 3018
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00171616537545028,
+      "loss": 1.7109,
+      "step": 3019
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017159642438505376,
+      "loss": 1.668,
+      "step": 3020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001715763052808024,
+      "loss": 1.6914,
+      "step": 3021
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001715561802339442,
+      "loss": 1.541,
+      "step": 3022
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017153604924615015,
+      "loss": 1.6621,
+      "step": 3023
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001715159123190915,
+      "loss": 1.7559,
+      "step": 3024
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001714957694544402,
+      "loss": 1.6328,
+      "step": 3025
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017147562065386857,
+      "loss": 1.6387,
+      "step": 3026
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017145546591904948,
+      "loss": 1.543,
+      "step": 3027
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017143530525165628,
+      "loss": 1.5918,
+      "step": 3028
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017141513865336275,
+      "loss": 1.6992,
+      "step": 3029
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017139496612584325,
+      "loss": 1.8379,
+      "step": 3030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017137478767077262,
+      "loss": 1.6543,
+      "step": 3031
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017135460328982614,
+      "loss": 1.707,
+      "step": 3032
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017133441298467964,
+      "loss": 1.6973,
+      "step": 3033
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001713142167570094,
+      "loss": 1.7363,
+      "step": 3034
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017129401460849222,
+      "loss": 1.75,
+      "step": 3035
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017127380654080536,
+      "loss": 1.6055,
+      "step": 3036
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017125359255562658,
+      "loss": 1.7266,
+      "step": 3037
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017123337265463419,
+      "loss": 1.7344,
+      "step": 3038
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017121314683950692,
+      "loss": 1.6543,
+      "step": 3039
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017119291511192398,
+      "loss": 1.6133,
+      "step": 3040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017117267747356517,
+      "loss": 1.6133,
+      "step": 3041
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017115243392611067,
+      "loss": 1.6211,
+      "step": 3042
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001711321844712412,
+      "loss": 1.6504,
+      "step": 3043
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017111192911063795,
+      "loss": 1.6875,
+      "step": 3044
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017109166784598266,
+      "loss": 1.6777,
+      "step": 3045
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017107140067895747,
+      "loss": 1.5527,
+      "step": 3046
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001710511276112451,
+      "loss": 1.5684,
+      "step": 3047
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017103084864452868,
+      "loss": 1.6855,
+      "step": 3048
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001710105637804919,
+      "loss": 1.6973,
+      "step": 3049
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017099027302081892,
+      "loss": 1.6523,
+      "step": 3050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001709699763671943,
+      "loss": 1.6426,
+      "step": 3051
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017094967382130324,
+      "loss": 1.6582,
+      "step": 3052
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017092936538483131,
+      "loss": 1.6582,
+      "step": 3053
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017090905105946463,
+      "loss": 1.6152,
+      "step": 3054
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017088873084688983,
+      "loss": 1.8848,
+      "step": 3055
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017086840474879392,
+      "loss": 1.543,
+      "step": 3056
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017084807276686452,
+      "loss": 1.6309,
+      "step": 3057
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017082773490278964,
+      "loss": 1.7285,
+      "step": 3058
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001708073911582579,
+      "loss": 1.6973,
+      "step": 3059
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017078704153495826,
+      "loss": 1.7461,
+      "step": 3060
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017076668603458029,
+      "loss": 1.6523,
+      "step": 3061
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017074632465881399,
+      "loss": 1.668,
+      "step": 3062
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017072595740934987,
+      "loss": 1.7168,
+      "step": 3063
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017070558428787885,
+      "loss": 1.6934,
+      "step": 3064
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001706852052960925,
+      "loss": 1.5742,
+      "step": 3065
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017066482043568275,
+      "loss": 1.5547,
+      "step": 3066
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00170644429708342,
+      "loss": 1.5059,
+      "step": 3067
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017062403311576323,
+      "loss": 1.6973,
+      "step": 3068
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017060363065963986,
+      "loss": 1.6094,
+      "step": 3069
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001705832223416658,
+      "loss": 1.6914,
+      "step": 3070
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017056280816353541,
+      "loss": 1.6465,
+      "step": 3071
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017054238812694365,
+      "loss": 1.7559,
+      "step": 3072
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001705219622335858,
+      "loss": 1.6797,
+      "step": 3073
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017050153048515778,
+      "loss": 1.707,
+      "step": 3074
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017048109288335586,
+      "loss": 1.7598,
+      "step": 3075
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00170460649429877,
+      "loss": 1.6797,
+      "step": 3076
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017044020012641835,
+      "loss": 1.5391,
+      "step": 3077
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017041974497467782,
+      "loss": 1.5215,
+      "step": 3078
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017039928397635368,
+      "loss": 1.5742,
+      "step": 3079
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017037881713314468,
+      "loss": 1.6543,
+      "step": 3080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017035834444675002,
+      "loss": 1.6953,
+      "step": 3081
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017033786591886956,
+      "loss": 1.582,
+      "step": 3082
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001703173815512034,
+      "loss": 1.5625,
+      "step": 3083
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017029689134545236,
+      "loss": 1.6465,
+      "step": 3084
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017027639530331755,
+      "loss": 1.707,
+      "step": 3085
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017025589342650074,
+      "loss": 1.6113,
+      "step": 3086
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00170235385716704,
+      "loss": 1.6523,
+      "step": 3087
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017021487217563002,
+      "loss": 1.582,
+      "step": 3088
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017019435280498194,
+      "loss": 1.7754,
+      "step": 3089
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017017382760646336,
+      "loss": 1.6973,
+      "step": 3090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0017015329658177838,
+      "loss": 1.541,
+      "step": 3091
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001701327597326316,
+      "loss": 1.6641,
+      "step": 3092
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0017011221706072807,
+      "loss": 1.5254,
+      "step": 3093
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0017009166856777336,
+      "loss": 1.8008,
+      "step": 3094
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0017007111425547349,
+      "loss": 1.7422,
+      "step": 3095
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0017005055412553497,
+      "loss": 1.6172,
+      "step": 3096
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001700299881796648,
+      "loss": 1.668,
+      "step": 3097
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001700094164195705,
+      "loss": 1.5586,
+      "step": 3098
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016998883884696,
+      "loss": 1.5664,
+      "step": 3099
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016996825546354173,
+      "loss": 1.5527,
+      "step": 3100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016994766627102466,
+      "loss": 1.5371,
+      "step": 3101
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001699270712711182,
+      "loss": 1.6543,
+      "step": 3102
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016990647046553225,
+      "loss": 1.6738,
+      "step": 3103
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016988586385597716,
+      "loss": 1.5449,
+      "step": 3104
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016986525144416379,
+      "loss": 1.6621,
+      "step": 3105
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001698446332318035,
+      "loss": 1.6797,
+      "step": 3106
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001698240092206081,
+      "loss": 1.5859,
+      "step": 3107
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016980337941228992,
+      "loss": 1.6484,
+      "step": 3108
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016978274380856169,
+      "loss": 1.7227,
+      "step": 3109
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016976210241113675,
+      "loss": 1.6953,
+      "step": 3110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016974145522172877,
+      "loss": 1.6445,
+      "step": 3111
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016972080224205205,
+      "loss": 1.6914,
+      "step": 3112
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016970014347382123,
+      "loss": 1.6602,
+      "step": 3113
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016967947891875156,
+      "loss": 1.5898,
+      "step": 3114
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001696588085785587,
+      "loss": 1.5488,
+      "step": 3115
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016963813245495878,
+      "loss": 1.6484,
+      "step": 3116
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016961745054966842,
+      "loss": 1.5664,
+      "step": 3117
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016959676286440475,
+      "loss": 1.6953,
+      "step": 3118
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016957606940088536,
+      "loss": 1.709,
+      "step": 3119
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016955537016082832,
+      "loss": 1.8027,
+      "step": 3120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016953466514595217,
+      "loss": 1.5625,
+      "step": 3121
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016951395435797596,
+      "loss": 1.6113,
+      "step": 3122
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016949323779861915,
+      "loss": 1.6152,
+      "step": 3123
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016947251546960184,
+      "loss": 1.7695,
+      "step": 3124
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016945178737264436,
+      "loss": 1.6816,
+      "step": 3125
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001694310535094677,
+      "loss": 1.6836,
+      "step": 3126
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016941031388179332,
+      "loss": 1.7324,
+      "step": 3127
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016938956849134313,
+      "loss": 1.7773,
+      "step": 3128
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016936881733983947,
+      "loss": 1.6387,
+      "step": 3129
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016934806042900518,
+      "loss": 1.6621,
+      "step": 3130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016932729776056366,
+      "loss": 1.6406,
+      "step": 3131
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001693065293362387,
+      "loss": 1.6934,
+      "step": 3132
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016928575515775459,
+      "loss": 1.6152,
+      "step": 3133
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016926497522683607,
+      "loss": 1.6309,
+      "step": 3134
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001692441895452085,
+      "loss": 1.623,
+      "step": 3135
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016922339811459746,
+      "loss": 1.6562,
+      "step": 3136
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016920260093672927,
+      "loss": 1.5664,
+      "step": 3137
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016918179801333052,
+      "loss": 1.6973,
+      "step": 3138
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016916098934612843,
+      "loss": 1.6484,
+      "step": 3139
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001691401749368506,
+      "loss": 1.6387,
+      "step": 3140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016911935478722518,
+      "loss": 1.625,
+      "step": 3141
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016909852889898072,
+      "loss": 1.7637,
+      "step": 3142
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016907769727384633,
+      "loss": 1.6621,
+      "step": 3143
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016905685991355148,
+      "loss": 1.7578,
+      "step": 3144
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016903601681982627,
+      "loss": 1.6895,
+      "step": 3145
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016901516799440112,
+      "loss": 1.623,
+      "step": 3146
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016899431343900706,
+      "loss": 1.7363,
+      "step": 3147
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001689734531553755,
+      "loss": 1.7598,
+      "step": 3148
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016895258714523839,
+      "loss": 1.6719,
+      "step": 3149
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016893171541032806,
+      "loss": 1.6836,
+      "step": 3150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016891083795237745,
+      "loss": 1.5703,
+      "step": 3151
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016888995477311985,
+      "loss": 1.625,
+      "step": 3152
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016886906587428917,
+      "loss": 1.5938,
+      "step": 3153
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016884817125761957,
+      "loss": 1.5156,
+      "step": 3154
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016882727092484597,
+      "loss": 1.7988,
+      "step": 3155
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001688063648777035,
+      "loss": 1.5801,
+      "step": 3156
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016878545311792795,
+      "loss": 1.6348,
+      "step": 3157
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016876453564725546,
+      "loss": 1.7324,
+      "step": 3158
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016874361246742276,
+      "loss": 1.7207,
+      "step": 3159
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016872268358016692,
+      "loss": 1.6328,
+      "step": 3160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016870174898722563,
+      "loss": 1.6465,
+      "step": 3161
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016868080869033692,
+      "loss": 1.7676,
+      "step": 3162
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001686598626912394,
+      "loss": 1.6035,
+      "step": 3163
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016863891099167209,
+      "loss": 1.6309,
+      "step": 3164
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016861795359337445,
+      "loss": 1.5879,
+      "step": 3165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016859699049808655,
+      "loss": 1.6348,
+      "step": 3166
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001685760217075488,
+      "loss": 1.5234,
+      "step": 3167
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016855504722350216,
+      "loss": 1.7188,
+      "step": 3168
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016853406704768798,
+      "loss": 1.6582,
+      "step": 3169
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016851308118184813,
+      "loss": 1.6895,
+      "step": 3170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016849208962772502,
+      "loss": 1.5781,
+      "step": 3171
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016847109238706142,
+      "loss": 1.6211,
+      "step": 3172
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016845008946160067,
+      "loss": 1.707,
+      "step": 3173
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016842908085308648,
+      "loss": 1.6797,
+      "step": 3174
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016840806656326312,
+      "loss": 1.7285,
+      "step": 3175
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016838704659387527,
+      "loss": 1.6543,
+      "step": 3176
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001683660209466681,
+      "loss": 1.5859,
+      "step": 3177
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016834498962338734,
+      "loss": 1.625,
+      "step": 3178
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00168323952625779,
+      "loss": 1.6875,
+      "step": 3179
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016830290995558973,
+      "loss": 1.6309,
+      "step": 3180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016828186161456662,
+      "loss": 1.5742,
+      "step": 3181
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016826080760445712,
+      "loss": 1.666,
+      "step": 3182
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016823974792700934,
+      "loss": 1.6016,
+      "step": 3183
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016821868258397163,
+      "loss": 1.4922,
+      "step": 3184
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016819761157709308,
+      "loss": 1.6445,
+      "step": 3185
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016817653490812296,
+      "loss": 1.5918,
+      "step": 3186
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001681554525788113,
+      "loss": 1.6914,
+      "step": 3187
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016813436459090835,
+      "loss": 1.6797,
+      "step": 3188
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016811327094616497,
+      "loss": 1.5879,
+      "step": 3189
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016809217164633244,
+      "loss": 1.5645,
+      "step": 3190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016807106669316254,
+      "loss": 1.584,
+      "step": 3191
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001680499560884075,
+      "loss": 1.6191,
+      "step": 3192
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016802883983382006,
+      "loss": 1.6602,
+      "step": 3193
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016800771793115332,
+      "loss": 1.6562,
+      "step": 3194
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00167986590382161,
+      "loss": 1.6523,
+      "step": 3195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016796545718859713,
+      "loss": 1.6797,
+      "step": 3196
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016794431835221633,
+      "loss": 1.6484,
+      "step": 3197
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016792317387477366,
+      "loss": 1.5918,
+      "step": 3198
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001679020237580246,
+      "loss": 1.6523,
+      "step": 3199
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016788086800372517,
+      "loss": 1.6602,
+      "step": 3200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016785970661363182,
+      "loss": 1.6758,
+      "step": 3201
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016783853958950142,
+      "loss": 1.6367,
+      "step": 3202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0016781736693309145,
+      "loss": 1.6797,
+      "step": 3203
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016779618864615964,
+      "loss": 1.6406,
+      "step": 3204
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016777500473046443,
+      "loss": 1.5996,
+      "step": 3205
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016775381518776453,
+      "loss": 1.6621,
+      "step": 3206
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016773262001981925,
+      "loss": 1.5996,
+      "step": 3207
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016771141922838826,
+      "loss": 1.6836,
+      "step": 3208
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016769021281523183,
+      "loss": 1.6875,
+      "step": 3209
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016766900078211053,
+      "loss": 1.6406,
+      "step": 3210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016764778313078552,
+      "loss": 1.6445,
+      "step": 3211
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016762655986301842,
+      "loss": 1.7148,
+      "step": 3212
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016760533098057124,
+      "loss": 1.6406,
+      "step": 3213
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016758409648520657,
+      "loss": 1.6074,
+      "step": 3214
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001675628563786873,
+      "loss": 1.7148,
+      "step": 3215
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016754161066277699,
+      "loss": 1.7207,
+      "step": 3216
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016752035933923948,
+      "loss": 1.6562,
+      "step": 3217
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016749910240983921,
+      "loss": 1.6348,
+      "step": 3218
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016747783987634097,
+      "loss": 1.6621,
+      "step": 3219
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016745657174051015,
+      "loss": 1.502,
+      "step": 3220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016743529800411251,
+      "loss": 1.6094,
+      "step": 3221
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016741401866891424,
+      "loss": 1.6758,
+      "step": 3222
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001673927337366821,
+      "loss": 1.6367,
+      "step": 3223
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016737144320918332,
+      "loss": 1.5957,
+      "step": 3224
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016735014708818543,
+      "loss": 1.625,
+      "step": 3225
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016732884537545662,
+      "loss": 1.5195,
+      "step": 3226
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016730753807276545,
+      "loss": 1.6465,
+      "step": 3227
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016728622518188093,
+      "loss": 1.5957,
+      "step": 3228
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016726490670457253,
+      "loss": 1.627,
+      "step": 3229
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016724358264261026,
+      "loss": 1.6426,
+      "step": 3230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001672222529977645,
+      "loss": 1.6406,
+      "step": 3231
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016720091777180622,
+      "loss": 1.543,
+      "step": 3232
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001671795769665067,
+      "loss": 1.6777,
+      "step": 3233
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001671582305836378,
+      "loss": 1.5801,
+      "step": 3234
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016713687862497176,
+      "loss": 1.584,
+      "step": 3235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016711552109228132,
+      "loss": 1.6406,
+      "step": 3236
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016709415798733974,
+      "loss": 1.7461,
+      "step": 3237
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001670727893119206,
+      "loss": 1.5996,
+      "step": 3238
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001670514150677981,
+      "loss": 1.5957,
+      "step": 3239
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016703003525674683,
+      "loss": 1.7148,
+      "step": 3240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016700864988054183,
+      "loss": 1.7656,
+      "step": 3241
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016698725894095858,
+      "loss": 1.5605,
+      "step": 3242
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001669658624397731,
+      "loss": 1.5547,
+      "step": 3243
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016694446037876184,
+      "loss": 1.7168,
+      "step": 3244
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016692305275970165,
+      "loss": 1.7715,
+      "step": 3245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016690163958436993,
+      "loss": 1.5859,
+      "step": 3246
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001668802208545445,
+      "loss": 1.5527,
+      "step": 3247
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016685879657200365,
+      "loss": 1.7656,
+      "step": 3248
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016683736673852611,
+      "loss": 1.6758,
+      "step": 3249
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001668159313558911,
+      "loss": 1.6348,
+      "step": 3250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016679449042587829,
+      "loss": 1.7051,
+      "step": 3251
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001667730439502678,
+      "loss": 1.7344,
+      "step": 3252
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016675159193084022,
+      "loss": 1.7578,
+      "step": 3253
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001667301343693766,
+      "loss": 1.6074,
+      "step": 3254
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016670867126765843,
+      "loss": 1.6055,
+      "step": 3255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016668720262746773,
+      "loss": 1.5449,
+      "step": 3256
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016666572845058687,
+      "loss": 1.623,
+      "step": 3257
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016664424873879878,
+      "loss": 1.5352,
+      "step": 3258
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001666227634938868,
+      "loss": 1.6113,
+      "step": 3259
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016660127271763473,
+      "loss": 1.6543,
+      "step": 3260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016657977641182681,
+      "loss": 1.5293,
+      "step": 3261
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016655827457824784,
+      "loss": 1.6406,
+      "step": 3262
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016653676721868293,
+      "loss": 1.5957,
+      "step": 3263
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016651525433491778,
+      "loss": 1.709,
+      "step": 3264
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016649373592873841,
+      "loss": 1.7031,
+      "step": 3265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001664722120019315,
+      "loss": 1.5469,
+      "step": 3266
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016645068255628397,
+      "loss": 1.5469,
+      "step": 3267
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016642914759358333,
+      "loss": 1.8086,
+      "step": 3268
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016640760711561758,
+      "loss": 1.6914,
+      "step": 3269
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016638606112417498,
+      "loss": 1.6387,
+      "step": 3270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016636450962104448,
+      "loss": 1.7891,
+      "step": 3271
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001663429526080154,
+      "loss": 1.5469,
+      "step": 3272
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001663213900868774,
+      "loss": 1.5273,
+      "step": 3273
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016629982205942082,
+      "loss": 1.5879,
+      "step": 3274
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016627824852743628,
+      "loss": 1.6426,
+      "step": 3275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016625666949271491,
+      "loss": 1.7324,
+      "step": 3276
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016623508495704838,
+      "loss": 1.5469,
+      "step": 3277
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016621349492222863,
+      "loss": 1.6426,
+      "step": 3278
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016619189939004824,
+      "loss": 1.5547,
+      "step": 3279
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016617029836230018,
+      "loss": 1.6113,
+      "step": 3280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016614869184077784,
+      "loss": 1.5918,
+      "step": 3281
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001661270798272751,
+      "loss": 1.5449,
+      "step": 3282
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016610546232358628,
+      "loss": 1.6914,
+      "step": 3283
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001660838393315062,
+      "loss": 1.707,
+      "step": 3284
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016606221085283008,
+      "loss": 1.7773,
+      "step": 3285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016604057688935365,
+      "loss": 1.6836,
+      "step": 3286
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016601893744287307,
+      "loss": 1.7129,
+      "step": 3287
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016599729251518488,
+      "loss": 1.582,
+      "step": 3288
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001659756421080862,
+      "loss": 1.5547,
+      "step": 3289
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016595398622337454,
+      "loss": 1.6738,
+      "step": 3290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016593232486284788,
+      "loss": 1.5566,
+      "step": 3291
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016591065802830465,
+      "loss": 1.6719,
+      "step": 3292
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016588898572154373,
+      "loss": 1.6055,
+      "step": 3293
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016586730794436446,
+      "loss": 1.6445,
+      "step": 3294
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016584562469856663,
+      "loss": 1.584,
+      "step": 3295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001658239359859505,
+      "loss": 1.627,
+      "step": 3296
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016580224180831674,
+      "loss": 1.6328,
+      "step": 3297
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016578054216746653,
+      "loss": 1.6523,
+      "step": 3298
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016575883706520145,
+      "loss": 1.7461,
+      "step": 3299
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001657371265033236,
+      "loss": 1.7617,
+      "step": 3300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016571541048363548,
+      "loss": 1.6328,
+      "step": 3301
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016569368900794005,
+      "loss": 1.4863,
+      "step": 3302
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016567196207804074,
+      "loss": 1.7305,
+      "step": 3303
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001656502296957414,
+      "loss": 1.7031,
+      "step": 3304
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001656284918628464,
+      "loss": 1.6211,
+      "step": 3305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016560674858116045,
+      "loss": 1.582,
+      "step": 3306
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016558499985248885,
+      "loss": 1.5469,
+      "step": 3307
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016556324567863726,
+      "loss": 1.6816,
+      "step": 3308
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016554148606141182,
+      "loss": 1.6289,
+      "step": 3309
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016551972100261908,
+      "loss": 1.5293,
+      "step": 3310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016549795050406616,
+      "loss": 1.6289,
+      "step": 3311
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016547617456756046,
+      "loss": 1.6172,
+      "step": 3312
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016545439319490996,
+      "loss": 1.5859,
+      "step": 3313
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016543260638792305,
+      "loss": 1.6289,
+      "step": 3314
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0016541081414840862,
+      "loss": 1.5586,
+      "step": 3315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001653890164781759,
+      "loss": 1.6738,
+      "step": 3316
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016536721337903466,
+      "loss": 1.6973,
+      "step": 3317
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016534540485279511,
+      "loss": 1.5996,
+      "step": 3318
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016532359090126792,
+      "loss": 1.6074,
+      "step": 3319
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016530177152626411,
+      "loss": 1.6836,
+      "step": 3320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001652799467295953,
+      "loss": 1.6816,
+      "step": 3321
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016525811651307346,
+      "loss": 1.5684,
+      "step": 3322
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016523628087851104,
+      "loss": 1.5977,
+      "step": 3323
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016521443982772093,
+      "loss": 1.6875,
+      "step": 3324
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001651925933625165,
+      "loss": 1.7793,
+      "step": 3325
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016517074148471153,
+      "loss": 1.8438,
+      "step": 3326
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001651488841961203,
+      "loss": 1.5762,
+      "step": 3327
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016512702149855747,
+      "loss": 1.6094,
+      "step": 3328
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016510515339383822,
+      "loss": 1.5762,
+      "step": 3329
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016508327988377807,
+      "loss": 1.5703,
+      "step": 3330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016506140097019319,
+      "loss": 1.7266,
+      "step": 3331
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016503951665489992,
+      "loss": 1.6348,
+      "step": 3332
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016501762693971532,
+      "loss": 1.6309,
+      "step": 3333
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016499573182645673,
+      "loss": 1.5352,
+      "step": 3334
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016497383131694199,
+      "loss": 1.666,
+      "step": 3335
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001649519254129894,
+      "loss": 1.6602,
+      "step": 3336
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016493001411641765,
+      "loss": 1.584,
+      "step": 3337
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00164908097429046,
+      "loss": 1.627,
+      "step": 3338
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016488617535269402,
+      "loss": 1.8262,
+      "step": 3339
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016486424788918176,
+      "loss": 1.582,
+      "step": 3340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001648423150403298,
+      "loss": 1.6641,
+      "step": 3341
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016482037680795908,
+      "loss": 1.748,
+      "step": 3342
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016479843319389103,
+      "loss": 1.5566,
+      "step": 3343
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001647764841999475,
+      "loss": 1.6719,
+      "step": 3344
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016475452982795086,
+      "loss": 1.498,
+      "step": 3345
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016473257007972376,
+      "loss": 1.6211,
+      "step": 3346
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016471060495708947,
+      "loss": 1.625,
+      "step": 3347
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016468863446187166,
+      "loss": 1.8262,
+      "step": 3348
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016466665859589437,
+      "loss": 1.6992,
+      "step": 3349
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016464467736098214,
+      "loss": 1.6426,
+      "step": 3350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016462269075896002,
+      "loss": 1.6992,
+      "step": 3351
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016460069879165337,
+      "loss": 1.6113,
+      "step": 3352
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016457870146088813,
+      "loss": 1.7422,
+      "step": 3353
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001645566987684906,
+      "loss": 1.6562,
+      "step": 3354
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016453469071628754,
+      "loss": 1.7637,
+      "step": 3355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016451267730610618,
+      "loss": 1.7246,
+      "step": 3356
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016449065853977414,
+      "loss": 1.6172,
+      "step": 3357
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016446863441911957,
+      "loss": 1.541,
+      "step": 3358
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016444660494597102,
+      "loss": 1.6367,
+      "step": 3359
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001644245701221574,
+      "loss": 1.6914,
+      "step": 3360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016440252994950825,
+      "loss": 1.6992,
+      "step": 3361
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001643804844298534,
+      "loss": 1.6875,
+      "step": 3362
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016435843356502318,
+      "loss": 1.7305,
+      "step": 3363
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016433637735684835,
+      "loss": 1.6836,
+      "step": 3364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016431431580716015,
+      "loss": 1.6172,
+      "step": 3365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016429224891779022,
+      "loss": 1.6465,
+      "step": 3366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016427017669057064,
+      "loss": 1.6367,
+      "step": 3367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016424809912733397,
+      "loss": 1.5098,
+      "step": 3368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001642260162299132,
+      "loss": 1.582,
+      "step": 3369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016420392800014173,
+      "loss": 1.5723,
+      "step": 3370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016418183443985348,
+      "loss": 1.6484,
+      "step": 3371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001641597355508827,
+      "loss": 1.5898,
+      "step": 3372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001641376313350642,
+      "loss": 1.5508,
+      "step": 3373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016411552179423315,
+      "loss": 1.6426,
+      "step": 3374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001640934069302252,
+      "loss": 1.6445,
+      "step": 3375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016407128674487642,
+      "loss": 1.7031,
+      "step": 3376
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016404916124002336,
+      "loss": 1.7188,
+      "step": 3377
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016402703041750292,
+      "loss": 1.7344,
+      "step": 3378
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001640048942791526,
+      "loss": 1.6445,
+      "step": 3379
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016398275282681017,
+      "loss": 1.5488,
+      "step": 3380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016396060606231396,
+      "loss": 1.6895,
+      "step": 3381
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001639384539875027,
+      "loss": 1.5977,
+      "step": 3382
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016391629660421553,
+      "loss": 1.6914,
+      "step": 3383
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016389413391429209,
+      "loss": 1.6562,
+      "step": 3384
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016387196591957243,
+      "loss": 1.5391,
+      "step": 3385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016384979262189705,
+      "loss": 1.5137,
+      "step": 3386
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016382761402310684,
+      "loss": 1.791,
+      "step": 3387
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016380543012504325,
+      "loss": 1.6738,
+      "step": 3388
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016378324092954802,
+      "loss": 1.6133,
+      "step": 3389
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016376104643846343,
+      "loss": 1.6504,
+      "step": 3390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016373884665363218,
+      "loss": 1.7188,
+      "step": 3391
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016371664157689737,
+      "loss": 1.6973,
+      "step": 3392
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016369443121010262,
+      "loss": 1.627,
+      "step": 3393
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016367221555509191,
+      "loss": 1.6172,
+      "step": 3394
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016364999461370969,
+      "loss": 1.6816,
+      "step": 3395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016362776838780087,
+      "loss": 1.6016,
+      "step": 3396
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016360553687921073,
+      "loss": 1.5723,
+      "step": 3397
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016358330008978512,
+      "loss": 1.5918,
+      "step": 3398
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016356105802137018,
+      "loss": 1.5879,
+      "step": 3399
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016353881067581255,
+      "loss": 1.752,
+      "step": 3400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016351655805495933,
+      "loss": 1.7148,
+      "step": 3401
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016349430016065806,
+      "loss": 1.6152,
+      "step": 3402
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016347203699475664,
+      "loss": 1.625,
+      "step": 3403
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016344976855910353,
+      "loss": 1.6602,
+      "step": 3404
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001634274948555475,
+      "loss": 1.6074,
+      "step": 3405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016340521588593785,
+      "loss": 1.7559,
+      "step": 3406
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001633829316521243,
+      "loss": 1.6367,
+      "step": 3407
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00163360642155957,
+      "loss": 1.627,
+      "step": 3408
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016333834739928647,
+      "loss": 1.6641,
+      "step": 3409
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016331604738396377,
+      "loss": 1.5762,
+      "step": 3410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016329374211184037,
+      "loss": 1.7129,
+      "step": 3411
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016327143158476813,
+      "loss": 1.6992,
+      "step": 3412
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016324911580459938,
+      "loss": 1.6367,
+      "step": 3413
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001632267947731869,
+      "loss": 1.6992,
+      "step": 3414
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016320446849238386,
+      "loss": 1.668,
+      "step": 3415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016318213696404397,
+      "loss": 1.7031,
+      "step": 3416
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016315980019002117,
+      "loss": 1.6191,
+      "step": 3417
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016313745817217006,
+      "loss": 1.8066,
+      "step": 3418
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016311511091234559,
+      "loss": 1.6602,
+      "step": 3419
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016309275841240307,
+      "loss": 1.7207,
+      "step": 3420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016307040067419837,
+      "loss": 1.6758,
+      "step": 3421
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016304803769958771,
+      "loss": 1.5527,
+      "step": 3422
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016302566949042776,
+      "loss": 1.6777,
+      "step": 3423
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016300329604857567,
+      "loss": 1.7305,
+      "step": 3424
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016298091737588897,
+      "loss": 1.623,
+      "step": 3425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0016295853347422563,
+      "loss": 1.5898,
+      "step": 3426
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001629361443454441,
+      "loss": 1.6191,
+      "step": 3427
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001629137499914032,
+      "loss": 1.6699,
+      "step": 3428
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016289135041396223,
+      "loss": 1.6992,
+      "step": 3429
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016286894561498091,
+      "loss": 1.627,
+      "step": 3430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001628465355963194,
+      "loss": 1.668,
+      "step": 3431
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001628241203598383,
+      "loss": 1.7109,
+      "step": 3432
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001628016999073986,
+      "loss": 1.6855,
+      "step": 3433
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016277927424086177,
+      "loss": 1.8164,
+      "step": 3434
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016275684336208966,
+      "loss": 1.6133,
+      "step": 3435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016273440727294463,
+      "loss": 1.584,
+      "step": 3436
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016271196597528945,
+      "loss": 1.6367,
+      "step": 3437
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016268951947098725,
+      "loss": 1.5664,
+      "step": 3438
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016266706776190169,
+      "loss": 1.625,
+      "step": 3439
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016264461084989678,
+      "loss": 1.5996,
+      "step": 3440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00162622148736837,
+      "loss": 1.5566,
+      "step": 3441
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016259968142458732,
+      "loss": 1.7051,
+      "step": 3442
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016257720891501301,
+      "loss": 1.7461,
+      "step": 3443
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001625547312099799,
+      "loss": 1.6152,
+      "step": 3444
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016253224831135418,
+      "loss": 1.7051,
+      "step": 3445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016250976022100247,
+      "loss": 1.6562,
+      "step": 3446
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016248726694079187,
+      "loss": 1.6289,
+      "step": 3447
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016246476847258987,
+      "loss": 1.6289,
+      "step": 3448
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016244226481826436,
+      "loss": 1.6875,
+      "step": 3449
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016241975597968375,
+      "loss": 1.5879,
+      "step": 3450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001623972419587168,
+      "loss": 1.5625,
+      "step": 3451
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016237472275723278,
+      "loss": 1.6328,
+      "step": 3452
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016235219837710127,
+      "loss": 1.6504,
+      "step": 3453
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001623296688201924,
+      "loss": 1.6016,
+      "step": 3454
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016230713408837668,
+      "loss": 1.5527,
+      "step": 3455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016228459418352503,
+      "loss": 1.7422,
+      "step": 3456
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016226204910750883,
+      "loss": 1.6934,
+      "step": 3457
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001622394988621999,
+      "loss": 1.6074,
+      "step": 3458
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016221694344947042,
+      "loss": 1.5273,
+      "step": 3459
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016219438287119311,
+      "loss": 1.584,
+      "step": 3460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00162171817129241,
+      "loss": 1.6504,
+      "step": 3461
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016214924622548767,
+      "loss": 1.6797,
+      "step": 3462
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00162126670161807,
+      "loss": 1.7441,
+      "step": 3463
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001621040889400734,
+      "loss": 1.6973,
+      "step": 3464
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016208150256216164,
+      "loss": 1.543,
+      "step": 3465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016205891102994698,
+      "loss": 1.5859,
+      "step": 3466
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001620363143453051,
+      "loss": 1.7031,
+      "step": 3467
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016201371251011198,
+      "loss": 1.5586,
+      "step": 3468
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016199110552624428,
+      "loss": 1.6484,
+      "step": 3469
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001619684933955788,
+      "loss": 1.6641,
+      "step": 3470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016194587611999302,
+      "loss": 1.7031,
+      "step": 3471
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016192325370136462,
+      "loss": 1.6328,
+      "step": 3472
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016190062614157195,
+      "loss": 1.6777,
+      "step": 3473
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016187799344249357,
+      "loss": 1.5938,
+      "step": 3474
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016185535560600856,
+      "loss": 1.5977,
+      "step": 3475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016183271263399647,
+      "loss": 1.668,
+      "step": 3476
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016181006452833718,
+      "loss": 1.5215,
+      "step": 3477
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016178741129091105,
+      "loss": 1.6504,
+      "step": 3478
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016176475292359889,
+      "loss": 1.5801,
+      "step": 3479
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016174208942828186,
+      "loss": 1.5605,
+      "step": 3480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016171942080684166,
+      "loss": 1.8066,
+      "step": 3481
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016169674706116028,
+      "loss": 1.6035,
+      "step": 3482
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016167406819312022,
+      "loss": 1.6914,
+      "step": 3483
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016165138420460444,
+      "loss": 1.5898,
+      "step": 3484
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016162869509749617,
+      "loss": 1.6133,
+      "step": 3485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016160600087367926,
+      "loss": 1.7148,
+      "step": 3486
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016158330153503786,
+      "loss": 1.5234,
+      "step": 3487
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001615605970834566,
+      "loss": 1.5586,
+      "step": 3488
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016153788752082045,
+      "loss": 1.6465,
+      "step": 3489
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001615151728490149,
+      "loss": 1.7305,
+      "step": 3490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016149245306992588,
+      "loss": 1.7637,
+      "step": 3491
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016146972818543964,
+      "loss": 1.7734,
+      "step": 3492
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016144699819744292,
+      "loss": 1.7422,
+      "step": 3493
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016142426310782284,
+      "loss": 1.6816,
+      "step": 3494
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016140152291846703,
+      "loss": 1.5332,
+      "step": 3495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016137877763126348,
+      "loss": 1.5098,
+      "step": 3496
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016135602724810059,
+      "loss": 1.541,
+      "step": 3497
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001613332717708672,
+      "loss": 1.6016,
+      "step": 3498
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016131051120145262,
+      "loss": 1.6797,
+      "step": 3499
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016128774554174648,
+      "loss": 1.6387,
+      "step": 3500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016126497479363895,
+      "loss": 1.6348,
+      "step": 3501
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016124219895902057,
+      "loss": 1.6992,
+      "step": 3502
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016121941803978225,
+      "loss": 1.6172,
+      "step": 3503
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016119663203781537,
+      "loss": 1.582,
+      "step": 3504
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016117384095501177,
+      "loss": 1.6074,
+      "step": 3505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016115104479326368,
+      "loss": 1.75,
+      "step": 3506
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016112824355446374,
+      "loss": 1.5352,
+      "step": 3507
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016110543724050497,
+      "loss": 1.623,
+      "step": 3508
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016108262585328092,
+      "loss": 1.7344,
+      "step": 3509
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016105980939468546,
+      "loss": 1.6328,
+      "step": 3510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016103698786661294,
+      "loss": 1.6348,
+      "step": 3511
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001610141612709581,
+      "loss": 1.8145,
+      "step": 3512
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016099132960961614,
+      "loss": 1.6641,
+      "step": 3513
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001609684928844826,
+      "loss": 1.7012,
+      "step": 3514
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001609456510974536,
+      "loss": 1.6836,
+      "step": 3515
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016092280425042543,
+      "loss": 1.5996,
+      "step": 3516
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016089995234529505,
+      "loss": 1.7539,
+      "step": 3517
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001608770953839597,
+      "loss": 1.6211,
+      "step": 3518
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016085423336831709,
+      "loss": 1.8418,
+      "step": 3519
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016083136630026527,
+      "loss": 1.6191,
+      "step": 3520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016080849418170287,
+      "loss": 1.7012,
+      "step": 3521
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016078561701452876,
+      "loss": 1.6699,
+      "step": 3522
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001607627348006424,
+      "loss": 1.7227,
+      "step": 3523
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016073984754194348,
+      "loss": 1.6953,
+      "step": 3524
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001607169552403323,
+      "loss": 1.6777,
+      "step": 3525
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016069405789770942,
+      "loss": 1.7539,
+      "step": 3526
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016067115551597591,
+      "loss": 1.6582,
+      "step": 3527
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016064824809703321,
+      "loss": 1.75,
+      "step": 3528
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001606253356427833,
+      "loss": 1.7461,
+      "step": 3529
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016060241815512835,
+      "loss": 1.7148,
+      "step": 3530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001605794956359712,
+      "loss": 1.7051,
+      "step": 3531
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016055656808721488,
+      "loss": 1.6602,
+      "step": 3532
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016053363551076303,
+      "loss": 1.6504,
+      "step": 3533
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016051069790851955,
+      "loss": 1.459,
+      "step": 3534
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001604877552823889,
+      "loss": 1.6562,
+      "step": 3535
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016046480763427583,
+      "loss": 1.709,
+      "step": 3536
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001604418549660856,
+      "loss": 1.4922,
+      "step": 3537
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001604188972797238,
+      "loss": 1.6934,
+      "step": 3538
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0016039593457709655,
+      "loss": 1.7305,
+      "step": 3539
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001603729668601103,
+      "loss": 1.5938,
+      "step": 3540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0016034999413067192,
+      "loss": 1.6367,
+      "step": 3541
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0016032701639068874,
+      "loss": 1.6309,
+      "step": 3542
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0016030403364206847,
+      "loss": 1.6738,
+      "step": 3543
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0016028104588671924,
+      "loss": 1.5215,
+      "step": 3544
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0016025805312654962,
+      "loss": 1.6992,
+      "step": 3545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0016023505536346856,
+      "loss": 1.5898,
+      "step": 3546
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0016021205259938549,
+      "loss": 1.6016,
+      "step": 3547
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0016018904483621015,
+      "loss": 1.6191,
+      "step": 3548
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0016016603207585278,
+      "loss": 1.7266,
+      "step": 3549
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00160143014320224,
+      "loss": 1.6738,
+      "step": 3550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0016011999157123488,
+      "loss": 1.7129,
+      "step": 3551
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0016009696383079686,
+      "loss": 1.6953,
+      "step": 3552
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0016007393110082182,
+      "loss": 1.6289,
+      "step": 3553
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0016005089338322204,
+      "loss": 1.6758,
+      "step": 3554
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0016002785067991022,
+      "loss": 1.7129,
+      "step": 3555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0016000480299279947,
+      "loss": 1.6328,
+      "step": 3556
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015998175032380334,
+      "loss": 1.7402,
+      "step": 3557
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015995869267483576,
+      "loss": 1.5156,
+      "step": 3558
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015993563004781106,
+      "loss": 1.6035,
+      "step": 3559
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015991256244464408,
+      "loss": 1.6934,
+      "step": 3560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001598894898672499,
+      "loss": 1.7051,
+      "step": 3561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015986641231754421,
+      "loss": 1.5293,
+      "step": 3562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00159843329797443,
+      "loss": 1.5156,
+      "step": 3563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015982024230886266,
+      "loss": 1.5664,
+      "step": 3564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015979714985372004,
+      "loss": 1.7559,
+      "step": 3565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015977405243393235,
+      "loss": 1.6895,
+      "step": 3566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001597509500514173,
+      "loss": 1.6348,
+      "step": 3567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015972784270809296,
+      "loss": 1.6055,
+      "step": 3568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015970473040587776,
+      "loss": 1.5996,
+      "step": 3569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015968161314669068,
+      "loss": 1.6934,
+      "step": 3570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015965849093245092,
+      "loss": 1.6191,
+      "step": 3571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001596353637650783,
+      "loss": 1.7109,
+      "step": 3572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015961223164649286,
+      "loss": 1.7734,
+      "step": 3573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015958909457861519,
+      "loss": 1.5996,
+      "step": 3574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001595659525633662,
+      "loss": 1.6953,
+      "step": 3575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015954280560266732,
+      "loss": 1.6777,
+      "step": 3576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015951965369844024,
+      "loss": 1.5723,
+      "step": 3577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015949649685260718,
+      "loss": 1.752,
+      "step": 3578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015947333506709075,
+      "loss": 1.5547,
+      "step": 3579
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015945016834381395,
+      "loss": 1.6367,
+      "step": 3580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015942699668470013,
+      "loss": 1.7715,
+      "step": 3581
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001594038200916732,
+      "loss": 1.5879,
+      "step": 3582
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015938063856665732,
+      "loss": 1.709,
+      "step": 3583
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015935745211157718,
+      "loss": 1.6543,
+      "step": 3584
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001593342607283578,
+      "loss": 1.5723,
+      "step": 3585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015931106441892467,
+      "loss": 1.4922,
+      "step": 3586
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001592878631852036,
+      "loss": 1.5703,
+      "step": 3587
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015926465702912093,
+      "loss": 1.5684,
+      "step": 3588
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015924144595260332,
+      "loss": 1.6211,
+      "step": 3589
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015921822995757786,
+      "loss": 1.7031,
+      "step": 3590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015919500904597206,
+      "loss": 1.5996,
+      "step": 3591
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015917178321971383,
+      "loss": 1.5391,
+      "step": 3592
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001591485524807315,
+      "loss": 1.6094,
+      "step": 3593
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001591253168309538,
+      "loss": 1.7207,
+      "step": 3594
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015910207627230979,
+      "loss": 1.5332,
+      "step": 3595
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015907883080672912,
+      "loss": 1.6074,
+      "step": 3596
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015905558043614167,
+      "loss": 1.6426,
+      "step": 3597
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015903232516247786,
+      "loss": 1.5781,
+      "step": 3598
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015900906498766837,
+      "loss": 1.584,
+      "step": 3599
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015898579991364442,
+      "loss": 1.6543,
+      "step": 3600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015896252994233757,
+      "loss": 1.502,
+      "step": 3601
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015893925507567985,
+      "loss": 1.6641,
+      "step": 3602
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015891597531560357,
+      "loss": 1.6055,
+      "step": 3603
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015889269066404164,
+      "loss": 1.7715,
+      "step": 3604
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015886940112292713,
+      "loss": 1.6797,
+      "step": 3605
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015884610669419374,
+      "loss": 1.6387,
+      "step": 3606
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015882280737977545,
+      "loss": 1.5879,
+      "step": 3607
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015879950318160674,
+      "loss": 1.5938,
+      "step": 3608
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001587761941016223,
+      "loss": 1.7051,
+      "step": 3609
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015875288014175753,
+      "loss": 1.6094,
+      "step": 3610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015872956130394793,
+      "loss": 1.5488,
+      "step": 3611
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015870623759012964,
+      "loss": 1.5703,
+      "step": 3612
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00158682909002239,
+      "loss": 1.6777,
+      "step": 3613
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015865957554221297,
+      "loss": 1.7324,
+      "step": 3614
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015863623721198872,
+      "loss": 1.6562,
+      "step": 3615
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00158612894013504,
+      "loss": 1.6602,
+      "step": 3616
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015858954594869677,
+      "loss": 1.5312,
+      "step": 3617
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015856619301950557,
+      "loss": 1.5645,
+      "step": 3618
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015854283522786924,
+      "loss": 1.627,
+      "step": 3619
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015851947257572707,
+      "loss": 1.6582,
+      "step": 3620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015849610506501874,
+      "loss": 1.6152,
+      "step": 3621
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015847273269768431,
+      "loss": 1.7637,
+      "step": 3622
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001584493554756643,
+      "loss": 1.7793,
+      "step": 3623
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015842597340089957,
+      "loss": 1.6602,
+      "step": 3624
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015840258647533143,
+      "loss": 1.8086,
+      "step": 3625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015837919470090159,
+      "loss": 1.5977,
+      "step": 3626
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015835579807955207,
+      "loss": 1.6562,
+      "step": 3627
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015833239661322547,
+      "loss": 1.6289,
+      "step": 3628
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001583089903038646,
+      "loss": 1.541,
+      "step": 3629
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015828557915341284,
+      "loss": 1.6602,
+      "step": 3630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015826216316381387,
+      "loss": 1.709,
+      "step": 3631
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015823874233701178,
+      "loss": 1.7109,
+      "step": 3632
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015821531667495109,
+      "loss": 1.6895,
+      "step": 3633
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015819188617957672,
+      "loss": 1.75,
+      "step": 3634
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015816845085283395,
+      "loss": 1.4961,
+      "step": 3635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015814501069666852,
+      "loss": 1.6504,
+      "step": 3636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015812156571302653,
+      "loss": 1.7188,
+      "step": 3637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015809811590385453,
+      "loss": 1.6582,
+      "step": 3638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001580746612710994,
+      "loss": 1.6719,
+      "step": 3639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015805120181670844,
+      "loss": 1.5156,
+      "step": 3640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015802773754262942,
+      "loss": 1.7012,
+      "step": 3641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001580042684508104,
+      "loss": 1.7871,
+      "step": 3642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015798079454319995,
+      "loss": 1.5332,
+      "step": 3643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015795731582174692,
+      "loss": 1.6504,
+      "step": 3644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001579338322884007,
+      "loss": 1.5312,
+      "step": 3645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015791034394511097,
+      "loss": 1.6113,
+      "step": 3646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015788685079382782,
+      "loss": 1.6543,
+      "step": 3647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001578633528365018,
+      "loss": 1.7012,
+      "step": 3648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001578398500750838,
+      "loss": 1.6484,
+      "step": 3649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015781634251152514,
+      "loss": 1.6855,
+      "step": 3650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015779283014777756,
+      "loss": 1.6426,
+      "step": 3651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001577693129857931,
+      "loss": 1.7578,
+      "step": 3652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0015774579102752436,
+      "loss": 1.7598,
+      "step": 3653
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015772226427492413,
+      "loss": 1.5312,
+      "step": 3654
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015769873272994581,
+      "loss": 1.7949,
+      "step": 3655
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015767519639454306,
+      "loss": 1.7988,
+      "step": 3656
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015765165527067,
+      "loss": 1.8379,
+      "step": 3657
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001576281093602811,
+      "loss": 1.5898,
+      "step": 3658
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015760455866533128,
+      "loss": 1.6055,
+      "step": 3659
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001575810031877758,
+      "loss": 1.7559,
+      "step": 3660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015755744292957035,
+      "loss": 1.6562,
+      "step": 3661
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015753387789267106,
+      "loss": 1.5938,
+      "step": 3662
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015751030807903436,
+      "loss": 1.666,
+      "step": 3663
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001574867334906172,
+      "loss": 1.6309,
+      "step": 3664
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015746315412937678,
+      "loss": 1.6582,
+      "step": 3665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015743956999727078,
+      "loss": 1.5723,
+      "step": 3666
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001574159810962573,
+      "loss": 1.6406,
+      "step": 3667
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015739238742829478,
+      "loss": 1.6738,
+      "step": 3668
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001573687889953421,
+      "loss": 1.6387,
+      "step": 3669
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015734518579935848,
+      "loss": 1.6289,
+      "step": 3670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001573215778423036,
+      "loss": 1.668,
+      "step": 3671
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015729796512613753,
+      "loss": 1.6387,
+      "step": 3672
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015727434765282066,
+      "loss": 1.5703,
+      "step": 3673
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015725072542431382,
+      "loss": 1.6113,
+      "step": 3674
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001572270984425783,
+      "loss": 1.6289,
+      "step": 3675
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015720346670957564,
+      "loss": 1.5527,
+      "step": 3676
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015717983022726797,
+      "loss": 1.6133,
+      "step": 3677
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015715618899761758,
+      "loss": 1.5488,
+      "step": 3678
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001571325430225874,
+      "loss": 1.5762,
+      "step": 3679
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015710889230414052,
+      "loss": 1.6543,
+      "step": 3680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015708523684424063,
+      "loss": 1.5938,
+      "step": 3681
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015706157664485162,
+      "loss": 1.5488,
+      "step": 3682
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00157037911707938,
+      "loss": 1.6387,
+      "step": 3683
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015701424203546442,
+      "loss": 1.752,
+      "step": 3684
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015699056762939613,
+      "loss": 1.6191,
+      "step": 3685
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015696688849169865,
+      "loss": 1.582,
+      "step": 3686
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015694320462433797,
+      "loss": 1.7266,
+      "step": 3687
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001569195160292804,
+      "loss": 1.5938,
+      "step": 3688
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001568958227084927,
+      "loss": 1.7207,
+      "step": 3689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015687212466394204,
+      "loss": 1.6816,
+      "step": 3690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015684842189759587,
+      "loss": 1.5527,
+      "step": 3691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015682471441142215,
+      "loss": 1.6133,
+      "step": 3692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001568010022073892,
+      "loss": 1.6445,
+      "step": 3693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015677728528746568,
+      "loss": 1.6133,
+      "step": 3694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015675356365362073,
+      "loss": 1.6367,
+      "step": 3695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001567298373078238,
+      "loss": 1.6426,
+      "step": 3696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015670610625204478,
+      "loss": 1.6309,
+      "step": 3697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001566823704882539,
+      "loss": 1.7188,
+      "step": 3698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015665863001842189,
+      "loss": 1.5938,
+      "step": 3699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001566348848445197,
+      "loss": 1.5723,
+      "step": 3700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015661113496851886,
+      "loss": 1.5781,
+      "step": 3701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015658738039239113,
+      "loss": 1.5215,
+      "step": 3702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001565636211181088,
+      "loss": 1.6348,
+      "step": 3703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001565398571476444,
+      "loss": 1.7402,
+      "step": 3704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00156516088482971,
+      "loss": 1.6172,
+      "step": 3705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001564923151260619,
+      "loss": 1.6641,
+      "step": 3706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015646853707889097,
+      "loss": 1.6777,
+      "step": 3707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015644475434343232,
+      "loss": 1.6914,
+      "step": 3708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015642096692166055,
+      "loss": 1.7148,
+      "step": 3709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015639717481555054,
+      "loss": 1.5293,
+      "step": 3710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001563733780270777,
+      "loss": 1.666,
+      "step": 3711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015634957655821767,
+      "loss": 1.6074,
+      "step": 3712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015632577041094664,
+      "loss": 1.627,
+      "step": 3713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015630195958724108,
+      "loss": 1.498,
+      "step": 3714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015627814408907786,
+      "loss": 1.6211,
+      "step": 3715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001562543239184343,
+      "loss": 1.5508,
+      "step": 3716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00156230499077288,
+      "loss": 1.6387,
+      "step": 3717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015620666956761707,
+      "loss": 1.5996,
+      "step": 3718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015618283539139987,
+      "loss": 1.5078,
+      "step": 3719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001561589965506154,
+      "loss": 1.6348,
+      "step": 3720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015613515304724267,
+      "loss": 1.5156,
+      "step": 3721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001561113048832614,
+      "loss": 1.6621,
+      "step": 3722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015608745206065154,
+      "loss": 1.7617,
+      "step": 3723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015606359458139349,
+      "loss": 1.5605,
+      "step": 3724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00156039732447468,
+      "loss": 1.6602,
+      "step": 3725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001560158656608562,
+      "loss": 1.627,
+      "step": 3726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015599199422353966,
+      "loss": 1.584,
+      "step": 3727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015596811813750027,
+      "loss": 1.6426,
+      "step": 3728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015594423740472034,
+      "loss": 1.5703,
+      "step": 3729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015592035202718257,
+      "loss": 1.7109,
+      "step": 3730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015589646200687007,
+      "loss": 1.6582,
+      "step": 3731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015587256734576624,
+      "loss": 1.498,
+      "step": 3732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015584866804585497,
+      "loss": 1.5938,
+      "step": 3733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015582476410912049,
+      "loss": 1.6172,
+      "step": 3734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015580085553754742,
+      "loss": 1.6973,
+      "step": 3735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015577694233312074,
+      "loss": 1.4746,
+      "step": 3736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015575302449782587,
+      "loss": 1.6191,
+      "step": 3737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015572910203364857,
+      "loss": 1.5859,
+      "step": 3738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015570517494257498,
+      "loss": 1.6523,
+      "step": 3739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015568124322659167,
+      "loss": 1.5215,
+      "step": 3740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015565730688768555,
+      "loss": 1.5996,
+      "step": 3741
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015563336592784394,
+      "loss": 1.6777,
+      "step": 3742
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015560942034905452,
+      "loss": 1.6172,
+      "step": 3743
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001555854701533054,
+      "loss": 1.6035,
+      "step": 3744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015556151534258495,
+      "loss": 1.5801,
+      "step": 3745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015553755591888212,
+      "loss": 1.5742,
+      "step": 3746
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015551359188418607,
+      "loss": 1.6621,
+      "step": 3747
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015548962324048645,
+      "loss": 1.6348,
+      "step": 3748
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015546564998977321,
+      "loss": 1.7227,
+      "step": 3749
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015544167213403677,
+      "loss": 1.6367,
+      "step": 3750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015541768967526782,
+      "loss": 1.5898,
+      "step": 3751
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015539370261545757,
+      "loss": 1.748,
+      "step": 3752
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015536971095659752,
+      "loss": 1.6133,
+      "step": 3753
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001553457147006795,
+      "loss": 1.6934,
+      "step": 3754
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015532171384969589,
+      "loss": 1.6133,
+      "step": 3755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015529770840563934,
+      "loss": 1.5859,
+      "step": 3756
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015527369837050285,
+      "loss": 1.7148,
+      "step": 3757
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001552496837462799,
+      "loss": 1.6758,
+      "step": 3758
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015522566453496426,
+      "loss": 1.6777,
+      "step": 3759
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015520164073855012,
+      "loss": 1.4297,
+      "step": 3760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015517761235903206,
+      "loss": 1.6504,
+      "step": 3761
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015515357939840508,
+      "loss": 1.6582,
+      "step": 3762
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015512954185866441,
+      "loss": 1.498,
+      "step": 3763
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015510549974180586,
+      "loss": 1.7148,
+      "step": 3764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0015508145304982545,
+      "loss": 1.7305,
+      "step": 3765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015505740178471967,
+      "loss": 1.6934,
+      "step": 3766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001550333459484854,
+      "loss": 1.5117,
+      "step": 3767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015500928554311986,
+      "loss": 1.7031,
+      "step": 3768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015498522057062064,
+      "loss": 1.5703,
+      "step": 3769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015496115103298571,
+      "loss": 1.6602,
+      "step": 3770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001549370769322135,
+      "loss": 1.6211,
+      "step": 3771
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015491299827030272,
+      "loss": 1.6875,
+      "step": 3772
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001548889150492525,
+      "loss": 1.6191,
+      "step": 3773
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015486482727106235,
+      "loss": 1.75,
+      "step": 3774
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015484073493773213,
+      "loss": 1.5,
+      "step": 3775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001548166380512621,
+      "loss": 1.6367,
+      "step": 3776
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015479253661365296,
+      "loss": 1.666,
+      "step": 3777
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015476843062690567,
+      "loss": 1.5801,
+      "step": 3778
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001547443200930216,
+      "loss": 1.6445,
+      "step": 3779
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001547202050140026,
+      "loss": 1.6602,
+      "step": 3780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015469608539185075,
+      "loss": 1.7207,
+      "step": 3781
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015467196122856861,
+      "loss": 1.5801,
+      "step": 3782
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001546478325261591,
+      "loss": 1.5918,
+      "step": 3783
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015462369928662546,
+      "loss": 1.5293,
+      "step": 3784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015459956151197135,
+      "loss": 1.6113,
+      "step": 3785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015457541920420085,
+      "loss": 1.5254,
+      "step": 3786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001545512723653183,
+      "loss": 1.7598,
+      "step": 3787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015452712099732855,
+      "loss": 1.6152,
+      "step": 3788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015450296510223672,
+      "loss": 1.6797,
+      "step": 3789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015447880468204835,
+      "loss": 1.5137,
+      "step": 3790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015445463973876937,
+      "loss": 1.666,
+      "step": 3791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015443047027440607,
+      "loss": 1.6055,
+      "step": 3792
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015440629629096512,
+      "loss": 1.5938,
+      "step": 3793
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015438211779045354,
+      "loss": 1.5625,
+      "step": 3794
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015435793477487874,
+      "loss": 1.6504,
+      "step": 3795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015433374724624857,
+      "loss": 1.6094,
+      "step": 3796
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015430955520657108,
+      "loss": 1.5508,
+      "step": 3797
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015428535865785493,
+      "loss": 1.627,
+      "step": 3798
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015426115760210895,
+      "loss": 1.5898,
+      "step": 3799
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015423695204134246,
+      "loss": 1.6973,
+      "step": 3800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001542127419775651,
+      "loss": 1.6387,
+      "step": 3801
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015418852741278695,
+      "loss": 1.5684,
+      "step": 3802
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001541643083490184,
+      "loss": 1.6094,
+      "step": 3803
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001541400847882702,
+      "loss": 1.6055,
+      "step": 3804
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015411585673255352,
+      "loss": 1.7344,
+      "step": 3805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015409162418387992,
+      "loss": 1.6797,
+      "step": 3806
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015406738714426127,
+      "loss": 1.6914,
+      "step": 3807
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015404314561570987,
+      "loss": 1.6836,
+      "step": 3808
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015401889960023833,
+      "loss": 1.543,
+      "step": 3809
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001539946490998597,
+      "loss": 1.6602,
+      "step": 3810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015397039411658739,
+      "loss": 1.5898,
+      "step": 3811
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001539461346524351,
+      "loss": 1.6602,
+      "step": 3812
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015392187070941703,
+      "loss": 1.6641,
+      "step": 3813
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015389760228954766,
+      "loss": 1.5469,
+      "step": 3814
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015387332939484186,
+      "loss": 1.6484,
+      "step": 3815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015384905202731494,
+      "loss": 1.6875,
+      "step": 3816
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015382477018898242,
+      "loss": 1.6777,
+      "step": 3817
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015380048388186041,
+      "loss": 1.6172,
+      "step": 3818
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015377619310796519,
+      "loss": 1.5469,
+      "step": 3819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001537518978693135,
+      "loss": 1.541,
+      "step": 3820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015372759816792252,
+      "loss": 1.6328,
+      "step": 3821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015370329400580967,
+      "loss": 1.6465,
+      "step": 3822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015367898538499278,
+      "loss": 1.5449,
+      "step": 3823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015365467230749011,
+      "loss": 1.5352,
+      "step": 3824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015363035477532023,
+      "loss": 1.6465,
+      "step": 3825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001536060327905021,
+      "loss": 1.5605,
+      "step": 3826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015358170635505503,
+      "loss": 1.5977,
+      "step": 3827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015355737547099878,
+      "loss": 1.6289,
+      "step": 3828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015353304014035331,
+      "loss": 1.5645,
+      "step": 3829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015350870036513916,
+      "loss": 1.6719,
+      "step": 3830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015348435614737708,
+      "loss": 1.6934,
+      "step": 3831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015346000748908827,
+      "loss": 1.6152,
+      "step": 3832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001534356543922942,
+      "loss": 1.6875,
+      "step": 3833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015341129685901688,
+      "loss": 1.6543,
+      "step": 3834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015338693489127856,
+      "loss": 1.4922,
+      "step": 3835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015336256849110184,
+      "loss": 1.6133,
+      "step": 3836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015333819766050977,
+      "loss": 1.5645,
+      "step": 3837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015331382240152576,
+      "loss": 1.6641,
+      "step": 3838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015328944271617352,
+      "loss": 1.5723,
+      "step": 3839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015326505860647714,
+      "loss": 1.5508,
+      "step": 3840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015324067007446117,
+      "loss": 1.5879,
+      "step": 3841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001532162771221504,
+      "loss": 1.5957,
+      "step": 3842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015319187975157012,
+      "loss": 1.5605,
+      "step": 3843
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015316747796474588,
+      "loss": 1.6348,
+      "step": 3844
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015314307176370362,
+      "loss": 1.707,
+      "step": 3845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001531186611504697,
+      "loss": 1.6172,
+      "step": 3846
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015309424612707074,
+      "loss": 1.668,
+      "step": 3847
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015306982669553387,
+      "loss": 1.5898,
+      "step": 3848
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015304540285788644,
+      "loss": 1.6172,
+      "step": 3849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001530209746161563,
+      "loss": 1.6504,
+      "step": 3850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001529965419723715,
+      "loss": 1.7656,
+      "step": 3851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015297210492856066,
+      "loss": 1.7324,
+      "step": 3852
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015294766348675265,
+      "loss": 1.6387,
+      "step": 3853
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001529232176489766,
+      "loss": 1.5605,
+      "step": 3854
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015289876741726226,
+      "loss": 1.7539,
+      "step": 3855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015287431279363951,
+      "loss": 1.7305,
+      "step": 3856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015284985378013877,
+      "loss": 1.5469,
+      "step": 3857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015282539037879067,
+      "loss": 1.5957,
+      "step": 3858
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015280092259162634,
+      "loss": 1.623,
+      "step": 3859
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015277645042067714,
+      "loss": 1.6172,
+      "step": 3860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015275197386797495,
+      "loss": 1.6426,
+      "step": 3861
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015272749293555187,
+      "loss": 1.5547,
+      "step": 3862
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015270300762544045,
+      "loss": 1.6387,
+      "step": 3863
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015267851793967354,
+      "loss": 1.5762,
+      "step": 3864
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015265402388028445,
+      "loss": 1.6133,
+      "step": 3865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015262952544930675,
+      "loss": 1.7051,
+      "step": 3866
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015260502264877445,
+      "loss": 1.5117,
+      "step": 3867
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015258051548072186,
+      "loss": 1.6309,
+      "step": 3868
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001525560039471837,
+      "loss": 1.6562,
+      "step": 3869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015253148805019498,
+      "loss": 1.582,
+      "step": 3870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015250696779179123,
+      "loss": 1.7969,
+      "step": 3871
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015248244317400818,
+      "loss": 1.6914,
+      "step": 3872
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015245791419888197,
+      "loss": 1.6074,
+      "step": 3873
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015243338086844914,
+      "loss": 1.6309,
+      "step": 3874
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015240884318474652,
+      "loss": 1.6504,
+      "step": 3875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001523843011498114,
+      "loss": 1.6309,
+      "step": 3876
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015235975476568138,
+      "loss": 1.5938,
+      "step": 3877
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0015233520403439439,
+      "loss": 1.7012,
+      "step": 3878
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015231064895798872,
+      "loss": 1.7988,
+      "step": 3879
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015228608953850307,
+      "loss": 1.748,
+      "step": 3880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015226152577797651,
+      "loss": 1.5898,
+      "step": 3881
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015223695767844845,
+      "loss": 1.6973,
+      "step": 3882
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015221238524195859,
+      "loss": 1.5215,
+      "step": 3883
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015218780847054707,
+      "loss": 1.7129,
+      "step": 3884
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015216322736625442,
+      "loss": 1.6504,
+      "step": 3885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015213864193112144,
+      "loss": 1.6699,
+      "step": 3886
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015211405216718932,
+      "loss": 1.502,
+      "step": 3887
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015208945807649965,
+      "loss": 1.7207,
+      "step": 3888
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015206485966109433,
+      "loss": 1.5371,
+      "step": 3889
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015204025692301564,
+      "loss": 1.6055,
+      "step": 3890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015201564986430621,
+      "loss": 1.5586,
+      "step": 3891
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015199103848700908,
+      "loss": 1.5879,
+      "step": 3892
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015196642279316755,
+      "loss": 1.5293,
+      "step": 3893
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015194180278482535,
+      "loss": 1.5762,
+      "step": 3894
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015191717846402655,
+      "loss": 1.6562,
+      "step": 3895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015189254983281558,
+      "loss": 1.7832,
+      "step": 3896
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015186791689323722,
+      "loss": 1.6191,
+      "step": 3897
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015184327964733665,
+      "loss": 1.7285,
+      "step": 3898
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015181863809715932,
+      "loss": 1.5176,
+      "step": 3899
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015179399224475114,
+      "loss": 1.6699,
+      "step": 3900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015176934209215826,
+      "loss": 1.6094,
+      "step": 3901
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015174468764142732,
+      "loss": 1.6621,
+      "step": 3902
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015172002889460522,
+      "loss": 1.6465,
+      "step": 3903
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015169536585373927,
+      "loss": 1.7168,
+      "step": 3904
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015167069852087703,
+      "loss": 1.5645,
+      "step": 3905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015164602689806662,
+      "loss": 1.7422,
+      "step": 3906
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015162135098735633,
+      "loss": 1.5566,
+      "step": 3907
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001515966707907949,
+      "loss": 1.5566,
+      "step": 3908
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015157198631043133,
+      "loss": 1.7324,
+      "step": 3909
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015154729754831512,
+      "loss": 1.6543,
+      "step": 3910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00151522604506496,
+      "loss": 1.6172,
+      "step": 3911
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015149790718702416,
+      "loss": 1.6543,
+      "step": 3912
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015147320559195003,
+      "loss": 1.7402,
+      "step": 3913
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015144849972332452,
+      "loss": 1.6445,
+      "step": 3914
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015142378958319875,
+      "loss": 1.6406,
+      "step": 3915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015139907517362431,
+      "loss": 1.7656,
+      "step": 3916
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015137435649665314,
+      "loss": 1.6719,
+      "step": 3917
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015134963355433748,
+      "loss": 1.6094,
+      "step": 3918
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015132490634872991,
+      "loss": 1.5215,
+      "step": 3919
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015130017488188346,
+      "loss": 1.7871,
+      "step": 3920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015127543915585137,
+      "loss": 1.6562,
+      "step": 3921
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015125069917268743,
+      "loss": 1.6562,
+      "step": 3922
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001512259549344456,
+      "loss": 1.5859,
+      "step": 3923
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015120120644318026,
+      "loss": 1.6387,
+      "step": 3924
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001511764537009462,
+      "loss": 1.6719,
+      "step": 3925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015115169670979847,
+      "loss": 1.5938,
+      "step": 3926
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015112693547179252,
+      "loss": 1.5898,
+      "step": 3927
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015110216998898416,
+      "loss": 1.6914,
+      "step": 3928
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001510774002634295,
+      "loss": 1.6035,
+      "step": 3929
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001510526262971851,
+      "loss": 1.6719,
+      "step": 3930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015102784809230775,
+      "loss": 1.5488,
+      "step": 3931
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001510030656508547,
+      "loss": 1.6777,
+      "step": 3932
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015097827897488352,
+      "loss": 1.7891,
+      "step": 3933
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015095348806645208,
+      "loss": 1.5898,
+      "step": 3934
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015092869292761863,
+      "loss": 1.709,
+      "step": 3935
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015090389356044182,
+      "loss": 1.5977,
+      "step": 3936
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015087908996698062,
+      "loss": 1.6445,
+      "step": 3937
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015085428214929428,
+      "loss": 1.6152,
+      "step": 3938
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001508294701094425,
+      "loss": 1.6016,
+      "step": 3939
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015080465384948532,
+      "loss": 1.6523,
+      "step": 3940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015077983337148306,
+      "loss": 1.6328,
+      "step": 3941
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015075500867749647,
+      "loss": 1.6055,
+      "step": 3942
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015073017976958657,
+      "loss": 1.4766,
+      "step": 3943
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015070534664981482,
+      "loss": 1.6055,
+      "step": 3944
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015068050932024297,
+      "loss": 1.6387,
+      "step": 3945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015065566778293313,
+      "loss": 1.7266,
+      "step": 3946
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015063082203994776,
+      "loss": 1.7715,
+      "step": 3947
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001506059720933497,
+      "loss": 1.4805,
+      "step": 3948
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015058111794520204,
+      "loss": 1.582,
+      "step": 3949
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015055625959756837,
+      "loss": 1.6621,
+      "step": 3950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001505313970525125,
+      "loss": 1.5938,
+      "step": 3951
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001505065303120987,
+      "loss": 1.6484,
+      "step": 3952
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015048165937839141,
+      "loss": 1.7188,
+      "step": 3953
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015045678425345565,
+      "loss": 1.6152,
+      "step": 3954
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015043190493935661,
+      "loss": 1.5605,
+      "step": 3955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015040702143815992,
+      "loss": 1.5762,
+      "step": 3956
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001503821337519315,
+      "loss": 1.6152,
+      "step": 3957
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001503572418827377,
+      "loss": 1.6211,
+      "step": 3958
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015033234583264506,
+      "loss": 1.6504,
+      "step": 3959
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015030744560372066,
+      "loss": 1.6895,
+      "step": 3960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015028254119803174,
+      "loss": 1.5977,
+      "step": 3961
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001502576326176461,
+      "loss": 1.6035,
+      "step": 3962
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015023271986463172,
+      "loss": 1.6641,
+      "step": 3963
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015020780294105697,
+      "loss": 1.6367,
+      "step": 3964
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015018288184899054,
+      "loss": 1.707,
+      "step": 3965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015015795659050158,
+      "loss": 1.7168,
+      "step": 3966
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001501330271676594,
+      "loss": 1.5957,
+      "step": 3967
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015010809358253386,
+      "loss": 1.5625,
+      "step": 3968
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015008315583719498,
+      "loss": 1.5977,
+      "step": 3969
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015005821393371328,
+      "loss": 1.6504,
+      "step": 3970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001500332678741595,
+      "loss": 1.4844,
+      "step": 3971
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0015000831766060485,
+      "loss": 1.6973,
+      "step": 3972
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0014998336329512074,
+      "loss": 1.5801,
+      "step": 3973
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0014995840477977907,
+      "loss": 1.6289,
+      "step": 3974
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0014993344211665194,
+      "loss": 1.6094,
+      "step": 3975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0014990847530781193,
+      "loss": 1.5645,
+      "step": 3976
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001498835043553319,
+      "loss": 1.752,
+      "step": 3977
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00149858529261285,
+      "loss": 1.7031,
+      "step": 3978
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0014983355002774485,
+      "loss": 1.6387,
+      "step": 3979
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0014980856665678531,
+      "loss": 1.6504,
+      "step": 3980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0014978357915048065,
+      "loss": 1.5918,
+      "step": 3981
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0014975858751090539,
+      "loss": 1.6387,
+      "step": 3982
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0014973359174013452,
+      "loss": 1.7598,
+      "step": 3983
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0014970859184024327,
+      "loss": 1.5234,
+      "step": 3984
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0014968358781330725,
+      "loss": 1.6855,
+      "step": 3985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0014965857966140245,
+      "loss": 1.5039,
+      "step": 3986
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0014963356738660514,
+      "loss": 1.5566,
+      "step": 3987
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0014960855099099195,
+      "loss": 1.7285,
+      "step": 3988
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0014958353047663988,
+      "loss": 1.5488,
+      "step": 3989
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0014955850584562622,
+      "loss": 1.6504,
+      "step": 3990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014953347710002871,
+      "loss": 1.5957,
+      "step": 3991
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014950844424192527,
+      "loss": 1.6406,
+      "step": 3992
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014948340727339432,
+      "loss": 1.6426,
+      "step": 3993
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014945836619651447,
+      "loss": 1.582,
+      "step": 3994
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001494333210133648,
+      "loss": 1.5352,
+      "step": 3995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001494082717260247,
+      "loss": 1.6484,
+      "step": 3996
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014938321833657385,
+      "loss": 1.6992,
+      "step": 3997
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014935816084709226,
+      "loss": 1.7051,
+      "step": 3998
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001493330992596604,
+      "loss": 1.6758,
+      "step": 3999
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014930803357635898,
+      "loss": 1.5762,
+      "step": 4000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014928296379926906,
+      "loss": 1.6934,
+      "step": 4001
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014925788993047204,
+      "loss": 1.6387,
+      "step": 4002
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001492328119720497,
+      "loss": 1.6562,
+      "step": 4003
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014920772992608412,
+      "loss": 1.4922,
+      "step": 4004
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014918264379465774,
+      "loss": 1.5703,
+      "step": 4005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001491575535798533,
+      "loss": 1.6445,
+      "step": 4006
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014913245928375397,
+      "loss": 1.5801,
+      "step": 4007
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001491073609084431,
+      "loss": 1.5605,
+      "step": 4008
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014908225845600461,
+      "loss": 1.6973,
+      "step": 4009
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014905715192852248,
+      "loss": 1.5586,
+      "step": 4010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001490320413280813,
+      "loss": 1.6328,
+      "step": 4011
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001490069266567658,
+      "loss": 1.5605,
+      "step": 4012
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014898180791666116,
+      "loss": 1.6719,
+      "step": 4013
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014895668510985281,
+      "loss": 1.7227,
+      "step": 4014
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014893155823842662,
+      "loss": 1.5957,
+      "step": 4015
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001489064273044687,
+      "loss": 1.6855,
+      "step": 4016
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014888129231006555,
+      "loss": 1.6953,
+      "step": 4017
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00148856153257304,
+      "loss": 1.627,
+      "step": 4018
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014883101014827123,
+      "loss": 1.7422,
+      "step": 4019
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001488058629850547,
+      "loss": 1.8203,
+      "step": 4020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001487807117697423,
+      "loss": 1.5879,
+      "step": 4021
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014875555650442217,
+      "loss": 1.6289,
+      "step": 4022
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014873039719118283,
+      "loss": 1.8008,
+      "step": 4023
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001487052338321131,
+      "loss": 1.6016,
+      "step": 4024
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014868006642930224,
+      "loss": 1.5879,
+      "step": 4025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014865489498483968,
+      "loss": 1.7227,
+      "step": 4026
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014862971950081534,
+      "loss": 1.7148,
+      "step": 4027
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014860453997931936,
+      "loss": 1.5781,
+      "step": 4028
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014857935642244228,
+      "loss": 1.709,
+      "step": 4029
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014855416883227497,
+      "loss": 1.7578,
+      "step": 4030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014852897721090862,
+      "loss": 1.6113,
+      "step": 4031
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014850378156043473,
+      "loss": 1.6445,
+      "step": 4032
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014847858188294523,
+      "loss": 1.5762,
+      "step": 4033
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014845337818053224,
+      "loss": 1.6348,
+      "step": 4034
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014842817045528833,
+      "loss": 1.6562,
+      "step": 4035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001484029587093064,
+      "loss": 1.5781,
+      "step": 4036
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014837774294467958,
+      "loss": 1.6211,
+      "step": 4037
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014835252316350144,
+      "loss": 1.6914,
+      "step": 4038
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014832729936786586,
+      "loss": 1.6641,
+      "step": 4039
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00148302071559867,
+      "loss": 1.584,
+      "step": 4040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014827683974159944,
+      "loss": 1.5859,
+      "step": 4041
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00148251603915158,
+      "loss": 1.6484,
+      "step": 4042
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014822636408263793,
+      "loss": 1.6133,
+      "step": 4043
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014820112024613468,
+      "loss": 1.6602,
+      "step": 4044
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014817587240774422,
+      "loss": 1.6211,
+      "step": 4045
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014815062056956264,
+      "loss": 1.6191,
+      "step": 4046
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014812536473368656,
+      "loss": 1.5664,
+      "step": 4047
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001481001049022128,
+      "loss": 1.6387,
+      "step": 4048
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014807484107723853,
+      "loss": 1.7285,
+      "step": 4049
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014804957326086132,
+      "loss": 1.7812,
+      "step": 4050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00148024301455179,
+      "loss": 1.5547,
+      "step": 4051
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014799902566228972,
+      "loss": 1.5176,
+      "step": 4052
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001479737458842921,
+      "loss": 1.5957,
+      "step": 4053
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014794846212328485,
+      "loss": 1.5469,
+      "step": 4054
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014792317438136729,
+      "loss": 1.5566,
+      "step": 4055
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001478978826606388,
+      "loss": 1.6328,
+      "step": 4056
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014787258696319931,
+      "loss": 1.6582,
+      "step": 4057
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014784728729114898,
+      "loss": 1.6133,
+      "step": 4058
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014782198364658825,
+      "loss": 1.5371,
+      "step": 4059
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00147796676031618,
+      "loss": 1.6445,
+      "step": 4060
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014777136444833938,
+      "loss": 1.6562,
+      "step": 4061
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014774604889885388,
+      "loss": 1.5664,
+      "step": 4062
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001477207293852633,
+      "loss": 1.6699,
+      "step": 4063
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014769540590966982,
+      "loss": 1.6074,
+      "step": 4064
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014767007847417586,
+      "loss": 1.6016,
+      "step": 4065
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001476447470808843,
+      "loss": 1.6582,
+      "step": 4066
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014761941173189821,
+      "loss": 1.5332,
+      "step": 4067
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001475940724293211,
+      "loss": 1.6504,
+      "step": 4068
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014756872917525672,
+      "loss": 1.5938,
+      "step": 4069
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014754338197180922,
+      "loss": 1.7148,
+      "step": 4070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014751803082108299,
+      "loss": 1.7227,
+      "step": 4071
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014749267572518288,
+      "loss": 1.752,
+      "step": 4072
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014746731668621395,
+      "loss": 1.6562,
+      "step": 4073
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014744195370628161,
+      "loss": 1.5566,
+      "step": 4074
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014741658678749168,
+      "loss": 1.8262,
+      "step": 4075
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001473912159319502,
+      "loss": 1.5918,
+      "step": 4076
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014736584114176355,
+      "loss": 1.6406,
+      "step": 4077
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014734046241903851,
+      "loss": 1.6992,
+      "step": 4078
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014731507976588213,
+      "loss": 1.625,
+      "step": 4079
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001472896931844018,
+      "loss": 1.5859,
+      "step": 4080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014726430267670524,
+      "loss": 1.6172,
+      "step": 4081
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014723890824490048,
+      "loss": 1.6699,
+      "step": 4082
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014721350989109587,
+      "loss": 1.6426,
+      "step": 4083
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014718810761740018,
+      "loss": 1.7422,
+      "step": 4084
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014716270142592235,
+      "loss": 1.6348,
+      "step": 4085
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014713729131877174,
+      "loss": 1.6289,
+      "step": 4086
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014711187729805801,
+      "loss": 1.6641,
+      "step": 4087
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001470864593658912,
+      "loss": 1.6699,
+      "step": 4088
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014706103752438159,
+      "loss": 1.584,
+      "step": 4089
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014703561177563983,
+      "loss": 1.6777,
+      "step": 4090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014701018212177686,
+      "loss": 1.5996,
+      "step": 4091
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014698474856490403,
+      "loss": 1.5566,
+      "step": 4092
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014695931110713292,
+      "loss": 1.5156,
+      "step": 4093
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014693386975057545,
+      "loss": 1.5938,
+      "step": 4094
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014690842449734394,
+      "loss": 1.5762,
+      "step": 4095
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014688297534955095,
+      "loss": 1.748,
+      "step": 4096
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014685752230930937,
+      "loss": 1.5898,
+      "step": 4097
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014683206537873241,
+      "loss": 1.7012,
+      "step": 4098
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014680660455993369,
+      "loss": 1.5879,
+      "step": 4099
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014678113985502708,
+      "loss": 1.4707,
+      "step": 4100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014675567126612673,
+      "loss": 1.6191,
+      "step": 4101
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0014673019879534721,
+      "loss": 1.5645,
+      "step": 4102
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014670472244480335,
+      "loss": 1.6055,
+      "step": 4103
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001466792422166103,
+      "loss": 1.5625,
+      "step": 4104
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001466537581128836,
+      "loss": 1.5586,
+      "step": 4105
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014662827013573903,
+      "loss": 1.6172,
+      "step": 4106
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001466027782872927,
+      "loss": 1.6777,
+      "step": 4107
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014657728256966113,
+      "loss": 1.5977,
+      "step": 4108
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014655178298496102,
+      "loss": 1.5938,
+      "step": 4109
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014652627953530952,
+      "loss": 1.5859,
+      "step": 4110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014650077222282403,
+      "loss": 1.5664,
+      "step": 4111
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014647526104962232,
+      "loss": 1.5371,
+      "step": 4112
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014644974601782238,
+      "loss": 1.6094,
+      "step": 4113
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014642422712954266,
+      "loss": 1.7246,
+      "step": 4114
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014639870438690184,
+      "loss": 1.7793,
+      "step": 4115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001463731777920189,
+      "loss": 1.7734,
+      "step": 4116
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014634764734701323,
+      "loss": 1.5508,
+      "step": 4117
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014632211305400447,
+      "loss": 1.709,
+      "step": 4118
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001462965749151126,
+      "loss": 1.6816,
+      "step": 4119
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014627103293245792,
+      "loss": 1.5547,
+      "step": 4120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00146245487108161,
+      "loss": 1.5293,
+      "step": 4121
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014621993744434287,
+      "loss": 1.6504,
+      "step": 4122
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014619438394312474,
+      "loss": 1.6719,
+      "step": 4123
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014616882660662817,
+      "loss": 1.5898,
+      "step": 4124
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014614326543697506,
+      "loss": 1.666,
+      "step": 4125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014611770043628763,
+      "loss": 1.5977,
+      "step": 4126
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014609213160668838,
+      "loss": 1.6602,
+      "step": 4127
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001460665589503002,
+      "loss": 1.6953,
+      "step": 4128
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014604098246924623,
+      "loss": 1.6211,
+      "step": 4129
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014601540216564995,
+      "loss": 1.666,
+      "step": 4130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014598981804163514,
+      "loss": 1.627,
+      "step": 4131
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014596423009932598,
+      "loss": 1.498,
+      "step": 4132
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014593863834084682,
+      "loss": 1.5566,
+      "step": 4133
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014591304276832249,
+      "loss": 1.6543,
+      "step": 4134
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014588744338387802,
+      "loss": 1.7207,
+      "step": 4135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014586184018963878,
+      "loss": 1.6191,
+      "step": 4136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014583623318773047,
+      "loss": 1.6719,
+      "step": 4137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014581062238027914,
+      "loss": 1.7266,
+      "step": 4138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014578500776941106,
+      "loss": 1.582,
+      "step": 4139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014575938935725294,
+      "loss": 1.5352,
+      "step": 4140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014573376714593174,
+      "loss": 1.5391,
+      "step": 4141
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014570814113757467,
+      "loss": 1.6895,
+      "step": 4142
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014568251133430942,
+      "loss": 1.5918,
+      "step": 4143
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014565687773826382,
+      "loss": 1.4941,
+      "step": 4144
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014563124035156614,
+      "loss": 1.6074,
+      "step": 4145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014560559917634489,
+      "loss": 1.6777,
+      "step": 4146
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014557995421472893,
+      "loss": 1.6562,
+      "step": 4147
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014555430546884743,
+      "loss": 1.7402,
+      "step": 4148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014552865294082992,
+      "loss": 1.5312,
+      "step": 4149
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014550299663280608,
+      "loss": 1.6133,
+      "step": 4150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014547733654690614,
+      "loss": 1.7637,
+      "step": 4151
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014545167268526042,
+      "loss": 1.6172,
+      "step": 4152
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014542600504999976,
+      "loss": 1.5508,
+      "step": 4153
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014540033364325512,
+      "loss": 1.5293,
+      "step": 4154
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014537465846715793,
+      "loss": 1.6855,
+      "step": 4155
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014534897952383982,
+      "loss": 1.6543,
+      "step": 4156
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014532329681543282,
+      "loss": 1.6641,
+      "step": 4157
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014529761034406917,
+      "loss": 1.625,
+      "step": 4158
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014527192011188153,
+      "loss": 1.6523,
+      "step": 4159
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001452462261210028,
+      "loss": 1.6895,
+      "step": 4160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014522052837356626,
+      "loss": 1.6191,
+      "step": 4161
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014519482687170542,
+      "loss": 1.6699,
+      "step": 4162
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014516912161755416,
+      "loss": 1.6328,
+      "step": 4163
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014514341261324662,
+      "loss": 1.5645,
+      "step": 4164
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014511769986091734,
+      "loss": 1.5625,
+      "step": 4165
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014509198336270107,
+      "loss": 1.5391,
+      "step": 4166
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014506626312073294,
+      "loss": 1.6016,
+      "step": 4167
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014504053913714837,
+      "loss": 1.5938,
+      "step": 4168
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014501481141408305,
+      "loss": 1.6543,
+      "step": 4169
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014498907995367306,
+      "loss": 1.6543,
+      "step": 4170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014496334475805476,
+      "loss": 1.6543,
+      "step": 4171
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014493760582936475,
+      "loss": 1.6289,
+      "step": 4172
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014491186316974004,
+      "loss": 1.5625,
+      "step": 4173
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014488611678131794,
+      "loss": 1.6328,
+      "step": 4174
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014486036666623598,
+      "loss": 1.6055,
+      "step": 4175
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014483461282663206,
+      "loss": 1.5527,
+      "step": 4176
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014480885526464442,
+      "loss": 1.6699,
+      "step": 4177
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014478309398241154,
+      "loss": 1.584,
+      "step": 4178
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001447573289820723,
+      "loss": 1.6914,
+      "step": 4179
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014473156026576577,
+      "loss": 1.7305,
+      "step": 4180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014470578783563144,
+      "loss": 1.5684,
+      "step": 4181
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00144680011693809,
+      "loss": 1.7617,
+      "step": 4182
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001446542318424386,
+      "loss": 1.6699,
+      "step": 4183
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014462844828366053,
+      "loss": 1.623,
+      "step": 4184
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014460266101961549,
+      "loss": 1.5996,
+      "step": 4185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014457687005244445,
+      "loss": 1.6562,
+      "step": 4186
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014455107538428873,
+      "loss": 1.5605,
+      "step": 4187
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001445252770172899,
+      "loss": 1.6875,
+      "step": 4188
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014449947495358989,
+      "loss": 1.6465,
+      "step": 4189
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014447366919533087,
+      "loss": 1.5879,
+      "step": 4190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014444785974465537,
+      "loss": 1.6406,
+      "step": 4191
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014442204660370624,
+      "loss": 1.6367,
+      "step": 4192
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014439622977462658,
+      "loss": 1.4785,
+      "step": 4193
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014437040925955986,
+      "loss": 1.5723,
+      "step": 4194
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001443445850606498,
+      "loss": 1.627,
+      "step": 4195
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014431875718004046,
+      "loss": 1.6504,
+      "step": 4196
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014429292561987618,
+      "loss": 1.6172,
+      "step": 4197
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014426709038230158,
+      "loss": 1.668,
+      "step": 4198
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014424125146946171,
+      "loss": 1.5312,
+      "step": 4199
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001442154088835018,
+      "loss": 1.5566,
+      "step": 4200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014418956262656742,
+      "loss": 1.6055,
+      "step": 4201
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014416371270080442,
+      "loss": 1.666,
+      "step": 4202
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014413785910835906,
+      "loss": 1.6191,
+      "step": 4203
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014411200185137777,
+      "loss": 1.6191,
+      "step": 4204
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014408614093200736,
+      "loss": 1.582,
+      "step": 4205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014406027635239492,
+      "loss": 1.6758,
+      "step": 4206
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014403440811468786,
+      "loss": 1.6191,
+      "step": 4207
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001440085362210339,
+      "loss": 1.543,
+      "step": 4208
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014398266067358102,
+      "loss": 1.7305,
+      "step": 4209
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014395678147447752,
+      "loss": 1.4863,
+      "step": 4210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014393089862587208,
+      "loss": 1.6348,
+      "step": 4211
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014390501212991353,
+      "loss": 1.4766,
+      "step": 4212
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014387912198875117,
+      "loss": 1.5391,
+      "step": 4213
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014385322820453447,
+      "loss": 1.5059,
+      "step": 4214
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0014382733077941327,
+      "loss": 1.6758,
+      "step": 4215
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014380142971553774,
+      "loss": 1.6719,
+      "step": 4216
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014377552501505825,
+      "loss": 1.6719,
+      "step": 4217
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014374961668012554,
+      "loss": 1.5762,
+      "step": 4218
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014372370471289067,
+      "loss": 1.6113,
+      "step": 4219
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014369778911550496,
+      "loss": 1.7363,
+      "step": 4220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014367186989012004,
+      "loss": 1.6855,
+      "step": 4221
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014364594703888786,
+      "loss": 1.4707,
+      "step": 4222
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014362002056396066,
+      "loss": 1.5566,
+      "step": 4223
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014359409046749098,
+      "loss": 1.707,
+      "step": 4224
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014356815675163166,
+      "loss": 1.7285,
+      "step": 4225
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014354221941853583,
+      "loss": 1.6875,
+      "step": 4226
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014351627847035693,
+      "loss": 1.709,
+      "step": 4227
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014349033390924873,
+      "loss": 1.6172,
+      "step": 4228
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014346438573736523,
+      "loss": 1.6055,
+      "step": 4229
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001434384339568608,
+      "loss": 1.7168,
+      "step": 4230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014341247856989009,
+      "loss": 1.6387,
+      "step": 4231
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014338651957860802,
+      "loss": 1.5801,
+      "step": 4232
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014336055698516983,
+      "loss": 1.7383,
+      "step": 4233
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014333459079173107,
+      "loss": 1.625,
+      "step": 4234
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014330862100044755,
+      "loss": 1.6016,
+      "step": 4235
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014328264761347546,
+      "loss": 1.5781,
+      "step": 4236
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014325667063297117,
+      "loss": 1.5957,
+      "step": 4237
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014323069006109143,
+      "loss": 1.5527,
+      "step": 4238
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001432047058999933,
+      "loss": 1.6191,
+      "step": 4239
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014317871815183413,
+      "loss": 1.668,
+      "step": 4240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014315272681877146,
+      "loss": 1.6113,
+      "step": 4241
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001431267319029633,
+      "loss": 1.6934,
+      "step": 4242
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014310073340656779,
+      "loss": 1.6367,
+      "step": 4243
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014307473133174356,
+      "loss": 1.6055,
+      "step": 4244
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001430487256806493,
+      "loss": 1.5566,
+      "step": 4245
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014302271645544424,
+      "loss": 1.5508,
+      "step": 4246
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001429967036582877,
+      "loss": 1.6309,
+      "step": 4247
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014297068729133943,
+      "loss": 1.6641,
+      "step": 4248
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014294466735675945,
+      "loss": 1.502,
+      "step": 4249
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014291864385670803,
+      "loss": 1.627,
+      "step": 4250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014289261679334576,
+      "loss": 1.6074,
+      "step": 4251
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014286658616883355,
+      "loss": 1.6797,
+      "step": 4252
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001428405519853326,
+      "loss": 1.6094,
+      "step": 4253
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014281451424500433,
+      "loss": 1.6602,
+      "step": 4254
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014278847295001056,
+      "loss": 1.5566,
+      "step": 4255
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014276242810251339,
+      "loss": 1.498,
+      "step": 4256
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014273637970467514,
+      "loss": 1.6973,
+      "step": 4257
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001427103277586585,
+      "loss": 1.6602,
+      "step": 4258
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014268427226662639,
+      "loss": 1.7559,
+      "step": 4259
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014265821323074213,
+      "loss": 1.6641,
+      "step": 4260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014263215065316922,
+      "loss": 1.7305,
+      "step": 4261
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014260608453607149,
+      "loss": 1.6543,
+      "step": 4262
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001425800148816131,
+      "loss": 1.6895,
+      "step": 4263
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014255394169195845,
+      "loss": 1.6191,
+      "step": 4264
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014252786496927229,
+      "loss": 1.7148,
+      "step": 4265
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014250178471571964,
+      "loss": 1.5352,
+      "step": 4266
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014247570093346576,
+      "loss": 1.6875,
+      "step": 4267
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014244961362467627,
+      "loss": 1.498,
+      "step": 4268
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001424235227915171,
+      "loss": 1.6738,
+      "step": 4269
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001423974284361544,
+      "loss": 1.5703,
+      "step": 4270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014237133056075465,
+      "loss": 1.6602,
+      "step": 4271
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014234522916748463,
+      "loss": 1.7891,
+      "step": 4272
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001423191242585114,
+      "loss": 1.6289,
+      "step": 4273
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001422930158360023,
+      "loss": 1.7598,
+      "step": 4274
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00142266903902125,
+      "loss": 1.6816,
+      "step": 4275
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014224078845904745,
+      "loss": 1.5391,
+      "step": 4276
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001422146695089378,
+      "loss": 1.6133,
+      "step": 4277
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014218854705396464,
+      "loss": 1.5938,
+      "step": 4278
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014216242109629678,
+      "loss": 1.5898,
+      "step": 4279
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014213629163810331,
+      "loss": 1.7148,
+      "step": 4280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001421101586815536,
+      "loss": 1.6289,
+      "step": 4281
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014208402222881738,
+      "loss": 1.7031,
+      "step": 4282
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014205788228206456,
+      "loss": 1.5273,
+      "step": 4283
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014203173884346548,
+      "loss": 1.4902,
+      "step": 4284
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001420055919151906,
+      "loss": 1.498,
+      "step": 4285
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014197944149941086,
+      "loss": 1.6562,
+      "step": 4286
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014195328759829733,
+      "loss": 1.7227,
+      "step": 4287
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014192713021402147,
+      "loss": 1.6172,
+      "step": 4288
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014190096934875494,
+      "loss": 1.6035,
+      "step": 4289
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014187480500466977,
+      "loss": 1.6504,
+      "step": 4290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014184863718393825,
+      "loss": 1.5723,
+      "step": 4291
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00141822465888733,
+      "loss": 1.584,
+      "step": 4292
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001417962911212268,
+      "loss": 1.6465,
+      "step": 4293
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014177011288359286,
+      "loss": 1.6895,
+      "step": 4294
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014174393117800462,
+      "loss": 1.6914,
+      "step": 4295
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001417177460066358,
+      "loss": 1.6094,
+      "step": 4296
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014169155737166039,
+      "loss": 1.5781,
+      "step": 4297
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014166536527525273,
+      "loss": 1.6348,
+      "step": 4298
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001416391697195874,
+      "loss": 1.7832,
+      "step": 4299
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014161297070683933,
+      "loss": 1.6191,
+      "step": 4300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014158676823918359,
+      "loss": 1.5762,
+      "step": 4301
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014156056231879573,
+      "loss": 1.6328,
+      "step": 4302
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001415343529478514,
+      "loss": 1.5625,
+      "step": 4303
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014150814012852673,
+      "loss": 1.6543,
+      "step": 4304
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014148192386299794,
+      "loss": 1.6172,
+      "step": 4305
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014145570415344168,
+      "loss": 1.6016,
+      "step": 4306
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001414294810020348,
+      "loss": 1.5703,
+      "step": 4307
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014140325441095456,
+      "loss": 1.8379,
+      "step": 4308
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001413770243823783,
+      "loss": 1.543,
+      "step": 4309
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014135079091848381,
+      "loss": 1.5684,
+      "step": 4310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014132455402144915,
+      "loss": 1.6133,
+      "step": 4311
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001412983136934526,
+      "loss": 1.5332,
+      "step": 4312
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014127206993667276,
+      "loss": 1.7441,
+      "step": 4313
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001412458227532885,
+      "loss": 1.5195,
+      "step": 4314
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014121957214547902,
+      "loss": 1.709,
+      "step": 4315
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014119331811542372,
+      "loss": 1.6289,
+      "step": 4316
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001411670606653024,
+      "loss": 1.6562,
+      "step": 4317
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00141140799797295,
+      "loss": 1.4746,
+      "step": 4318
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014111453551358193,
+      "loss": 1.7363,
+      "step": 4319
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014108826781634365,
+      "loss": 1.6484,
+      "step": 4320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014106199670776115,
+      "loss": 1.6582,
+      "step": 4321
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001410357221900155,
+      "loss": 1.6426,
+      "step": 4322
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014100944426528818,
+      "loss": 1.668,
+      "step": 4323
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014098316293576086,
+      "loss": 1.6797,
+      "step": 4324
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001409568782036156,
+      "loss": 1.6113,
+      "step": 4325
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014093059007103462,
+      "loss": 1.6328,
+      "step": 4326
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0014090429854020058,
+      "loss": 1.5879,
+      "step": 4327
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014087800361329622,
+      "loss": 1.5488,
+      "step": 4328
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014085170529250476,
+      "loss": 1.6465,
+      "step": 4329
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014082540358000956,
+      "loss": 1.5703,
+      "step": 4330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014079909847799432,
+      "loss": 1.627,
+      "step": 4331
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014077278998864305,
+      "loss": 1.6602,
+      "step": 4332
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014074647811413994,
+      "loss": 1.7129,
+      "step": 4333
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001407201628566696,
+      "loss": 1.5508,
+      "step": 4334
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014069384421841681,
+      "loss": 1.75,
+      "step": 4335
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014066752220156667,
+      "loss": 1.6426,
+      "step": 4336
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014064119680830458,
+      "loss": 1.6074,
+      "step": 4337
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014061486804081615,
+      "loss": 1.7461,
+      "step": 4338
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014058853590128738,
+      "loss": 1.5938,
+      "step": 4339
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014056220039190446,
+      "loss": 1.5137,
+      "step": 4340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014053586151485389,
+      "loss": 1.5176,
+      "step": 4341
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014050951927232246,
+      "loss": 1.6367,
+      "step": 4342
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014048317366649723,
+      "loss": 1.5586,
+      "step": 4343
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014045682469956552,
+      "loss": 1.6172,
+      "step": 4344
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014043047237371495,
+      "loss": 1.6895,
+      "step": 4345
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014040411669113343,
+      "loss": 1.7422,
+      "step": 4346
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014037775765400914,
+      "loss": 1.6113,
+      "step": 4347
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014035139526453052,
+      "loss": 1.6172,
+      "step": 4348
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001403250295248863,
+      "loss": 1.7344,
+      "step": 4349
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014029866043726547,
+      "loss": 1.5898,
+      "step": 4350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014027228800385738,
+      "loss": 1.5391,
+      "step": 4351
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014024591222685154,
+      "loss": 1.6465,
+      "step": 4352
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001402195331084378,
+      "loss": 1.7402,
+      "step": 4353
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014019315065080628,
+      "loss": 1.6309,
+      "step": 4354
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014016676485614744,
+      "loss": 1.6113,
+      "step": 4355
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014014037572665184,
+      "loss": 1.6602,
+      "step": 4356
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014011398326451052,
+      "loss": 1.5898,
+      "step": 4357
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014008758747191467,
+      "loss": 1.6465,
+      "step": 4358
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014006118835105583,
+      "loss": 1.6113,
+      "step": 4359
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014003478590412573,
+      "loss": 1.498,
+      "step": 4360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0014000838013331646,
+      "loss": 1.6582,
+      "step": 4361
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013998197104082035,
+      "loss": 1.6367,
+      "step": 4362
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013995555862883003,
+      "loss": 1.6133,
+      "step": 4363
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013992914289953833,
+      "loss": 1.748,
+      "step": 4364
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013990272385513848,
+      "loss": 1.6035,
+      "step": 4365
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013987630149782382,
+      "loss": 1.7285,
+      "step": 4366
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013984987582978817,
+      "loss": 1.6875,
+      "step": 4367
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013982344685322544,
+      "loss": 1.6191,
+      "step": 4368
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013979701457032992,
+      "loss": 1.7109,
+      "step": 4369
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013977057898329611,
+      "loss": 1.627,
+      "step": 4370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013974414009431885,
+      "loss": 1.6211,
+      "step": 4371
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013971769790559329,
+      "loss": 1.4844,
+      "step": 4372
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013969125241931463,
+      "loss": 1.6035,
+      "step": 4373
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013966480363767862,
+      "loss": 1.6094,
+      "step": 4374
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013963835156288113,
+      "loss": 1.6602,
+      "step": 4375
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013961189619711833,
+      "loss": 1.5957,
+      "step": 4376
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013958543754258668,
+      "loss": 1.5586,
+      "step": 4377
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001395589756014829,
+      "loss": 1.7227,
+      "step": 4378
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00139532510376004,
+      "loss": 1.7539,
+      "step": 4379
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013950604186834723,
+      "loss": 1.6797,
+      "step": 4380
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013947957008071015,
+      "loss": 1.5977,
+      "step": 4381
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013945309501529053,
+      "loss": 1.623,
+      "step": 4382
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013942661667428653,
+      "loss": 1.7578,
+      "step": 4383
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013940013505989645,
+      "loss": 1.6562,
+      "step": 4384
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013937365017431896,
+      "loss": 1.5566,
+      "step": 4385
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013934716201975293,
+      "loss": 1.5352,
+      "step": 4386
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013932067059839754,
+      "loss": 1.5879,
+      "step": 4387
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013929417591245223,
+      "loss": 1.6719,
+      "step": 4388
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013926767796411674,
+      "loss": 1.6172,
+      "step": 4389
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013924117675559102,
+      "loss": 1.6035,
+      "step": 4390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001392146722890754,
+      "loss": 1.7695,
+      "step": 4391
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001391881645667703,
+      "loss": 1.6992,
+      "step": 4392
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013916165359087657,
+      "loss": 1.5469,
+      "step": 4393
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013913513936359533,
+      "loss": 1.6504,
+      "step": 4394
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013910862188712782,
+      "loss": 1.6504,
+      "step": 4395
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013908210116367575,
+      "loss": 1.5312,
+      "step": 4396
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001390555771954409,
+      "loss": 1.5039,
+      "step": 4397
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013902904998462546,
+      "loss": 1.6758,
+      "step": 4398
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013900251953343186,
+      "loss": 1.4824,
+      "step": 4399
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013897598584406278,
+      "loss": 1.5918,
+      "step": 4400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013894944891872112,
+      "loss": 1.6328,
+      "step": 4401
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013892290875961019,
+      "loss": 1.5332,
+      "step": 4402
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001388963653689334,
+      "loss": 1.7188,
+      "step": 4403
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013886981874889457,
+      "loss": 1.6094,
+      "step": 4404
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001388432689016977,
+      "loss": 1.6426,
+      "step": 4405
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001388167158295471,
+      "loss": 1.6777,
+      "step": 4406
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013879015953464728,
+      "loss": 1.5898,
+      "step": 4407
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013876360001920315,
+      "loss": 1.5254,
+      "step": 4408
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013873703728541976,
+      "loss": 1.6133,
+      "step": 4409
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013871047133550247,
+      "loss": 1.7109,
+      "step": 4410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013868390217165697,
+      "loss": 1.6875,
+      "step": 4411
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013865732979608907,
+      "loss": 1.4395,
+      "step": 4412
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013863075421100502,
+      "loss": 1.6348,
+      "step": 4413
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001386041754186112,
+      "loss": 1.7031,
+      "step": 4414
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013857759342111428,
+      "loss": 1.5938,
+      "step": 4415
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001385510082207213,
+      "loss": 1.8203,
+      "step": 4416
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013852441981963945,
+      "loss": 1.5215,
+      "step": 4417
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013849782822007626,
+      "loss": 1.6406,
+      "step": 4418
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013847123342423941,
+      "loss": 1.6484,
+      "step": 4419
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013844463543433703,
+      "loss": 1.6328,
+      "step": 4420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013841803425257732,
+      "loss": 1.5,
+      "step": 4421
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013839142988116893,
+      "loss": 1.6699,
+      "step": 4422
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013836482232232059,
+      "loss": 1.6699,
+      "step": 4423
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013833821157824147,
+      "loss": 1.6543,
+      "step": 4424
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013831159765114083,
+      "loss": 1.6738,
+      "step": 4425
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013828498054322838,
+      "loss": 1.6309,
+      "step": 4426
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013825836025671391,
+      "loss": 1.5898,
+      "step": 4427
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013823173679380766,
+      "loss": 1.5059,
+      "step": 4428
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013820511015671996,
+      "loss": 1.4844,
+      "step": 4429
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001381784803476615,
+      "loss": 1.5293,
+      "step": 4430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013815184736884322,
+      "loss": 1.6777,
+      "step": 4431
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013812521122247633,
+      "loss": 1.7266,
+      "step": 4432
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013809857191077227,
+      "loss": 1.6172,
+      "step": 4433
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013807192943594278,
+      "loss": 1.584,
+      "step": 4434
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001380452838001998,
+      "loss": 1.6523,
+      "step": 4435
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013801863500575568,
+      "loss": 1.668,
+      "step": 4436
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001379919830548228,
+      "loss": 1.5879,
+      "step": 4437
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013796532794961402,
+      "loss": 1.5312,
+      "step": 4438
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013793866969234233,
+      "loss": 1.502,
+      "step": 4439
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0013791200828522107,
+      "loss": 1.5977,
+      "step": 4440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013788534373046376,
+      "loss": 1.666,
+      "step": 4441
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013785867603028424,
+      "loss": 1.5996,
+      "step": 4442
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013783200518689659,
+      "loss": 1.6367,
+      "step": 4443
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013780533120251513,
+      "loss": 1.6309,
+      "step": 4444
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013777865407935447,
+      "loss": 1.6836,
+      "step": 4445
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001377519738196295,
+      "loss": 1.5566,
+      "step": 4446
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013772529042555528,
+      "loss": 1.6152,
+      "step": 4447
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013769860389934727,
+      "loss": 1.6465,
+      "step": 4448
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013767191424322105,
+      "loss": 1.7598,
+      "step": 4449
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013764522145939258,
+      "loss": 1.4902,
+      "step": 4450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013761852555007796,
+      "loss": 1.5488,
+      "step": 4451
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013759182651749365,
+      "loss": 1.6953,
+      "step": 4452
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013756512436385634,
+      "loss": 1.6758,
+      "step": 4453
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013753841909138293,
+      "loss": 1.7012,
+      "step": 4454
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013751171070229065,
+      "loss": 1.6055,
+      "step": 4455
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013748499919879697,
+      "loss": 1.8105,
+      "step": 4456
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013745828458311953,
+      "loss": 1.6387,
+      "step": 4457
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013743156685747642,
+      "loss": 1.5527,
+      "step": 4458
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001374048460240858,
+      "loss": 1.6895,
+      "step": 4459
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013737812208516615,
+      "loss": 1.5195,
+      "step": 4460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013735139504293628,
+      "loss": 1.5488,
+      "step": 4461
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013732466489961516,
+      "loss": 1.6855,
+      "step": 4462
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013729793165742203,
+      "loss": 1.5488,
+      "step": 4463
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001372711953185765,
+      "loss": 1.6289,
+      "step": 4464
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013724445588529821,
+      "loss": 1.6875,
+      "step": 4465
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013721771335980732,
+      "loss": 1.666,
+      "step": 4466
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013719096774432405,
+      "loss": 1.6445,
+      "step": 4467
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00137164219041069,
+      "loss": 1.6211,
+      "step": 4468
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013713746725226292,
+      "loss": 1.6836,
+      "step": 4469
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013711071238012692,
+      "loss": 1.8164,
+      "step": 4470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001370839544268823,
+      "loss": 1.5352,
+      "step": 4471
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013705719339475063,
+      "loss": 1.7129,
+      "step": 4472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013703042928595373,
+      "loss": 1.6875,
+      "step": 4473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001370036621027137,
+      "loss": 1.666,
+      "step": 4474
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013697689184725287,
+      "loss": 1.6309,
+      "step": 4475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013695011852179384,
+      "loss": 1.7441,
+      "step": 4476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013692334212855944,
+      "loss": 1.75,
+      "step": 4477
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013689656266977282,
+      "loss": 1.6797,
+      "step": 4478
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013686978014765727,
+      "loss": 1.6211,
+      "step": 4479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013684299456443645,
+      "loss": 1.6523,
+      "step": 4480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013681620592233426,
+      "loss": 1.6211,
+      "step": 4481
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013678941422357474,
+      "loss": 1.582,
+      "step": 4482
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013676261947038232,
+      "loss": 1.666,
+      "step": 4483
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013673582166498159,
+      "loss": 1.6035,
+      "step": 4484
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013670902080959748,
+      "loss": 1.5547,
+      "step": 4485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001366822169064551,
+      "loss": 1.6367,
+      "step": 4486
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013665540995777986,
+      "loss": 1.6914,
+      "step": 4487
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013662859996579732,
+      "loss": 1.7168,
+      "step": 4488
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001366017869327335,
+      "loss": 1.7891,
+      "step": 4489
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013657497086081445,
+      "loss": 1.5215,
+      "step": 4490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013654815175226658,
+      "loss": 1.5508,
+      "step": 4491
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013652132960931657,
+      "loss": 1.5293,
+      "step": 4492
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013649450443419134,
+      "loss": 1.5293,
+      "step": 4493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013646767622911798,
+      "loss": 1.6758,
+      "step": 4494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013644084499632395,
+      "loss": 1.6348,
+      "step": 4495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013641401073803685,
+      "loss": 1.6914,
+      "step": 4496
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013638717345648467,
+      "loss": 1.5859,
+      "step": 4497
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013636033315389548,
+      "loss": 1.6094,
+      "step": 4498
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013633348983249779,
+      "loss": 1.5977,
+      "step": 4499
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013630664349452016,
+      "loss": 1.6289,
+      "step": 4500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001362797941421916,
+      "loss": 1.7344,
+      "step": 4501
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013625294177774115,
+      "loss": 1.6719,
+      "step": 4502
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013622608640339834,
+      "loss": 1.5586,
+      "step": 4503
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013619922802139276,
+      "loss": 1.7012,
+      "step": 4504
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013617236663395437,
+      "loss": 1.6426,
+      "step": 4505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013614550224331325,
+      "loss": 1.5762,
+      "step": 4506
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001361186348516999,
+      "loss": 1.6309,
+      "step": 4507
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013609176446134493,
+      "loss": 1.5762,
+      "step": 4508
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013606489107447928,
+      "loss": 1.6133,
+      "step": 4509
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013603801469333407,
+      "loss": 1.5098,
+      "step": 4510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013601113532014076,
+      "loss": 1.7207,
+      "step": 4511
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013598425295713093,
+      "loss": 1.498,
+      "step": 4512
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013595736760653654,
+      "loss": 1.707,
+      "step": 4513
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001359304792705897,
+      "loss": 1.668,
+      "step": 4514
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013590358795152284,
+      "loss": 1.5859,
+      "step": 4515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001358766936515686,
+      "loss": 1.6211,
+      "step": 4516
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013584979637295987,
+      "loss": 1.6602,
+      "step": 4517
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013582289611792976,
+      "loss": 1.6172,
+      "step": 4518
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001357959928887117,
+      "loss": 1.6289,
+      "step": 4519
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001357690866875393,
+      "loss": 1.6602,
+      "step": 4520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013574217751664648,
+      "loss": 1.6133,
+      "step": 4521
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001357152653782673,
+      "loss": 1.6309,
+      "step": 4522
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013568835027463623,
+      "loss": 1.5977,
+      "step": 4523
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013566143220798777,
+      "loss": 1.6777,
+      "step": 4524
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001356345111805569,
+      "loss": 1.7305,
+      "step": 4525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013560758719457864,
+      "loss": 1.666,
+      "step": 4526
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013558066025228844,
+      "loss": 1.6699,
+      "step": 4527
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001355537303559218,
+      "loss": 1.7285,
+      "step": 4528
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013552679750771467,
+      "loss": 1.5703,
+      "step": 4529
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013549986170990308,
+      "loss": 1.6348,
+      "step": 4530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013547292296472336,
+      "loss": 1.6055,
+      "step": 4531
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013544598127441217,
+      "loss": 1.5898,
+      "step": 4532
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013541903664120625,
+      "loss": 1.5527,
+      "step": 4533
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013539208906734271,
+      "loss": 1.6074,
+      "step": 4534
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013536513855505888,
+      "loss": 1.5391,
+      "step": 4535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013533818510659229,
+      "loss": 1.6816,
+      "step": 4536
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013531122872418077,
+      "loss": 1.6191,
+      "step": 4537
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013528426941006233,
+      "loss": 1.6172,
+      "step": 4538
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013525730716647533,
+      "loss": 1.6035,
+      "step": 4539
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013523034199565823,
+      "loss": 1.5273,
+      "step": 4540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013520337389984986,
+      "loss": 1.5957,
+      "step": 4541
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013517640288128918,
+      "loss": 1.7578,
+      "step": 4542
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013514942894221553,
+      "loss": 1.6289,
+      "step": 4543
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013512245208486837,
+      "loss": 1.6777,
+      "step": 4544
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013509547231148748,
+      "loss": 1.5137,
+      "step": 4545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013506848962431277,
+      "loss": 1.4863,
+      "step": 4546
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013504150402558457,
+      "loss": 1.5664,
+      "step": 4547
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013501451551754327,
+      "loss": 1.6543,
+      "step": 4548
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013498752410242966,
+      "loss": 1.6348,
+      "step": 4549
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013496052978248466,
+      "loss": 1.5938,
+      "step": 4550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013493353255994945,
+      "loss": 1.627,
+      "step": 4551
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0013490653243706549,
+      "loss": 1.5586,
+      "step": 4552
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013487952941607449,
+      "loss": 1.6172,
+      "step": 4553
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013485252349921828,
+      "loss": 1.7617,
+      "step": 4554
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013482551468873912,
+      "loss": 1.8828,
+      "step": 4555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013479850298687933,
+      "loss": 1.5879,
+      "step": 4556
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013477148839588162,
+      "loss": 1.5117,
+      "step": 4557
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001347444709179888,
+      "loss": 1.5625,
+      "step": 4558
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013471745055544406,
+      "loss": 1.5332,
+      "step": 4559
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001346904273104907,
+      "loss": 1.5566,
+      "step": 4560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013466340118537237,
+      "loss": 1.6719,
+      "step": 4561
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013463637218233287,
+      "loss": 1.6016,
+      "step": 4562
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001346093403036163,
+      "loss": 1.5625,
+      "step": 4563
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013458230555146696,
+      "loss": 1.4902,
+      "step": 4564
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001345552679281294,
+      "loss": 1.6348,
+      "step": 4565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013452822743584844,
+      "loss": 1.6367,
+      "step": 4566
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001345011840768691,
+      "loss": 1.6543,
+      "step": 4567
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013447413785343664,
+      "loss": 1.5488,
+      "step": 4568
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013444708876779656,
+      "loss": 1.6016,
+      "step": 4569
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001344200368221946,
+      "loss": 1.6367,
+      "step": 4570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013439298201887678,
+      "loss": 1.6289,
+      "step": 4571
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001343659243600893,
+      "loss": 1.7031,
+      "step": 4572
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013433886384807859,
+      "loss": 1.6328,
+      "step": 4573
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013431180048509136,
+      "loss": 1.5938,
+      "step": 4574
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013428473427337456,
+      "loss": 1.6797,
+      "step": 4575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013425766521517532,
+      "loss": 1.6543,
+      "step": 4576
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013423059331274109,
+      "loss": 1.5781,
+      "step": 4577
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013420351856831946,
+      "loss": 1.6504,
+      "step": 4578
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013417644098415833,
+      "loss": 1.7051,
+      "step": 4579
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013414936056250582,
+      "loss": 1.623,
+      "step": 4580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013412227730561024,
+      "loss": 1.5879,
+      "step": 4581
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001340951912157202,
+      "loss": 1.6797,
+      "step": 4582
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001340681022950845,
+      "loss": 1.6895,
+      "step": 4583
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013404101054595224,
+      "loss": 1.6094,
+      "step": 4584
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013401391597057262,
+      "loss": 1.7051,
+      "step": 4585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013398681857119525,
+      "loss": 1.5957,
+      "step": 4586
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013395971835006982,
+      "loss": 1.5801,
+      "step": 4587
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013393261530944636,
+      "loss": 1.7363,
+      "step": 4588
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013390550945157507,
+      "loss": 1.6133,
+      "step": 4589
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001338784007787064,
+      "loss": 1.5273,
+      "step": 4590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013385128929309107,
+      "loss": 1.5957,
+      "step": 4591
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013382417499698,
+      "loss": 1.5098,
+      "step": 4592
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013379705789262433,
+      "loss": 1.707,
+      "step": 4593
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013376993798227548,
+      "loss": 1.6133,
+      "step": 4594
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013374281526818503,
+      "loss": 1.7422,
+      "step": 4595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013371568975260489,
+      "loss": 1.582,
+      "step": 4596
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001336885614377871,
+      "loss": 1.6504,
+      "step": 4597
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013366143032598403,
+      "loss": 1.5312,
+      "step": 4598
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013363429641944818,
+      "loss": 1.4902,
+      "step": 4599
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001336071597204324,
+      "loss": 1.625,
+      "step": 4600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013358002023118965,
+      "loss": 1.7324,
+      "step": 4601
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013355287795397324,
+      "loss": 1.5488,
+      "step": 4602
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013352573289103658,
+      "loss": 1.5723,
+      "step": 4603
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013349858504463344,
+      "loss": 1.5137,
+      "step": 4604
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013347143441701772,
+      "loss": 1.6035,
+      "step": 4605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013344428101044366,
+      "loss": 1.6426,
+      "step": 4606
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001334171248271656,
+      "loss": 1.5176,
+      "step": 4607
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001333899658694382,
+      "loss": 1.543,
+      "step": 4608
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013336280413951635,
+      "loss": 1.5508,
+      "step": 4609
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001333356396396551,
+      "loss": 1.6016,
+      "step": 4610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013330847237210978,
+      "loss": 1.6953,
+      "step": 4611
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013328130233913602,
+      "loss": 1.5098,
+      "step": 4612
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013325412954298952,
+      "loss": 1.6934,
+      "step": 4613
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013322695398592633,
+      "loss": 1.582,
+      "step": 4614
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013319977567020267,
+      "loss": 1.6094,
+      "step": 4615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013317259459807507,
+      "loss": 1.627,
+      "step": 4616
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013314541077180018,
+      "loss": 1.6523,
+      "step": 4617
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013311822419363498,
+      "loss": 1.6289,
+      "step": 4618
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013309103486583657,
+      "loss": 1.6094,
+      "step": 4619
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013306384279066236,
+      "loss": 1.6523,
+      "step": 4620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013303664797036999,
+      "loss": 1.6211,
+      "step": 4621
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013300945040721729,
+      "loss": 1.5527,
+      "step": 4622
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001329822501034623,
+      "loss": 1.5879,
+      "step": 4623
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013295504706136337,
+      "loss": 1.625,
+      "step": 4624
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00132927841283179,
+      "loss": 1.5957,
+      "step": 4625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013290063277116796,
+      "loss": 1.5762,
+      "step": 4626
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001328734215275892,
+      "loss": 1.7324,
+      "step": 4627
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013284620755470195,
+      "loss": 1.7598,
+      "step": 4628
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013281899085476564,
+      "loss": 1.5332,
+      "step": 4629
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013279177143003993,
+      "loss": 1.5684,
+      "step": 4630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013276454928278472,
+      "loss": 1.5918,
+      "step": 4631
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001327373244152601,
+      "loss": 1.4395,
+      "step": 4632
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001327100968297264,
+      "loss": 1.5723,
+      "step": 4633
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013268286652844425,
+      "loss": 1.6289,
+      "step": 4634
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013265563351367434,
+      "loss": 1.6016,
+      "step": 4635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013262839778767777,
+      "loss": 1.5801,
+      "step": 4636
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013260115935271573,
+      "loss": 1.6562,
+      "step": 4637
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001325739182110497,
+      "loss": 1.5859,
+      "step": 4638
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001325466743649414,
+      "loss": 1.6328,
+      "step": 4639
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001325194278166527,
+      "loss": 1.5898,
+      "step": 4640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013249217856844573,
+      "loss": 1.6211,
+      "step": 4641
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013246492662258292,
+      "loss": 1.668,
+      "step": 4642
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013243767198132677,
+      "loss": 1.625,
+      "step": 4643
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013241041464694017,
+      "loss": 1.5,
+      "step": 4644
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001323831546216861,
+      "loss": 1.6758,
+      "step": 4645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013235589190782784,
+      "loss": 1.8438,
+      "step": 4646
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013232862650762887,
+      "loss": 1.5684,
+      "step": 4647
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001323013584233529,
+      "loss": 1.7031,
+      "step": 4648
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001322740876572638,
+      "loss": 1.6738,
+      "step": 4649
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013224681421162583,
+      "loss": 1.5195,
+      "step": 4650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013221953808870328,
+      "loss": 1.5723,
+      "step": 4651
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013219225929076075,
+      "loss": 1.584,
+      "step": 4652
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001321649778200631,
+      "loss": 1.666,
+      "step": 4653
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013213769367887527,
+      "loss": 1.4688,
+      "step": 4654
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013211040686946266,
+      "loss": 1.6074,
+      "step": 4655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013208311739409063,
+      "loss": 1.6543,
+      "step": 4656
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013205582525502493,
+      "loss": 1.6621,
+      "step": 4657
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001320285304545315,
+      "loss": 1.5527,
+      "step": 4658
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013200123299487647,
+      "loss": 1.6777,
+      "step": 4659
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001319739328783262,
+      "loss": 1.5234,
+      "step": 4660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001319466301071473,
+      "loss": 1.4531,
+      "step": 4661
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013191932468360652,
+      "loss": 1.6602,
+      "step": 4662
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013189201660997094,
+      "loss": 1.5859,
+      "step": 4663
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013186470588850779,
+      "loss": 1.5234,
+      "step": 4664
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0013183739252148453,
+      "loss": 1.6758,
+      "step": 4665
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013181007651116888,
+      "loss": 1.7363,
+      "step": 4666
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001317827578598287,
+      "loss": 1.6309,
+      "step": 4667
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013175543656973215,
+      "loss": 1.5176,
+      "step": 4668
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013172811264314753,
+      "loss": 1.543,
+      "step": 4669
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013170078608234347,
+      "loss": 1.6504,
+      "step": 4670
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013167345688958867,
+      "loss": 1.7051,
+      "step": 4671
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013164612506715216,
+      "loss": 1.5176,
+      "step": 4672
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013161879061730321,
+      "loss": 1.6016,
+      "step": 4673
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013159145354231122,
+      "loss": 1.709,
+      "step": 4674
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013156411384444581,
+      "loss": 1.5098,
+      "step": 4675
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001315367715259769,
+      "loss": 1.4863,
+      "step": 4676
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013150942658917458,
+      "loss": 1.5996,
+      "step": 4677
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013148207903630911,
+      "loss": 1.6094,
+      "step": 4678
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013145472886965108,
+      "loss": 1.6523,
+      "step": 4679
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013142737609147118,
+      "loss": 1.5449,
+      "step": 4680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013140002070404038,
+      "loss": 1.6992,
+      "step": 4681
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013137266270962986,
+      "loss": 1.5312,
+      "step": 4682
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013134530211051103,
+      "loss": 1.5996,
+      "step": 4683
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013131793890895547,
+      "loss": 1.6074,
+      "step": 4684
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00131290573107235,
+      "loss": 1.6543,
+      "step": 4685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013126320470762168,
+      "loss": 1.7188,
+      "step": 4686
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013123583371238779,
+      "loss": 1.668,
+      "step": 4687
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013120846012380572,
+      "loss": 1.6621,
+      "step": 4688
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013118108394414823,
+      "loss": 1.7168,
+      "step": 4689
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013115370517568822,
+      "loss": 1.6504,
+      "step": 4690
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013112632382069876,
+      "loss": 1.4863,
+      "step": 4691
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013109893988145318,
+      "loss": 1.6523,
+      "step": 4692
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001310715533602251,
+      "loss": 1.7324,
+      "step": 4693
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013104416425928815,
+      "loss": 1.4746,
+      "step": 4694
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013101677258091646,
+      "loss": 1.7129,
+      "step": 4695
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013098937832738409,
+      "loss": 1.666,
+      "step": 4696
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013096198150096552,
+      "loss": 1.5957,
+      "step": 4697
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013093458210393534,
+      "loss": 1.7168,
+      "step": 4698
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013090718013856837,
+      "loss": 1.6875,
+      "step": 4699
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013087977560713965,
+      "loss": 1.5508,
+      "step": 4700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013085236851192446,
+      "loss": 1.6484,
+      "step": 4701
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013082495885519824,
+      "loss": 1.6211,
+      "step": 4702
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013079754663923669,
+      "loss": 1.7246,
+      "step": 4703
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013077013186631565,
+      "loss": 1.6895,
+      "step": 4704
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013074271453871135,
+      "loss": 1.5957,
+      "step": 4705
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013071529465869995,
+      "loss": 1.707,
+      "step": 4706
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013068787222855812,
+      "loss": 1.6289,
+      "step": 4707
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001306604472505625,
+      "loss": 1.5312,
+      "step": 4708
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001306330197269901,
+      "loss": 1.707,
+      "step": 4709
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013060558966011809,
+      "loss": 1.6113,
+      "step": 4710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001305781570522238,
+      "loss": 1.6133,
+      "step": 4711
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013055072190558481,
+      "loss": 1.5117,
+      "step": 4712
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013052328422247898,
+      "loss": 1.5625,
+      "step": 4713
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013049584400518426,
+      "loss": 1.6602,
+      "step": 4714
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013046840125597891,
+      "loss": 1.5762,
+      "step": 4715
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013044095597714131,
+      "loss": 1.7246,
+      "step": 4716
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013041350817095015,
+      "loss": 1.6035,
+      "step": 4717
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013038605783968426,
+      "loss": 1.6348,
+      "step": 4718
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013035860498562266,
+      "loss": 1.6992,
+      "step": 4719
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013033114961104467,
+      "loss": 1.5215,
+      "step": 4720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013030369171822974,
+      "loss": 1.6152,
+      "step": 4721
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013027623130945755,
+      "loss": 1.5566,
+      "step": 4722
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013024876838700803,
+      "loss": 1.7344,
+      "step": 4723
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013022130295316124,
+      "loss": 1.6543,
+      "step": 4724
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001301938350101975,
+      "loss": 1.6797,
+      "step": 4725
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013016636456039735,
+      "loss": 1.7559,
+      "step": 4726
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013013889160604151,
+      "loss": 1.5625,
+      "step": 4727
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013011141614941088,
+      "loss": 1.4375,
+      "step": 4728
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001300839381927867,
+      "loss": 1.5098,
+      "step": 4729
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013005645773845023,
+      "loss": 1.6055,
+      "step": 4730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013002897478868304,
+      "loss": 1.6445,
+      "step": 4731
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0013000148934576691,
+      "loss": 1.6094,
+      "step": 4732
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012997400141198385,
+      "loss": 1.6055,
+      "step": 4733
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012994651098961597,
+      "loss": 1.7383,
+      "step": 4734
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001299190180809457,
+      "loss": 1.6875,
+      "step": 4735
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012989152268825565,
+      "loss": 1.4668,
+      "step": 4736
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012986402481382855,
+      "loss": 1.7188,
+      "step": 4737
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012983652445994747,
+      "loss": 1.5977,
+      "step": 4738
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001298090216288956,
+      "loss": 1.6934,
+      "step": 4739
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012978151632295635,
+      "loss": 1.541,
+      "step": 4740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012975400854441333,
+      "loss": 1.582,
+      "step": 4741
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012972649829555041,
+      "loss": 1.5664,
+      "step": 4742
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012969898557865156,
+      "loss": 1.4961,
+      "step": 4743
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001296714703960011,
+      "loss": 1.709,
+      "step": 4744
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012964395274988338,
+      "loss": 1.6309,
+      "step": 4745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001296164326425831,
+      "loss": 1.5586,
+      "step": 4746
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012958891007638514,
+      "loss": 1.5938,
+      "step": 4747
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012956138505357448,
+      "loss": 1.6992,
+      "step": 4748
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012953385757643644,
+      "loss": 1.5254,
+      "step": 4749
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012950632764725644,
+      "loss": 1.7383,
+      "step": 4750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012947879526832016,
+      "loss": 1.666,
+      "step": 4751
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001294512604419135,
+      "loss": 1.7969,
+      "step": 4752
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001294237231703225,
+      "loss": 1.5586,
+      "step": 4753
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012939618345583345,
+      "loss": 1.5059,
+      "step": 4754
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012936864130073276,
+      "loss": 1.5117,
+      "step": 4755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012934109670730723,
+      "loss": 1.6602,
+      "step": 4756
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012931354967784365,
+      "loss": 1.6289,
+      "step": 4757
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012928600021462919,
+      "loss": 1.6816,
+      "step": 4758
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012925844831995106,
+      "loss": 1.5645,
+      "step": 4759
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001292308939960968,
+      "loss": 1.5996,
+      "step": 4760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012920333724535404,
+      "loss": 1.5605,
+      "step": 4761
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012917577807001078,
+      "loss": 1.4863,
+      "step": 4762
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00129148216472355,
+      "loss": 1.6543,
+      "step": 4763
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012912065245467505,
+      "loss": 1.6387,
+      "step": 4764
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012909308601925943,
+      "loss": 1.5781,
+      "step": 4765
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012906551716839682,
+      "loss": 1.6328,
+      "step": 4766
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012903794590437615,
+      "loss": 1.7168,
+      "step": 4767
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012901037222948649,
+      "loss": 1.6973,
+      "step": 4768
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012898279614601714,
+      "loss": 1.6934,
+      "step": 4769
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001289552176562576,
+      "loss": 1.6348,
+      "step": 4770
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012892763676249754,
+      "loss": 1.6074,
+      "step": 4771
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012890005346702692,
+      "loss": 1.5176,
+      "step": 4772
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001288724677721358,
+      "loss": 1.6211,
+      "step": 4773
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012884487968011444,
+      "loss": 1.6543,
+      "step": 4774
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001288172891932534,
+      "loss": 1.582,
+      "step": 4775
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012878969631384338,
+      "loss": 1.666,
+      "step": 4776
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0012876210104417521,
+      "loss": 1.6406,
+      "step": 4777
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012873450338654004,
+      "loss": 1.6777,
+      "step": 4778
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012870690334322909,
+      "loss": 1.6523,
+      "step": 4779
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012867930091653391,
+      "loss": 1.5605,
+      "step": 4780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012865169610874615,
+      "loss": 1.5703,
+      "step": 4781
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012862408892215774,
+      "loss": 1.7129,
+      "step": 4782
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001285964793590607,
+      "loss": 1.7246,
+      "step": 4783
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012856886742174734,
+      "loss": 1.5645,
+      "step": 4784
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012854125311251012,
+      "loss": 1.625,
+      "step": 4785
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012851363643364175,
+      "loss": 1.6289,
+      "step": 4786
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012848601738743503,
+      "loss": 1.6055,
+      "step": 4787
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012845839597618312,
+      "loss": 1.6953,
+      "step": 4788
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012843077220217923,
+      "loss": 1.6836,
+      "step": 4789
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012840314606771678,
+      "loss": 1.5977,
+      "step": 4790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012837551757508944,
+      "loss": 1.75,
+      "step": 4791
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012834788672659112,
+      "loss": 1.5742,
+      "step": 4792
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001283202535245158,
+      "loss": 1.5391,
+      "step": 4793
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012829261797115779,
+      "loss": 1.582,
+      "step": 4794
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012826498006881142,
+      "loss": 1.5156,
+      "step": 4795
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012823733981977142,
+      "loss": 1.6758,
+      "step": 4796
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012820969722633256,
+      "loss": 1.6543,
+      "step": 4797
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012818205229078988,
+      "loss": 1.5137,
+      "step": 4798
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012815440501543858,
+      "loss": 1.6602,
+      "step": 4799
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001281267554025741,
+      "loss": 1.5547,
+      "step": 4800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00128099103454492,
+      "loss": 1.6465,
+      "step": 4801
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012807144917348812,
+      "loss": 1.6543,
+      "step": 4802
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001280437925618584,
+      "loss": 1.5156,
+      "step": 4803
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001280161336218991,
+      "loss": 1.6445,
+      "step": 4804
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012798847235590649,
+      "loss": 1.7754,
+      "step": 4805
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012796080876617726,
+      "loss": 1.6641,
+      "step": 4806
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012793314285500807,
+      "loss": 1.582,
+      "step": 4807
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012790547462469598,
+      "loss": 1.6836,
+      "step": 4808
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012787780407753804,
+      "loss": 1.6055,
+      "step": 4809
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012785013121583162,
+      "loss": 1.5977,
+      "step": 4810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012782245604187428,
+      "loss": 1.625,
+      "step": 4811
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012779477855796376,
+      "loss": 1.7539,
+      "step": 4812
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012776709876639789,
+      "loss": 1.6152,
+      "step": 4813
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012773941666947486,
+      "loss": 1.5,
+      "step": 4814
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012771173226949295,
+      "loss": 1.5605,
+      "step": 4815
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012768404556875064,
+      "loss": 1.6602,
+      "step": 4816
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001276563565695466,
+      "loss": 1.7383,
+      "step": 4817
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012762866527417974,
+      "loss": 1.6582,
+      "step": 4818
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001276009716849491,
+      "loss": 1.668,
+      "step": 4819
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012757327580415394,
+      "loss": 1.5898,
+      "step": 4820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001275455776340937,
+      "loss": 1.5195,
+      "step": 4821
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012751787717706802,
+      "loss": 1.5938,
+      "step": 4822
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012749017443537674,
+      "loss": 1.5898,
+      "step": 4823
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012746246941131982,
+      "loss": 1.6582,
+      "step": 4824
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012743476210719754,
+      "loss": 1.6152,
+      "step": 4825
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001274070525253102,
+      "loss": 1.6875,
+      "step": 4826
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012737934066795846,
+      "loss": 1.8066,
+      "step": 4827
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001273516265374431,
+      "loss": 1.5547,
+      "step": 4828
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012732391013606501,
+      "loss": 1.6055,
+      "step": 4829
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012729619146612536,
+      "loss": 1.5996,
+      "step": 4830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012726847052992555,
+      "loss": 1.4922,
+      "step": 4831
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012724074732976699,
+      "loss": 1.6484,
+      "step": 4832
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012721302186795148,
+      "loss": 1.6367,
+      "step": 4833
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012718529414678088,
+      "loss": 1.584,
+      "step": 4834
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012715756416855732,
+      "loss": 1.5801,
+      "step": 4835
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012712983193558303,
+      "loss": 1.6641,
+      "step": 4836
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001271020974501605,
+      "loss": 1.6465,
+      "step": 4837
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012707436071459238,
+      "loss": 1.582,
+      "step": 4838
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012704662173118149,
+      "loss": 1.6465,
+      "step": 4839
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012701888050223086,
+      "loss": 1.6289,
+      "step": 4840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012699113703004369,
+      "loss": 1.5449,
+      "step": 4841
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012696339131692339,
+      "loss": 1.6133,
+      "step": 4842
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012693564336517353,
+      "loss": 1.7168,
+      "step": 4843
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012690789317709788,
+      "loss": 1.6973,
+      "step": 4844
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012688014075500043,
+      "loss": 1.627,
+      "step": 4845
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012685238610118527,
+      "loss": 1.5605,
+      "step": 4846
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012682462921795672,
+      "loss": 1.5977,
+      "step": 4847
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012679687010761934,
+      "loss": 1.7188,
+      "step": 4848
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012676910877247778,
+      "loss": 1.6387,
+      "step": 4849
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001267413452148369,
+      "loss": 1.6562,
+      "step": 4850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012671357943700183,
+      "loss": 1.5898,
+      "step": 4851
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012668581144127777,
+      "loss": 1.5977,
+      "step": 4852
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012665804122997018,
+      "loss": 1.5254,
+      "step": 4853
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012663026880538465,
+      "loss": 1.6973,
+      "step": 4854
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012660249416982699,
+      "loss": 1.5879,
+      "step": 4855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012657471732560317,
+      "loss": 1.6328,
+      "step": 4856
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012654693827501937,
+      "loss": 1.6816,
+      "step": 4857
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012651915702038194,
+      "loss": 1.7266,
+      "step": 4858
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012649137356399742,
+      "loss": 1.6562,
+      "step": 4859
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001264635879081725,
+      "loss": 1.5469,
+      "step": 4860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012643580005521412,
+      "loss": 1.5996,
+      "step": 4861
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001264080100074293,
+      "loss": 1.7285,
+      "step": 4862
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012638021776712536,
+      "loss": 1.6797,
+      "step": 4863
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001263524233366097,
+      "loss": 1.6582,
+      "step": 4864
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012632462671818999,
+      "loss": 1.5723,
+      "step": 4865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012629682791417398,
+      "loss": 1.5566,
+      "step": 4866
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012626902692686975,
+      "loss": 1.582,
+      "step": 4867
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001262412237585854,
+      "loss": 1.5801,
+      "step": 4868
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012621341841162928,
+      "loss": 1.6543,
+      "step": 4869
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012618561088830994,
+      "loss": 1.7227,
+      "step": 4870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001261578011909361,
+      "loss": 1.7227,
+      "step": 4871
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012612998932181663,
+      "loss": 1.4922,
+      "step": 4872
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012610217528326061,
+      "loss": 1.5762,
+      "step": 4873
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012607435907757733,
+      "loss": 1.5371,
+      "step": 4874
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012604654070707622,
+      "loss": 1.7305,
+      "step": 4875
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012601872017406683,
+      "loss": 1.6465,
+      "step": 4876
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012599089748085903,
+      "loss": 1.666,
+      "step": 4877
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012596307262976274,
+      "loss": 1.6309,
+      "step": 4878
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012593524562308817,
+      "loss": 1.5527,
+      "step": 4879
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012590741646314556,
+      "loss": 1.6133,
+      "step": 4880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001258795851522455,
+      "loss": 1.6074,
+      "step": 4881
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012585175169269865,
+      "loss": 1.582,
+      "step": 4882
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012582391608681588,
+      "loss": 1.7422,
+      "step": 4883
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012579607833690823,
+      "loss": 1.6191,
+      "step": 4884
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012576823844528693,
+      "loss": 1.6445,
+      "step": 4885
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012574039641426338,
+      "loss": 1.5859,
+      "step": 4886
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012571255224614916,
+      "loss": 1.5293,
+      "step": 4887
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012568470594325602,
+      "loss": 1.7617,
+      "step": 4888
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0012565685750789589,
+      "loss": 1.6309,
+      "step": 4889
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001256290069423809,
+      "loss": 1.6484,
+      "step": 4890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012560115424902329,
+      "loss": 1.4746,
+      "step": 4891
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012557329943013555,
+      "loss": 1.4961,
+      "step": 4892
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012554544248803037,
+      "loss": 1.5527,
+      "step": 4893
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012551758342502048,
+      "loss": 1.5801,
+      "step": 4894
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012548972224341894,
+      "loss": 1.6426,
+      "step": 4895
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012546185894553887,
+      "loss": 1.625,
+      "step": 4896
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012543399353369367,
+      "loss": 1.5449,
+      "step": 4897
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001254061260101968,
+      "loss": 1.6387,
+      "step": 4898
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00125378256377362,
+      "loss": 1.6523,
+      "step": 4899
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001253503846375031,
+      "loss": 1.6289,
+      "step": 4900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001253225107929342,
+      "loss": 1.5586,
+      "step": 4901
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012529463484596945,
+      "loss": 1.584,
+      "step": 4902
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012526675679892327,
+      "loss": 1.6523,
+      "step": 4903
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012523887665411028,
+      "loss": 1.625,
+      "step": 4904
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012521099441384516,
+      "loss": 1.6406,
+      "step": 4905
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012518311008044285,
+      "loss": 1.6055,
+      "step": 4906
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012515522365621847,
+      "loss": 1.5527,
+      "step": 4907
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012512733514348723,
+      "loss": 1.5547,
+      "step": 4908
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001250994445445646,
+      "loss": 1.5801,
+      "step": 4909
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012507155186176617,
+      "loss": 1.5156,
+      "step": 4910
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012504365709740772,
+      "loss": 1.6875,
+      "step": 4911
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012501576025380527,
+      "loss": 1.7695,
+      "step": 4912
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012498786133327484,
+      "loss": 1.6191,
+      "step": 4913
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012495996033813283,
+      "loss": 1.8145,
+      "step": 4914
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012493205727069565,
+      "loss": 1.5254,
+      "step": 4915
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012490415213328,
+      "loss": 1.5547,
+      "step": 4916
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012487624492820266,
+      "loss": 1.6543,
+      "step": 4917
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012484833565778064,
+      "loss": 1.5449,
+      "step": 4918
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012482042432433106,
+      "loss": 1.6914,
+      "step": 4919
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012479251093017135,
+      "loss": 1.6816,
+      "step": 4920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001247645954776189,
+      "loss": 1.5742,
+      "step": 4921
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012473667796899144,
+      "loss": 1.6816,
+      "step": 4922
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012470875840660679,
+      "loss": 1.5293,
+      "step": 4923
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00124680836792783,
+      "loss": 1.6152,
+      "step": 4924
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012465291312983822,
+      "loss": 1.5898,
+      "step": 4925
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012462498742009083,
+      "loss": 1.5996,
+      "step": 4926
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012459705966585937,
+      "loss": 1.707,
+      "step": 4927
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001245691298694625,
+      "loss": 1.5645,
+      "step": 4928
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012454119803321907,
+      "loss": 1.5156,
+      "step": 4929
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001245132641594482,
+      "loss": 1.6406,
+      "step": 4930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00124485328250469,
+      "loss": 1.5957,
+      "step": 4931
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012445739030860089,
+      "loss": 1.5176,
+      "step": 4932
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001244294503361634,
+      "loss": 1.6113,
+      "step": 4933
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012440150833547623,
+      "loss": 1.5879,
+      "step": 4934
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012437356430885927,
+      "loss": 1.5762,
+      "step": 4935
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001243456182586326,
+      "loss": 1.6699,
+      "step": 4936
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012431767018711635,
+      "loss": 1.6504,
+      "step": 4937
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00124289720096631,
+      "loss": 1.541,
+      "step": 4938
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012426176798949703,
+      "loss": 1.6387,
+      "step": 4939
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012423381386803519,
+      "loss": 1.7227,
+      "step": 4940
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012420585773456633,
+      "loss": 1.7871,
+      "step": 4941
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012417789959141157,
+      "loss": 1.623,
+      "step": 4942
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012414993944089205,
+      "loss": 1.707,
+      "step": 4943
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012412197728532924,
+      "loss": 1.6738,
+      "step": 4944
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012409401312704457,
+      "loss": 1.6797,
+      "step": 4945
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001240660469683599,
+      "loss": 1.7305,
+      "step": 4946
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012403807881159704,
+      "loss": 1.7148,
+      "step": 4947
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00124010108659078,
+      "loss": 1.752,
+      "step": 4948
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012398213651312506,
+      "loss": 1.6934,
+      "step": 4949
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012395416237606061,
+      "loss": 1.7383,
+      "step": 4950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012392618625020712,
+      "loss": 1.5254,
+      "step": 4951
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001238982081378874,
+      "loss": 1.6816,
+      "step": 4952
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012387022804142423,
+      "loss": 1.6914,
+      "step": 4953
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012384224596314072,
+      "loss": 1.6602,
+      "step": 4954
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012381426190536005,
+      "loss": 1.5684,
+      "step": 4955
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012378627587040561,
+      "loss": 1.6543,
+      "step": 4956
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012375828786060088,
+      "loss": 1.7461,
+      "step": 4957
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012373029787826964,
+      "loss": 1.5234,
+      "step": 4958
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012370230592573567,
+      "loss": 1.4824,
+      "step": 4959
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012367431200532306,
+      "loss": 1.502,
+      "step": 4960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012364631611935595,
+      "loss": 1.6348,
+      "step": 4961
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012361831827015874,
+      "loss": 1.5859,
+      "step": 4962
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001235903184600559,
+      "loss": 1.7012,
+      "step": 4963
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012356231669137217,
+      "loss": 1.6191,
+      "step": 4964
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001235343129664323,
+      "loss": 1.6621,
+      "step": 4965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012350630728756137,
+      "loss": 1.6484,
+      "step": 4966
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012347829965708452,
+      "loss": 1.5039,
+      "step": 4967
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012345029007732706,
+      "loss": 1.6855,
+      "step": 4968
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001234222785506145,
+      "loss": 1.6621,
+      "step": 4969
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012339426507927249,
+      "loss": 1.5742,
+      "step": 4970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012336624966562681,
+      "loss": 1.7012,
+      "step": 4971
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001233382323120035,
+      "loss": 1.6406,
+      "step": 4972
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012331021302072862,
+      "loss": 1.6641,
+      "step": 4973
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012328219179412853,
+      "loss": 1.5762,
+      "step": 4974
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012325416863452963,
+      "loss": 1.7305,
+      "step": 4975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012322614354425858,
+      "loss": 1.6426,
+      "step": 4976
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012319811652564212,
+      "loss": 1.752,
+      "step": 4977
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012317008758100723,
+      "loss": 1.6953,
+      "step": 4978
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012314205671268097,
+      "loss": 1.5645,
+      "step": 4979
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012311402392299062,
+      "loss": 1.5273,
+      "step": 4980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012308598921426358,
+      "loss": 1.6055,
+      "step": 4981
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012305795258882744,
+      "loss": 1.5918,
+      "step": 4982
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001230299140490099,
+      "loss": 1.625,
+      "step": 4983
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012300187359713891,
+      "loss": 1.5879,
+      "step": 4984
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012297383123554247,
+      "loss": 1.5117,
+      "step": 4985
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012294578696654883,
+      "loss": 1.582,
+      "step": 4986
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012291774079248635,
+      "loss": 1.6797,
+      "step": 4987
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012288969271568354,
+      "loss": 1.7305,
+      "step": 4988
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012286164273846909,
+      "loss": 1.5801,
+      "step": 4989
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012283359086317184,
+      "loss": 1.5723,
+      "step": 4990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001228055370921208,
+      "loss": 1.5898,
+      "step": 4991
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012277748142764512,
+      "loss": 1.6543,
+      "step": 4992
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012274942387207415,
+      "loss": 1.6133,
+      "step": 4993
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012272136442773732,
+      "loss": 1.6309,
+      "step": 4994
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012269330309696428,
+      "loss": 1.6387,
+      "step": 4995
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012266523988208478,
+      "loss": 1.748,
+      "step": 4996
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012263717478542882,
+      "loss": 1.5938,
+      "step": 4997
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012260910780932646,
+      "loss": 1.4629,
+      "step": 4998
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012258103895610797,
+      "loss": 1.6348,
+      "step": 4999
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012255296822810374,
+      "loss": 1.6445,
+      "step": 5000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012252489562764436,
+      "loss": 1.6387,
+      "step": 5001
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0012249682115706055,
+      "loss": 1.6621,
+      "step": 5002
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012246874481868317,
+      "loss": 1.6602,
+      "step": 5003
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012244066661484326,
+      "loss": 1.7109,
+      "step": 5004
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012241258654787201,
+      "loss": 1.5371,
+      "step": 5005
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012238450462010077,
+      "loss": 1.6211,
+      "step": 5006
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012235642083386104,
+      "loss": 1.5723,
+      "step": 5007
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001223283351914844,
+      "loss": 1.6113,
+      "step": 5008
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012230024769530274,
+      "loss": 1.5684,
+      "step": 5009
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012227215834764798,
+      "loss": 1.6465,
+      "step": 5010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012224406715085224,
+      "loss": 1.4805,
+      "step": 5011
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012221597410724777,
+      "loss": 1.5996,
+      "step": 5012
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012218787921916701,
+      "loss": 1.707,
+      "step": 5013
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012215978248894253,
+      "loss": 1.5137,
+      "step": 5014
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012213168391890705,
+      "loss": 1.5859,
+      "step": 5015
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012210358351139343,
+      "loss": 1.625,
+      "step": 5016
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012207548126873474,
+      "loss": 1.6113,
+      "step": 5017
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001220473771932641,
+      "loss": 1.6406,
+      "step": 5018
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012201927128731493,
+      "loss": 1.6816,
+      "step": 5019
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012199116355322065,
+      "loss": 1.6035,
+      "step": 5020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012196305399331492,
+      "loss": 1.5918,
+      "step": 5021
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001219349426099315,
+      "loss": 1.5664,
+      "step": 5022
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012190682940540442,
+      "loss": 1.6875,
+      "step": 5023
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012187871438206763,
+      "loss": 1.5957,
+      "step": 5024
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012185059754225552,
+      "loss": 1.5156,
+      "step": 5025
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012182247888830242,
+      "loss": 1.5605,
+      "step": 5026
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012179435842254284,
+      "loss": 1.5391,
+      "step": 5027
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012176623614731155,
+      "loss": 1.5762,
+      "step": 5028
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012173811206494334,
+      "loss": 1.6797,
+      "step": 5029
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001217099861777732,
+      "loss": 1.6309,
+      "step": 5030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012168185848813633,
+      "loss": 1.6094,
+      "step": 5031
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012165372899836796,
+      "loss": 1.6973,
+      "step": 5032
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012162559771080358,
+      "loss": 1.5508,
+      "step": 5033
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012159746462777877,
+      "loss": 1.5684,
+      "step": 5034
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001215693297516293,
+      "loss": 1.6133,
+      "step": 5035
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012154119308469098,
+      "loss": 1.5859,
+      "step": 5036
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012151305462929996,
+      "loss": 1.5781,
+      "step": 5037
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012148491438779233,
+      "loss": 1.5605,
+      "step": 5038
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001214567723625045,
+      "loss": 1.7832,
+      "step": 5039
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012142862855577293,
+      "loss": 1.6484,
+      "step": 5040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012140048296993425,
+      "loss": 1.4961,
+      "step": 5041
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012137233560732521,
+      "loss": 1.6699,
+      "step": 5042
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001213441864702828,
+      "loss": 1.6797,
+      "step": 5043
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012131603556114405,
+      "loss": 1.6094,
+      "step": 5044
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012128788288224624,
+      "loss": 1.5684,
+      "step": 5045
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012125972843592668,
+      "loss": 1.5605,
+      "step": 5046
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012123157222452287,
+      "loss": 1.6152,
+      "step": 5047
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012120341425037255,
+      "loss": 1.6777,
+      "step": 5048
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012117525451581347,
+      "loss": 1.623,
+      "step": 5049
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001211470930231836,
+      "loss": 1.6367,
+      "step": 5050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001211189297748211,
+      "loss": 1.6465,
+      "step": 5051
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012109076477306408,
+      "loss": 1.5859,
+      "step": 5052
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012106259802025106,
+      "loss": 1.6816,
+      "step": 5053
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012103442951872053,
+      "loss": 1.6074,
+      "step": 5054
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012100625927081118,
+      "loss": 1.5762,
+      "step": 5055
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012097808727886182,
+      "loss": 1.5996,
+      "step": 5056
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012094991354521145,
+      "loss": 1.6914,
+      "step": 5057
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001209217380721992,
+      "loss": 1.584,
+      "step": 5058
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012089356086216428,
+      "loss": 1.6113,
+      "step": 5059
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012086538191744614,
+      "loss": 1.6953,
+      "step": 5060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001208372012403843,
+      "loss": 1.6465,
+      "step": 5061
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001208090188333185,
+      "loss": 1.5098,
+      "step": 5062
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012078083469858854,
+      "loss": 1.5781,
+      "step": 5063
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001207526488385344,
+      "loss": 1.6094,
+      "step": 5064
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012072446125549624,
+      "loss": 1.7168,
+      "step": 5065
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001206962719518143,
+      "loss": 1.5762,
+      "step": 5066
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012066808092982897,
+      "loss": 1.6836,
+      "step": 5067
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012063988819188086,
+      "loss": 1.6758,
+      "step": 5068
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012061169374031062,
+      "loss": 1.627,
+      "step": 5069
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012058349757745907,
+      "loss": 1.5273,
+      "step": 5070
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012055529970566728,
+      "loss": 1.6074,
+      "step": 5071
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012052710012727627,
+      "loss": 1.4941,
+      "step": 5072
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012049889884462736,
+      "loss": 1.7266,
+      "step": 5073
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012047069586006195,
+      "loss": 1.5039,
+      "step": 5074
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012044249117592156,
+      "loss": 1.6562,
+      "step": 5075
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012041428479454793,
+      "loss": 1.6328,
+      "step": 5076
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001203860767182828,
+      "loss": 1.459,
+      "step": 5077
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012035786694946826,
+      "loss": 1.7305,
+      "step": 5078
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012032965549044633,
+      "loss": 1.793,
+      "step": 5079
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012030144234355929,
+      "loss": 1.5625,
+      "step": 5080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012027322751114948,
+      "loss": 1.5957,
+      "step": 5081
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012024501099555953,
+      "loss": 1.6094,
+      "step": 5082
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012021679279913202,
+      "loss": 1.4863,
+      "step": 5083
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012018857292420983,
+      "loss": 1.4688,
+      "step": 5084
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012016035137313585,
+      "loss": 1.6016,
+      "step": 5085
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012013212814825321,
+      "loss": 1.5664,
+      "step": 5086
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001201039032519051,
+      "loss": 1.5254,
+      "step": 5087
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012007567668643492,
+      "loss": 1.6504,
+      "step": 5088
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012004744845418613,
+      "loss": 1.5879,
+      "step": 5089
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0012001921855750244,
+      "loss": 1.8027,
+      "step": 5090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011999098699872755,
+      "loss": 1.7461,
+      "step": 5091
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011996275378020544,
+      "loss": 1.582,
+      "step": 5092
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011993451890428015,
+      "loss": 1.5977,
+      "step": 5093
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011990628237329589,
+      "loss": 1.7227,
+      "step": 5094
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011987804418959695,
+      "loss": 1.5684,
+      "step": 5095
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011984980435552782,
+      "loss": 1.5938,
+      "step": 5096
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011982156287343312,
+      "loss": 1.6562,
+      "step": 5097
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011979331974565756,
+      "loss": 1.6445,
+      "step": 5098
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011976507497454606,
+      "loss": 1.5918,
+      "step": 5099
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001197368285624436,
+      "loss": 1.4141,
+      "step": 5100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011970858051169534,
+      "loss": 1.5293,
+      "step": 5101
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011968033082464663,
+      "loss": 1.6836,
+      "step": 5102
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001196520795036428,
+      "loss": 1.6133,
+      "step": 5103
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011962382655102946,
+      "loss": 1.6094,
+      "step": 5104
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001195955719691523,
+      "loss": 1.5977,
+      "step": 5105
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011956731576035712,
+      "loss": 1.5859,
+      "step": 5106
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011953905792698994,
+      "loss": 1.7363,
+      "step": 5107
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011951079847139683,
+      "loss": 1.5664,
+      "step": 5108
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011948253739592401,
+      "loss": 1.6348,
+      "step": 5109
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011945427470291788,
+      "loss": 1.6094,
+      "step": 5110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001194260103947249,
+      "loss": 1.5762,
+      "step": 5111
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011939774447369178,
+      "loss": 1.6973,
+      "step": 5112
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001193694769421652,
+      "loss": 1.6465,
+      "step": 5113
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0011934120780249216,
+      "loss": 1.5684,
+      "step": 5114
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011931293705701963,
+      "loss": 1.5391,
+      "step": 5115
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011928466470809483,
+      "loss": 1.5938,
+      "step": 5116
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00119256390758065,
+      "loss": 1.6426,
+      "step": 5117
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011922811520927766,
+      "loss": 1.6758,
+      "step": 5118
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001191998380640803,
+      "loss": 1.6602,
+      "step": 5119
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001191715593248207,
+      "loss": 1.6543,
+      "step": 5120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011914327899384665,
+      "loss": 1.666,
+      "step": 5121
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011911499707350612,
+      "loss": 1.5605,
+      "step": 5122
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011908671356614723,
+      "loss": 1.6152,
+      "step": 5123
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001190584284741182,
+      "loss": 1.4746,
+      "step": 5124
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011903014179976742,
+      "loss": 1.7188,
+      "step": 5125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011900185354544331,
+      "loss": 1.5918,
+      "step": 5126
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011897356371349456,
+      "loss": 1.5332,
+      "step": 5127
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011894527230626995,
+      "loss": 1.6484,
+      "step": 5128
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011891697932611828,
+      "loss": 1.707,
+      "step": 5129
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011888868477538865,
+      "loss": 1.6133,
+      "step": 5130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011886038865643015,
+      "loss": 1.6113,
+      "step": 5131
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011883209097159212,
+      "loss": 1.707,
+      "step": 5132
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011880379172322388,
+      "loss": 1.5469,
+      "step": 5133
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011877549091367508,
+      "loss": 1.6836,
+      "step": 5134
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011874718854529532,
+      "loss": 1.5938,
+      "step": 5135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011871888462043438,
+      "loss": 1.5586,
+      "step": 5136
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011869057914144223,
+      "loss": 1.5938,
+      "step": 5137
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011866227211066896,
+      "loss": 1.6152,
+      "step": 5138
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011863396353046466,
+      "loss": 1.6777,
+      "step": 5139
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001186056534031797,
+      "loss": 1.5039,
+      "step": 5140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001185773417311645,
+      "loss": 1.6836,
+      "step": 5141
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011854902851676967,
+      "loss": 1.6152,
+      "step": 5142
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011852071376234586,
+      "loss": 1.6797,
+      "step": 5143
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011849239747024395,
+      "loss": 1.6172,
+      "step": 5144
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011846407964281485,
+      "loss": 1.623,
+      "step": 5145
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011843576028240962,
+      "loss": 1.752,
+      "step": 5146
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011840743939137957,
+      "loss": 1.6426,
+      "step": 5147
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011837911697207594,
+      "loss": 1.5781,
+      "step": 5148
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011835079302685019,
+      "loss": 1.6016,
+      "step": 5149
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011832246755805395,
+      "loss": 1.6426,
+      "step": 5150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011829414056803893,
+      "loss": 1.6875,
+      "step": 5151
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011826581205915698,
+      "loss": 1.5938,
+      "step": 5152
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011823748203376003,
+      "loss": 1.5918,
+      "step": 5153
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011820915049420022,
+      "loss": 1.6621,
+      "step": 5154
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011818081744282974,
+      "loss": 1.6367,
+      "step": 5155
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011815248288200094,
+      "loss": 1.6055,
+      "step": 5156
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011812414681406631,
+      "loss": 1.5371,
+      "step": 5157
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001180958092413784,
+      "loss": 1.6992,
+      "step": 5158
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011806747016629001,
+      "loss": 1.7402,
+      "step": 5159
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001180391295911539,
+      "loss": 1.6152,
+      "step": 5160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011801078751832311,
+      "loss": 1.6191,
+      "step": 5161
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011798244395015067,
+      "loss": 1.502,
+      "step": 5162
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011795409888898985,
+      "loss": 1.6523,
+      "step": 5163
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011792575233719397,
+      "loss": 1.6914,
+      "step": 5164
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011789740429711648,
+      "loss": 1.5293,
+      "step": 5165
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00117869054771111,
+      "loss": 1.498,
+      "step": 5166
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011784070376153126,
+      "loss": 1.7344,
+      "step": 5167
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011781235127073103,
+      "loss": 1.6523,
+      "step": 5168
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011778399730106434,
+      "loss": 1.5312,
+      "step": 5169
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011775564185488516,
+      "loss": 1.627,
+      "step": 5170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011772728493454785,
+      "loss": 1.5957,
+      "step": 5171
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011769892654240664,
+      "loss": 1.5801,
+      "step": 5172
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00117670566680816,
+      "loss": 1.5195,
+      "step": 5173
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001176422053521305,
+      "loss": 1.6094,
+      "step": 5174
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011761384255870482,
+      "loss": 1.6172,
+      "step": 5175
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001175854783028938,
+      "loss": 1.5703,
+      "step": 5176
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011755711258705235,
+      "loss": 1.6836,
+      "step": 5177
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011752874541353555,
+      "loss": 1.5742,
+      "step": 5178
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011750037678469858,
+      "loss": 1.6562,
+      "step": 5179
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001174720067028967,
+      "loss": 1.6973,
+      "step": 5180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011744363517048539,
+      "loss": 1.5605,
+      "step": 5181
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011741526218982013,
+      "loss": 1.7305,
+      "step": 5182
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011738688776325662,
+      "loss": 1.6406,
+      "step": 5183
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011735851189315068,
+      "loss": 1.6875,
+      "step": 5184
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011733013458185809,
+      "loss": 1.543,
+      "step": 5185
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011730175583173498,
+      "loss": 1.6113,
+      "step": 5186
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011727337564513744,
+      "loss": 1.4902,
+      "step": 5187
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011724499402442172,
+      "loss": 1.6035,
+      "step": 5188
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011721661097194422,
+      "loss": 1.6641,
+      "step": 5189
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011718822649006144,
+      "loss": 1.5762,
+      "step": 5190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011715984058113,
+      "loss": 1.625,
+      "step": 5191
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011713145324750661,
+      "loss": 1.627,
+      "step": 5192
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011710306449154812,
+      "loss": 1.6309,
+      "step": 5193
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011707467431561153,
+      "loss": 1.5859,
+      "step": 5194
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001170462827220539,
+      "loss": 1.5996,
+      "step": 5195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011701788971323244,
+      "loss": 1.5527,
+      "step": 5196
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001169894952915045,
+      "loss": 1.6719,
+      "step": 5197
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011696109945922748,
+      "loss": 1.6992,
+      "step": 5198
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011693270221875898,
+      "loss": 1.6426,
+      "step": 5199
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011690430357245664,
+      "loss": 1.623,
+      "step": 5200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011687590352267828,
+      "loss": 1.5371,
+      "step": 5201
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011684750207178176,
+      "loss": 1.5293,
+      "step": 5202
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011681909922212515,
+      "loss": 1.5273,
+      "step": 5203
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001167906949760666,
+      "loss": 1.709,
+      "step": 5204
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001167622893359643,
+      "loss": 1.6543,
+      "step": 5205
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001167338823041767,
+      "loss": 1.6914,
+      "step": 5206
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011670547388306226,
+      "loss": 1.6406,
+      "step": 5207
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011667706407497956,
+      "loss": 1.5605,
+      "step": 5208
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011664865288228733,
+      "loss": 1.6426,
+      "step": 5209
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011662024030734438,
+      "loss": 1.6094,
+      "step": 5210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011659182635250973,
+      "loss": 1.6113,
+      "step": 5211
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011656341102014235,
+      "loss": 1.5625,
+      "step": 5212
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001165349943126015,
+      "loss": 1.6328,
+      "step": 5213
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011650657623224641,
+      "loss": 1.543,
+      "step": 5214
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011647815678143651,
+      "loss": 1.5586,
+      "step": 5215
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011644973596253133,
+      "loss": 1.5723,
+      "step": 5216
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001164213137778905,
+      "loss": 1.7363,
+      "step": 5217
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001163928902298737,
+      "loss": 1.5996,
+      "step": 5218
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001163644653208409,
+      "loss": 1.5859,
+      "step": 5219
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011633603905315197,
+      "loss": 1.707,
+      "step": 5220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011630761142916708,
+      "loss": 1.5957,
+      "step": 5221
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011627918245124636,
+      "loss": 1.6348,
+      "step": 5222
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011625075212175016,
+      "loss": 1.6152,
+      "step": 5223
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011622232044303886,
+      "loss": 1.5488,
+      "step": 5224
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011619388741747306,
+      "loss": 1.6426,
+      "step": 5225
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011616545304741336,
+      "loss": 1.7227,
+      "step": 5226
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0011613701733522053,
+      "loss": 1.6504,
+      "step": 5227
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011610858028325542,
+      "loss": 1.6641,
+      "step": 5228
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00116080141893879,
+      "loss": 1.6953,
+      "step": 5229
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001160517021694524,
+      "loss": 1.5859,
+      "step": 5230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011602326111233682,
+      "loss": 1.5273,
+      "step": 5231
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011599481872489354,
+      "loss": 1.6992,
+      "step": 5232
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011596637500948404,
+      "loss": 1.6113,
+      "step": 5233
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011593792996846977,
+      "loss": 1.5664,
+      "step": 5234
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011590948360421243,
+      "loss": 1.6133,
+      "step": 5235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011588103591907376,
+      "loss": 1.5723,
+      "step": 5236
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011585258691541563,
+      "loss": 1.4238,
+      "step": 5237
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011582413659559998,
+      "loss": 1.6152,
+      "step": 5238
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011579568496198895,
+      "loss": 1.5723,
+      "step": 5239
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011576723201694467,
+      "loss": 1.5723,
+      "step": 5240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011573877776282946,
+      "loss": 1.5625,
+      "step": 5241
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011571032220200578,
+      "loss": 1.6914,
+      "step": 5242
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011568186533683607,
+      "loss": 1.6367,
+      "step": 5243
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011565340716968298,
+      "loss": 1.5898,
+      "step": 5244
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011562494770290927,
+      "loss": 1.6621,
+      "step": 5245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011559648693887778,
+      "loss": 1.6582,
+      "step": 5246
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011556802487995136,
+      "loss": 1.7109,
+      "step": 5247
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011553956152849324,
+      "loss": 1.5391,
+      "step": 5248
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001155110968868664,
+      "loss": 1.5762,
+      "step": 5249
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011548263095743428,
+      "loss": 1.625,
+      "step": 5250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011545416374256013,
+      "loss": 1.627,
+      "step": 5251
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001154256952446075,
+      "loss": 1.5664,
+      "step": 5252
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001153972254659399,
+      "loss": 1.5996,
+      "step": 5253
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011536875440892116,
+      "loss": 1.6348,
+      "step": 5254
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011534028207591499,
+      "loss": 1.7148,
+      "step": 5255
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011531180846928533,
+      "loss": 1.6934,
+      "step": 5256
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011528333359139616,
+      "loss": 1.6465,
+      "step": 5257
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011525485744461163,
+      "loss": 1.6348,
+      "step": 5258
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011522638003129595,
+      "loss": 1.6582,
+      "step": 5259
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001151979013538135,
+      "loss": 1.6992,
+      "step": 5260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011516942141452863,
+      "loss": 1.541,
+      "step": 5261
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011514094021580598,
+      "loss": 1.5703,
+      "step": 5262
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011511245776001013,
+      "loss": 1.6074,
+      "step": 5263
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011508397404950584,
+      "loss": 1.6309,
+      "step": 5264
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011505548908665798,
+      "loss": 1.707,
+      "step": 5265
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011502700287383148,
+      "loss": 1.4805,
+      "step": 5266
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011499851541339138,
+      "loss": 1.5996,
+      "step": 5267
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011497002670770294,
+      "loss": 1.4512,
+      "step": 5268
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011494153675913133,
+      "loss": 1.6133,
+      "step": 5269
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011491304557004198,
+      "loss": 1.627,
+      "step": 5270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011488455314280033,
+      "loss": 1.6016,
+      "step": 5271
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00114856059479772,
+      "loss": 1.5312,
+      "step": 5272
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011482756458332261,
+      "loss": 1.6172,
+      "step": 5273
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011479906845581803,
+      "loss": 1.5273,
+      "step": 5274
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011477057109962403,
+      "loss": 1.582,
+      "step": 5275
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001147420725171067,
+      "loss": 1.5312,
+      "step": 5276
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011471357271063205,
+      "loss": 1.584,
+      "step": 5277
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011468507168256633,
+      "loss": 1.6641,
+      "step": 5278
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011465656943527578,
+      "loss": 1.666,
+      "step": 5279
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011462806597112684,
+      "loss": 1.6484,
+      "step": 5280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011459956129248599,
+      "loss": 1.6543,
+      "step": 5281
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001145710554017198,
+      "loss": 1.627,
+      "step": 5282
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011454254830119501,
+      "loss": 1.6719,
+      "step": 5283
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011451403999327837,
+      "loss": 1.5898,
+      "step": 5284
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001144855304803368,
+      "loss": 1.6055,
+      "step": 5285
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001144570197647373,
+      "loss": 1.5039,
+      "step": 5286
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011442850784884695,
+      "loss": 1.5938,
+      "step": 5287
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011439999473503296,
+      "loss": 1.5996,
+      "step": 5288
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011437148042566258,
+      "loss": 1.6328,
+      "step": 5289
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011434296492310327,
+      "loss": 1.6152,
+      "step": 5290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001143144482297225,
+      "loss": 1.6152,
+      "step": 5291
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011428593034788785,
+      "loss": 1.6113,
+      "step": 5292
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011425741127996701,
+      "loss": 1.6016,
+      "step": 5293
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011422889102832778,
+      "loss": 1.6289,
+      "step": 5294
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011420036959533804,
+      "loss": 1.5957,
+      "step": 5295
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001141718469833658,
+      "loss": 1.498,
+      "step": 5296
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001141433231947791,
+      "loss": 1.4707,
+      "step": 5297
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011411479823194618,
+      "loss": 1.5391,
+      "step": 5298
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011408627209723524,
+      "loss": 1.6621,
+      "step": 5299
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011405774479301472,
+      "loss": 1.5234,
+      "step": 5300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011402921632165306,
+      "loss": 1.6191,
+      "step": 5301
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011400068668551886,
+      "loss": 1.5469,
+      "step": 5302
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011397215588698076,
+      "loss": 1.5176,
+      "step": 5303
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011394362392840754,
+      "loss": 1.4805,
+      "step": 5304
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011391509081216805,
+      "loss": 1.6465,
+      "step": 5305
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011388655654063125,
+      "loss": 1.5879,
+      "step": 5306
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011385802111616617,
+      "loss": 1.7031,
+      "step": 5307
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011382948454114199,
+      "loss": 1.5293,
+      "step": 5308
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011380094681792791,
+      "loss": 1.7344,
+      "step": 5309
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011377240794889333,
+      "loss": 1.7246,
+      "step": 5310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011374386793640764,
+      "loss": 1.7246,
+      "step": 5311
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001137153267828404,
+      "loss": 1.5703,
+      "step": 5312
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011368678449056118,
+      "loss": 1.7363,
+      "step": 5313
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011365824106193973,
+      "loss": 1.5645,
+      "step": 5314
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011362969649934586,
+      "loss": 1.5586,
+      "step": 5315
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011360115080514951,
+      "loss": 1.5195,
+      "step": 5316
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001135726039817206,
+      "loss": 1.5625,
+      "step": 5317
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011354405603142932,
+      "loss": 1.6855,
+      "step": 5318
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011351550695664577,
+      "loss": 1.6191,
+      "step": 5319
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011348695675974028,
+      "loss": 1.5508,
+      "step": 5320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011345840544308321,
+      "loss": 1.5332,
+      "step": 5321
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011342985300904502,
+      "loss": 1.6914,
+      "step": 5322
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001134012994599963,
+      "loss": 1.5957,
+      "step": 5323
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011337274479830766,
+      "loss": 1.6621,
+      "step": 5324
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011334418902634988,
+      "loss": 1.6738,
+      "step": 5325
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011331563214649375,
+      "loss": 1.6055,
+      "step": 5326
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011328707416111026,
+      "loss": 1.6191,
+      "step": 5327
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011325851507257038,
+      "loss": 1.5859,
+      "step": 5328
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011322995488324526,
+      "loss": 1.5566,
+      "step": 5329
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011320139359550607,
+      "loss": 1.6602,
+      "step": 5330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011317283121172414,
+      "loss": 1.623,
+      "step": 5331
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001131442677342708,
+      "loss": 1.6172,
+      "step": 5332
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011311570316551761,
+      "loss": 1.6797,
+      "step": 5333
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011308713750783604,
+      "loss": 1.5918,
+      "step": 5334
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011305857076359782,
+      "loss": 1.666,
+      "step": 5335
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011303000293517466,
+      "loss": 1.6777,
+      "step": 5336
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011300143402493844,
+      "loss": 1.5996,
+      "step": 5337
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011297286403526101,
+      "loss": 1.6602,
+      "step": 5338
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0011294429296851449,
+      "loss": 1.6562,
+      "step": 5339
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011291572082707088,
+      "loss": 1.6074,
+      "step": 5340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011288714761330248,
+      "loss": 1.5723,
+      "step": 5341
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001128585733295815,
+      "loss": 1.5762,
+      "step": 5342
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001128299979782803,
+      "loss": 1.6016,
+      "step": 5343
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011280142156177145,
+      "loss": 1.6914,
+      "step": 5344
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001127728440824274,
+      "loss": 1.5918,
+      "step": 5345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011274426554262079,
+      "loss": 1.6094,
+      "step": 5346
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011271568594472439,
+      "loss": 1.5273,
+      "step": 5347
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00112687105291111,
+      "loss": 1.5859,
+      "step": 5348
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011265852358415351,
+      "loss": 1.5879,
+      "step": 5349
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011262994082622494,
+      "loss": 1.6719,
+      "step": 5350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011260135701969833,
+      "loss": 1.6914,
+      "step": 5351
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011257277216694685,
+      "loss": 1.6289,
+      "step": 5352
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011254418627034382,
+      "loss": 1.5996,
+      "step": 5353
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011251559933226248,
+      "loss": 1.5469,
+      "step": 5354
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001124870113550763,
+      "loss": 1.541,
+      "step": 5355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011245842234115877,
+      "loss": 1.7344,
+      "step": 5356
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011242983229288353,
+      "loss": 1.6309,
+      "step": 5357
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011240124121262422,
+      "loss": 1.5547,
+      "step": 5358
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011237264910275463,
+      "loss": 1.5938,
+      "step": 5359
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011234405596564862,
+      "loss": 1.6738,
+      "step": 5360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001123154618036801,
+      "loss": 1.6289,
+      "step": 5361
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001122868666192231,
+      "loss": 1.6367,
+      "step": 5362
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011225827041465176,
+      "loss": 1.5332,
+      "step": 5363
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011222967319234024,
+      "loss": 1.7266,
+      "step": 5364
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011220107495466282,
+      "loss": 1.7168,
+      "step": 5365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011217247570399387,
+      "loss": 1.7109,
+      "step": 5366
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011214387544270785,
+      "loss": 1.5195,
+      "step": 5367
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011211527417317928,
+      "loss": 1.6973,
+      "step": 5368
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011208667189778277,
+      "loss": 1.5586,
+      "step": 5369
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00112058068618893,
+      "loss": 1.6934,
+      "step": 5370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001120294643388848,
+      "loss": 1.5645,
+      "step": 5371
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00112000859060133,
+      "loss": 1.6016,
+      "step": 5372
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011197225278501254,
+      "loss": 1.6875,
+      "step": 5373
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011194364551589846,
+      "loss": 1.6992,
+      "step": 5374
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001119150372551659,
+      "loss": 1.5352,
+      "step": 5375
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011188642800518999,
+      "loss": 1.625,
+      "step": 5376
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011185781776834607,
+      "loss": 1.5703,
+      "step": 5377
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011182920654700945,
+      "loss": 1.6289,
+      "step": 5378
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011180059434355563,
+      "loss": 1.5312,
+      "step": 5379
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011177198116036004,
+      "loss": 1.6641,
+      "step": 5380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011174336699979839,
+      "loss": 1.6973,
+      "step": 5381
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001117147518642463,
+      "loss": 1.6055,
+      "step": 5382
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011168613575607955,
+      "loss": 1.6562,
+      "step": 5383
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011165751867767398,
+      "loss": 1.7246,
+      "step": 5384
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011162890063140553,
+      "loss": 1.6758,
+      "step": 5385
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011160028161965017,
+      "loss": 1.625,
+      "step": 5386
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011157166164478406,
+      "loss": 1.6094,
+      "step": 5387
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011154304070918326,
+      "loss": 1.6309,
+      "step": 5388
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011151441881522412,
+      "loss": 1.7559,
+      "step": 5389
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001114857959652829,
+      "loss": 1.5918,
+      "step": 5390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011145717216173604,
+      "loss": 1.625,
+      "step": 5391
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011142854740696002,
+      "loss": 1.7793,
+      "step": 5392
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011139992170333141,
+      "loss": 1.5703,
+      "step": 5393
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011137129505322683,
+      "loss": 1.7246,
+      "step": 5394
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00111342667459023,
+      "loss": 1.5938,
+      "step": 5395
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011131403892309674,
+      "loss": 1.7109,
+      "step": 5396
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011128540944782496,
+      "loss": 1.5918,
+      "step": 5397
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001112567790355845,
+      "loss": 1.6445,
+      "step": 5398
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011122814768875252,
+      "loss": 1.5996,
+      "step": 5399
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011119951540970608,
+      "loss": 1.4609,
+      "step": 5400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001111708822008224,
+      "loss": 1.6621,
+      "step": 5401
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011114224806447863,
+      "loss": 1.6699,
+      "step": 5402
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011111361300305228,
+      "loss": 1.6465,
+      "step": 5403
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011108497701892066,
+      "loss": 1.6816,
+      "step": 5404
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011105634011446131,
+      "loss": 1.5312,
+      "step": 5405
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011102770229205175,
+      "loss": 1.6094,
+      "step": 5406
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011099906355406968,
+      "loss": 1.6738,
+      "step": 5407
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001109704239028928,
+      "loss": 1.582,
+      "step": 5408
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011094178334089895,
+      "loss": 1.7422,
+      "step": 5409
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011091314187046594,
+      "loss": 1.6445,
+      "step": 5410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011088449949397177,
+      "loss": 1.6328,
+      "step": 5411
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011085585621379449,
+      "loss": 1.5703,
+      "step": 5412
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011082721203231212,
+      "loss": 1.584,
+      "step": 5413
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011079856695190291,
+      "loss": 1.6035,
+      "step": 5414
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011076992097494506,
+      "loss": 1.5938,
+      "step": 5415
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011074127410381695,
+      "loss": 1.3574,
+      "step": 5416
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011071262634089697,
+      "loss": 1.4961,
+      "step": 5417
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011068397768856356,
+      "loss": 1.6992,
+      "step": 5418
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011065532814919525,
+      "loss": 1.5391,
+      "step": 5419
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011062667772517075,
+      "loss": 1.625,
+      "step": 5420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011059802641886867,
+      "loss": 1.6504,
+      "step": 5421
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001105693742326678,
+      "loss": 1.6328,
+      "step": 5422
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00110540721168947,
+      "loss": 1.5977,
+      "step": 5423
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001105120672300852,
+      "loss": 1.623,
+      "step": 5424
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011048341241846134,
+      "loss": 1.7266,
+      "step": 5425
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011045475673645449,
+      "loss": 1.6367,
+      "step": 5426
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001104261001864438,
+      "loss": 1.5859,
+      "step": 5427
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011039744277080848,
+      "loss": 1.7305,
+      "step": 5428
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011036878449192775,
+      "loss": 1.6816,
+      "step": 5429
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011034012535218105,
+      "loss": 1.584,
+      "step": 5430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011031146535394772,
+      "loss": 1.4238,
+      "step": 5431
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001102828044996073,
+      "loss": 1.6133,
+      "step": 5432
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011025414279153929,
+      "loss": 1.6484,
+      "step": 5433
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001102254802321234,
+      "loss": 1.6094,
+      "step": 5434
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011019681682373928,
+      "loss": 1.6152,
+      "step": 5435
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011016815256876672,
+      "loss": 1.6582,
+      "step": 5436
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011013948746958554,
+      "loss": 1.6094,
+      "step": 5437
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011011082152857568,
+      "loss": 1.6152,
+      "step": 5438
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011008215474811712,
+      "loss": 1.5898,
+      "step": 5439
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011005348713058992,
+      "loss": 1.5547,
+      "step": 5440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0011002481867837418,
+      "loss": 1.6289,
+      "step": 5441
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0010999614939385012,
+      "loss": 1.5625,
+      "step": 5442
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0010996747927939797,
+      "loss": 1.6973,
+      "step": 5443
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0010993880833739809,
+      "loss": 1.5996,
+      "step": 5444
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0010991013657023084,
+      "loss": 1.5664,
+      "step": 5445
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0010988146398027674,
+      "loss": 1.6074,
+      "step": 5446
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0010985279056991626,
+      "loss": 1.6406,
+      "step": 5447
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0010982411634153006,
+      "loss": 1.6465,
+      "step": 5448
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001097954412974988,
+      "loss": 1.6289,
+      "step": 5449
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0010976676544020318,
+      "loss": 1.6055,
+      "step": 5450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0010973808877202404,
+      "loss": 1.6562,
+      "step": 5451
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001097094112953423,
+      "loss": 1.6152,
+      "step": 5452
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010968073301253877,
+      "loss": 1.6113,
+      "step": 5453
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001096520539259946,
+      "loss": 1.4766,
+      "step": 5454
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001096233740380908,
+      "loss": 1.6348,
+      "step": 5455
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010959469335120851,
+      "loss": 1.6387,
+      "step": 5456
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010956601186772892,
+      "loss": 1.6426,
+      "step": 5457
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010953732959003335,
+      "loss": 1.5703,
+      "step": 5458
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010950864652050313,
+      "loss": 1.6777,
+      "step": 5459
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010947996266151962,
+      "loss": 1.6934,
+      "step": 5460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010945127801546433,
+      "loss": 1.7188,
+      "step": 5461
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001094225925847188,
+      "loss": 1.6543,
+      "step": 5462
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010939390637166465,
+      "loss": 1.502,
+      "step": 5463
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001093652193786835,
+      "loss": 1.5664,
+      "step": 5464
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010933653160815706,
+      "loss": 1.6816,
+      "step": 5465
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001093078430624672,
+      "loss": 1.6582,
+      "step": 5466
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001092791537439957,
+      "loss": 1.7266,
+      "step": 5467
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001092504636551246,
+      "loss": 1.6523,
+      "step": 5468
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010922177279823575,
+      "loss": 1.5215,
+      "step": 5469
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010919308117571132,
+      "loss": 1.4824,
+      "step": 5470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010916438878993333,
+      "loss": 1.6738,
+      "step": 5471
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010913569564328404,
+      "loss": 1.5293,
+      "step": 5472
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010910700173814565,
+      "loss": 1.5723,
+      "step": 5473
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010907830707690047,
+      "loss": 1.7012,
+      "step": 5474
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010904961166193087,
+      "loss": 1.6523,
+      "step": 5475
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010902091549561926,
+      "loss": 1.6094,
+      "step": 5476
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010899221858034817,
+      "loss": 1.623,
+      "step": 5477
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010896352091850012,
+      "loss": 1.6504,
+      "step": 5478
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010893482251245774,
+      "loss": 1.6055,
+      "step": 5479
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010890612336460375,
+      "loss": 1.6523,
+      "step": 5480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001088774234773208,
+      "loss": 1.6172,
+      "step": 5481
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010884872285299178,
+      "loss": 1.7168,
+      "step": 5482
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010882002149399954,
+      "loss": 1.7051,
+      "step": 5483
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010879131940272695,
+      "loss": 1.668,
+      "step": 5484
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010876261658155704,
+      "loss": 1.582,
+      "step": 5485
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010873391303287284,
+      "loss": 1.5742,
+      "step": 5486
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010870520875905744,
+      "loss": 1.7891,
+      "step": 5487
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010867650376249402,
+      "loss": 1.6816,
+      "step": 5488
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010864779804556585,
+      "loss": 1.5508,
+      "step": 5489
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010861909161065614,
+      "loss": 1.5859,
+      "step": 5490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010859038446014825,
+      "loss": 1.5176,
+      "step": 5491
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010856167659642566,
+      "loss": 1.5723,
+      "step": 5492
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010853296802187174,
+      "loss": 1.7012,
+      "step": 5493
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010850425873887004,
+      "loss": 1.6309,
+      "step": 5494
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001084755487498042,
+      "loss": 1.6406,
+      "step": 5495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010844683805705778,
+      "loss": 1.8203,
+      "step": 5496
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010841812666301453,
+      "loss": 1.6895,
+      "step": 5497
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010838941457005818,
+      "loss": 1.5547,
+      "step": 5498
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010836070178057257,
+      "loss": 1.7109,
+      "step": 5499
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010833198829694154,
+      "loss": 1.5684,
+      "step": 5500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010830327412154905,
+      "loss": 1.6406,
+      "step": 5501
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001082745592567791,
+      "loss": 1.5664,
+      "step": 5502
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010824584370501568,
+      "loss": 1.5703,
+      "step": 5503
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010821712746864295,
+      "loss": 1.5625,
+      "step": 5504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010818841055004501,
+      "loss": 1.4453,
+      "step": 5505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010815969295160613,
+      "loss": 1.5762,
+      "step": 5506
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010813097467571055,
+      "loss": 1.5762,
+      "step": 5507
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010810225572474262,
+      "loss": 1.5977,
+      "step": 5508
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010807353610108672,
+      "loss": 1.7402,
+      "step": 5509
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010804481580712726,
+      "loss": 1.3809,
+      "step": 5510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001080160948452488,
+      "loss": 1.6875,
+      "step": 5511
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001079873732178358,
+      "loss": 1.5586,
+      "step": 5512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010795865092727297,
+      "loss": 1.6113,
+      "step": 5513
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001079299279759449,
+      "loss": 1.7539,
+      "step": 5514
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001079012043662363,
+      "loss": 1.7754,
+      "step": 5515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00107872480100532,
+      "loss": 1.6094,
+      "step": 5516
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001078437551812168,
+      "loss": 1.5801,
+      "step": 5517
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010781502961067558,
+      "loss": 1.5859,
+      "step": 5518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010778630339129324,
+      "loss": 1.6348,
+      "step": 5519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010775757652545482,
+      "loss": 1.5566,
+      "step": 5520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010772884901554534,
+      "loss": 1.5977,
+      "step": 5521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010770012086394989,
+      "loss": 1.7383,
+      "step": 5522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001076713920730536,
+      "loss": 1.6602,
+      "step": 5523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010764266264524172,
+      "loss": 1.5684,
+      "step": 5524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010761393258289947,
+      "loss": 1.6426,
+      "step": 5525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010758520188841215,
+      "loss": 1.5684,
+      "step": 5526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010755647056416514,
+      "loss": 1.6602,
+      "step": 5527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010752773861254386,
+      "loss": 1.5391,
+      "step": 5528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010749900603593374,
+      "loss": 1.5078,
+      "step": 5529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010747027283672032,
+      "loss": 1.627,
+      "step": 5530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010744153901728918,
+      "loss": 1.6543,
+      "step": 5531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001074128045800259,
+      "loss": 1.5254,
+      "step": 5532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001073840695273162,
+      "loss": 1.541,
+      "step": 5533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010735533386154574,
+      "loss": 1.4512,
+      "step": 5534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010732659758510034,
+      "loss": 1.6387,
+      "step": 5535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001072978607003658,
+      "loss": 1.7871,
+      "step": 5536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010726912320972804,
+      "loss": 1.707,
+      "step": 5537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001072403851155729,
+      "loss": 1.6172,
+      "step": 5538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010721164642028642,
+      "loss": 1.5703,
+      "step": 5539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001071829071262546,
+      "loss": 1.7637,
+      "step": 5540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010715416723586352,
+      "loss": 1.6953,
+      "step": 5541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010712542675149932,
+      "loss": 1.6406,
+      "step": 5542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010709668567554812,
+      "loss": 1.7109,
+      "step": 5543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001070679440103962,
+      "loss": 1.6621,
+      "step": 5544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010703920175842978,
+      "loss": 1.623,
+      "step": 5545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001070104589220352,
+      "loss": 1.5918,
+      "step": 5546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010698171550359886,
+      "loss": 1.6309,
+      "step": 5547
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010695297150550712,
+      "loss": 1.6191,
+      "step": 5548
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001069242269301465,
+      "loss": 1.6133,
+      "step": 5549
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010689548177990344,
+      "loss": 1.6504,
+      "step": 5550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010686673605716458,
+      "loss": 1.6875,
+      "step": 5551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010683798976431646,
+      "loss": 1.5684,
+      "step": 5552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010680924290374578,
+      "loss": 1.502,
+      "step": 5553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001067804954778392,
+      "loss": 1.6445,
+      "step": 5554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010675174748898356,
+      "loss": 1.6465,
+      "step": 5555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010672299893956555,
+      "loss": 1.5938,
+      "step": 5556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010669424983197205,
+      "loss": 1.7344,
+      "step": 5557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010666550016858994,
+      "loss": 1.4961,
+      "step": 5558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010663674995180616,
+      "loss": 1.6758,
+      "step": 5559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010660799918400772,
+      "loss": 1.5879,
+      "step": 5560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001065792478675816,
+      "loss": 1.6387,
+      "step": 5561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010655049600491492,
+      "loss": 1.5176,
+      "step": 5562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0010652174359839475,
+      "loss": 1.668,
+      "step": 5563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001064929906504083,
+      "loss": 1.4551,
+      "step": 5564
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010646423716334273,
+      "loss": 1.5293,
+      "step": 5565
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001064354831395853,
+      "loss": 1.6641,
+      "step": 5566
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010640672858152332,
+      "loss": 1.627,
+      "step": 5567
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001063779734915441,
+      "loss": 1.5508,
+      "step": 5568
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010634921787203509,
+      "loss": 1.5469,
+      "step": 5569
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010632046172538367,
+      "loss": 1.7051,
+      "step": 5570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010629170505397733,
+      "loss": 1.541,
+      "step": 5571
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010626294786020357,
+      "loss": 1.6738,
+      "step": 5572
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010623419014644997,
+      "loss": 1.7637,
+      "step": 5573
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010620543191510413,
+      "loss": 1.5762,
+      "step": 5574
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010617667316855366,
+      "loss": 1.6211,
+      "step": 5575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001061479139091863,
+      "loss": 1.5723,
+      "step": 5576
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010611915413938976,
+      "loss": 1.5469,
+      "step": 5577
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001060903938615518,
+      "loss": 1.7207,
+      "step": 5578
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010606163307806022,
+      "loss": 1.6367,
+      "step": 5579
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010603287179130297,
+      "loss": 1.5156,
+      "step": 5580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010600411000366784,
+      "loss": 1.5898,
+      "step": 5581
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010597534771754284,
+      "loss": 1.6758,
+      "step": 5582
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001059465849353159,
+      "loss": 1.5762,
+      "step": 5583
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010591782165937512,
+      "loss": 1.6816,
+      "step": 5584
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010588905789210845,
+      "loss": 1.541,
+      "step": 5585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001058602936359041,
+      "loss": 1.627,
+      "step": 5586
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010583152889315018,
+      "loss": 1.543,
+      "step": 5587
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010580276366623487,
+      "loss": 1.627,
+      "step": 5588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010577399795754637,
+      "loss": 1.6094,
+      "step": 5589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00105745231769473,
+      "loss": 1.4824,
+      "step": 5590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010571646510440305,
+      "loss": 1.6113,
+      "step": 5591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010568769796472485,
+      "loss": 1.5742,
+      "step": 5592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001056589303528268,
+      "loss": 1.6523,
+      "step": 5593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010563016227109732,
+      "loss": 1.4883,
+      "step": 5594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010560139372192486,
+      "loss": 1.6758,
+      "step": 5595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010557262470769794,
+      "loss": 1.7441,
+      "step": 5596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001055438552308051,
+      "loss": 1.5664,
+      "step": 5597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001055150852936349,
+      "loss": 1.543,
+      "step": 5598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00105486314898576,
+      "loss": 1.623,
+      "step": 5599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010545754404801702,
+      "loss": 1.5195,
+      "step": 5600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001054287727443467,
+      "loss": 1.6445,
+      "step": 5601
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010540000098995367,
+      "loss": 1.6504,
+      "step": 5602
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010537122878722677,
+      "loss": 1.498,
+      "step": 5603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010534245613855481,
+      "loss": 1.543,
+      "step": 5604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010531368304632662,
+      "loss": 1.668,
+      "step": 5605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010528490951293111,
+      "loss": 1.5977,
+      "step": 5606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001052561355407571,
+      "loss": 1.5605,
+      "step": 5607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010522736113219367,
+      "loss": 1.459,
+      "step": 5608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001051985862896297,
+      "loss": 1.6016,
+      "step": 5609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001051698110154543,
+      "loss": 1.5664,
+      "step": 5610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010514103531205645,
+      "loss": 1.6289,
+      "step": 5611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010511225918182535,
+      "loss": 1.6621,
+      "step": 5612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010508348262715003,
+      "loss": 1.7344,
+      "step": 5613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001050547056504197,
+      "loss": 1.5488,
+      "step": 5614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010502592825402356,
+      "loss": 1.6816,
+      "step": 5615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010499715044035086,
+      "loss": 1.6191,
+      "step": 5616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010496837221179083,
+      "loss": 1.6465,
+      "step": 5617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010493959357073283,
+      "loss": 1.6074,
+      "step": 5618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010491081451956617,
+      "loss": 1.6406,
+      "step": 5619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010488203506068023,
+      "loss": 1.6738,
+      "step": 5620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010485325519646442,
+      "loss": 1.5879,
+      "step": 5621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010482447492930818,
+      "loss": 1.666,
+      "step": 5622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010479569426160098,
+      "loss": 1.5781,
+      "step": 5623
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010476691319573235,
+      "loss": 1.5176,
+      "step": 5624
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010473813173409179,
+      "loss": 1.5723,
+      "step": 5625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010470934987906891,
+      "loss": 1.625,
+      "step": 5626
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010468056763305329,
+      "loss": 1.6348,
+      "step": 5627
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010465178499843462,
+      "loss": 1.6484,
+      "step": 5628
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001046230019776025,
+      "loss": 1.7344,
+      "step": 5629
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001045942185729467,
+      "loss": 1.7227,
+      "step": 5630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001045654347868569,
+      "loss": 1.5195,
+      "step": 5631
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010453665062172292,
+      "loss": 1.5625,
+      "step": 5632
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001045078660799345,
+      "loss": 1.6328,
+      "step": 5633
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010447908116388155,
+      "loss": 1.7188,
+      "step": 5634
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010445029587595384,
+      "loss": 1.6465,
+      "step": 5635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010442151021854134,
+      "loss": 1.5234,
+      "step": 5636
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010439272419403391,
+      "loss": 1.582,
+      "step": 5637
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010436393780482156,
+      "loss": 1.6719,
+      "step": 5638
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010433515105329424,
+      "loss": 1.7012,
+      "step": 5639
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010430636394184196,
+      "loss": 1.6074,
+      "step": 5640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001042775764728548,
+      "loss": 1.5566,
+      "step": 5641
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010424878864872277,
+      "loss": 1.5527,
+      "step": 5642
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010422000047183603,
+      "loss": 1.582,
+      "step": 5643
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010419121194458472,
+      "loss": 1.623,
+      "step": 5644
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010416242306935893,
+      "loss": 1.6641,
+      "step": 5645
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010413363384854891,
+      "loss": 1.5879,
+      "step": 5646
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010410484428454486,
+      "loss": 1.6016,
+      "step": 5647
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010407605437973701,
+      "loss": 1.5605,
+      "step": 5648
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010404726413651567,
+      "loss": 1.6191,
+      "step": 5649
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010401847355727112,
+      "loss": 1.4219,
+      "step": 5650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001039896826443937,
+      "loss": 1.623,
+      "step": 5651
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010396089140027379,
+      "loss": 1.6953,
+      "step": 5652
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010393209982730172,
+      "loss": 1.6094,
+      "step": 5653
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010390330792786795,
+      "loss": 1.7324,
+      "step": 5654
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001038745157043629,
+      "loss": 1.5215,
+      "step": 5655
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001038457231591771,
+      "loss": 1.5547,
+      "step": 5656
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001038169302947009,
+      "loss": 1.6016,
+      "step": 5657
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00103788137113325,
+      "loss": 1.5645,
+      "step": 5658
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001037593436174398,
+      "loss": 1.6426,
+      "step": 5659
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00103730549809436,
+      "loss": 1.5801,
+      "step": 5660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010370175569170408,
+      "loss": 1.6738,
+      "step": 5661
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010367296126663472,
+      "loss": 1.4199,
+      "step": 5662
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010364416653661861,
+      "loss": 1.7695,
+      "step": 5663
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010361537150404634,
+      "loss": 1.6934,
+      "step": 5664
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010358657617130869,
+      "loss": 1.5352,
+      "step": 5665
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010355778054079632,
+      "loss": 1.6406,
+      "step": 5666
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010352898461490003,
+      "loss": 1.5039,
+      "step": 5667
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010350018839601059,
+      "loss": 1.4922,
+      "step": 5668
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010347139188651878,
+      "loss": 1.6387,
+      "step": 5669
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001034425950888154,
+      "loss": 1.5859,
+      "step": 5670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010341379800529141,
+      "loss": 1.5762,
+      "step": 5671
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010338500063833754,
+      "loss": 1.6582,
+      "step": 5672
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010335620299034476,
+      "loss": 1.5469,
+      "step": 5673
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010332740506370397,
+      "loss": 1.5742,
+      "step": 5674
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010329860686080611,
+      "loss": 1.6152,
+      "step": 5675
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0010326980838404217,
+      "loss": 1.5645,
+      "step": 5676
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010324100963580312,
+      "loss": 1.5625,
+      "step": 5677
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010321221061847994,
+      "loss": 1.7324,
+      "step": 5678
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010318341133446372,
+      "loss": 1.5801,
+      "step": 5679
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010315461178614547,
+      "loss": 1.5879,
+      "step": 5680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001031258119759163,
+      "loss": 1.5762,
+      "step": 5681
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010309701190616724,
+      "loss": 1.6328,
+      "step": 5682
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010306821157928952,
+      "loss": 1.5859,
+      "step": 5683
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010303941099767415,
+      "loss": 1.5918,
+      "step": 5684
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001030106101637124,
+      "loss": 1.4941,
+      "step": 5685
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001029818090797954,
+      "loss": 1.7637,
+      "step": 5686
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010295300774831437,
+      "loss": 1.6172,
+      "step": 5687
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010292420617166054,
+      "loss": 1.5488,
+      "step": 5688
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010289540435222512,
+      "loss": 1.6113,
+      "step": 5689
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010286660229239942,
+      "loss": 1.7227,
+      "step": 5690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010283779999457469,
+      "loss": 1.6523,
+      "step": 5691
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010280899746114224,
+      "loss": 1.7129,
+      "step": 5692
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010278019469449344,
+      "loss": 1.5898,
+      "step": 5693
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010275139169701956,
+      "loss": 1.5527,
+      "step": 5694
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010272258847111202,
+      "loss": 1.6328,
+      "step": 5695
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010269378501916217,
+      "loss": 1.5996,
+      "step": 5696
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010266498134356141,
+      "loss": 1.666,
+      "step": 5697
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010263617744670117,
+      "loss": 1.6738,
+      "step": 5698
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001026073733309729,
+      "loss": 1.6367,
+      "step": 5699
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010257856899876805,
+      "loss": 1.623,
+      "step": 5700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010254976445247805,
+      "loss": 1.5957,
+      "step": 5701
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010252095969449447,
+      "loss": 1.5684,
+      "step": 5702
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010249215472720875,
+      "loss": 1.5977,
+      "step": 5703
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010246334955301241,
+      "loss": 1.5996,
+      "step": 5704
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010243454417429704,
+      "loss": 1.7402,
+      "step": 5705
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010240573859345418,
+      "loss": 1.5234,
+      "step": 5706
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010237693281287543,
+      "loss": 1.5508,
+      "step": 5707
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010234812683495235,
+      "loss": 1.5918,
+      "step": 5708
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010231932066207656,
+      "loss": 1.6094,
+      "step": 5709
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001022905142966397,
+      "loss": 1.6191,
+      "step": 5710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001022617077410334,
+      "loss": 1.5391,
+      "step": 5711
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001022329009976493,
+      "loss": 1.5898,
+      "step": 5712
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001022040940688791,
+      "loss": 1.5527,
+      "step": 5713
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010217528695711448,
+      "loss": 1.4336,
+      "step": 5714
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010214647966474717,
+      "loss": 1.5254,
+      "step": 5715
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010211767219416882,
+      "loss": 1.8105,
+      "step": 5716
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010208886454777127,
+      "loss": 1.5645,
+      "step": 5717
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010206005672794617,
+      "loss": 1.5469,
+      "step": 5718
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001020312487370853,
+      "loss": 1.6562,
+      "step": 5719
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010200244057758051,
+      "loss": 1.541,
+      "step": 5720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010197363225182348,
+      "loss": 1.541,
+      "step": 5721
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001019448237622061,
+      "loss": 1.6699,
+      "step": 5722
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010191601511112015,
+      "loss": 1.543,
+      "step": 5723
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010188720630095749,
+      "loss": 1.6719,
+      "step": 5724
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010185839733410995,
+      "loss": 1.5391,
+      "step": 5725
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010182958821296937,
+      "loss": 1.5586,
+      "step": 5726
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010180077893992762,
+      "loss": 1.6973,
+      "step": 5727
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010177196951737663,
+      "loss": 1.7344,
+      "step": 5728
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010174315994770826,
+      "loss": 1.6445,
+      "step": 5729
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010171435023331441,
+      "loss": 1.4727,
+      "step": 5730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010168554037658705,
+      "loss": 1.709,
+      "step": 5731
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010165673037991804,
+      "loss": 1.5625,
+      "step": 5732
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010162792024569936,
+      "loss": 1.5996,
+      "step": 5733
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010159910997632295,
+      "loss": 1.7148,
+      "step": 5734
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010157029957418083,
+      "loss": 1.5605,
+      "step": 5735
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010154148904166491,
+      "loss": 1.5781,
+      "step": 5736
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010151267838116721,
+      "loss": 1.5957,
+      "step": 5737
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010148386759507974,
+      "loss": 1.5957,
+      "step": 5738
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010145505668579448,
+      "loss": 1.5957,
+      "step": 5739
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001014262456557035,
+      "loss": 1.6055,
+      "step": 5740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010139743450719874,
+      "loss": 1.5586,
+      "step": 5741
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010136862324267232,
+      "loss": 1.5996,
+      "step": 5742
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010133981186451627,
+      "loss": 1.5625,
+      "step": 5743
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010131100037512261,
+      "loss": 1.4961,
+      "step": 5744
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010128218877688347,
+      "loss": 1.5488,
+      "step": 5745
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010125337707219085,
+      "loss": 1.6055,
+      "step": 5746
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001012245652634369,
+      "loss": 1.6816,
+      "step": 5747
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010119575335301367,
+      "loss": 1.5664,
+      "step": 5748
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001011669413433133,
+      "loss": 1.6172,
+      "step": 5749
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010113812923672793,
+      "loss": 1.6816,
+      "step": 5750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010110931703564956,
+      "loss": 1.543,
+      "step": 5751
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010108050474247044,
+      "loss": 1.5547,
+      "step": 5752
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010105169235958263,
+      "loss": 1.7051,
+      "step": 5753
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010102287988937831,
+      "loss": 1.6641,
+      "step": 5754
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010099406733424959,
+      "loss": 1.6367,
+      "step": 5755
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010096525469658869,
+      "loss": 1.4766,
+      "step": 5756
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001009364419787877,
+      "loss": 1.5996,
+      "step": 5757
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010090762918323886,
+      "loss": 1.5508,
+      "step": 5758
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010087881631233428,
+      "loss": 1.666,
+      "step": 5759
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010085000336846617,
+      "loss": 1.623,
+      "step": 5760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010082119035402675,
+      "loss": 1.584,
+      "step": 5761
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010079237727140817,
+      "loss": 1.6875,
+      "step": 5762
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010076356412300263,
+      "loss": 1.7031,
+      "step": 5763
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010073475091120236,
+      "loss": 1.5742,
+      "step": 5764
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010070593763839954,
+      "loss": 1.6465,
+      "step": 5765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010067712430698644,
+      "loss": 1.6074,
+      "step": 5766
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001006483109193552,
+      "loss": 1.5508,
+      "step": 5767
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010061949747789815,
+      "loss": 1.5703,
+      "step": 5768
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001005906839850074,
+      "loss": 1.5254,
+      "step": 5769
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010056187044307527,
+      "loss": 1.5898,
+      "step": 5770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010053305685449397,
+      "loss": 1.4492,
+      "step": 5771
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010050424322165575,
+      "loss": 1.5508,
+      "step": 5772
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010047542954695282,
+      "loss": 1.582,
+      "step": 5773
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001004466158327775,
+      "loss": 1.6113,
+      "step": 5774
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010041780208152199,
+      "loss": 1.5684,
+      "step": 5775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010038898829557855,
+      "loss": 1.6816,
+      "step": 5776
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010036017447733942,
+      "loss": 1.5684,
+      "step": 5777
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010033136062919691,
+      "loss": 1.5742,
+      "step": 5778
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010030254675354327,
+      "loss": 1.7246,
+      "step": 5779
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010027373285277073,
+      "loss": 1.6738,
+      "step": 5780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001002449189292716,
+      "loss": 1.6172,
+      "step": 5781
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010021610498543812,
+      "loss": 1.6133,
+      "step": 5782
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010018729102366254,
+      "loss": 1.6406,
+      "step": 5783
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010015847704633722,
+      "loss": 1.7559,
+      "step": 5784
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010012966305585434,
+      "loss": 1.6133,
+      "step": 5785
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010010084905460624,
+      "loss": 1.5723,
+      "step": 5786
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010007203504498515,
+      "loss": 1.7207,
+      "step": 5787
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010004322102938337,
+      "loss": 1.6953,
+      "step": 5788
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0010001440701019318,
+      "loss": 1.5801,
+      "step": 5789
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009998559298980685,
+      "loss": 1.584,
+      "step": 5790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009995677897061666,
+      "loss": 1.666,
+      "step": 5791
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009992796495501485,
+      "loss": 1.7246,
+      "step": 5792
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009989915094539379,
+      "loss": 1.5898,
+      "step": 5793
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009987033694414567,
+      "loss": 1.752,
+      "step": 5794
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000998415229536628,
+      "loss": 1.5293,
+      "step": 5795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009981270897633744,
+      "loss": 1.6309,
+      "step": 5796
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009978389501456189,
+      "loss": 1.5625,
+      "step": 5797
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009975508107072843,
+      "loss": 1.4766,
+      "step": 5798
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000997262671472293,
+      "loss": 1.5312,
+      "step": 5799
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009969745324645675,
+      "loss": 1.5137,
+      "step": 5800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000996686393708031,
+      "loss": 1.6094,
+      "step": 5801
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009963982552266058,
+      "loss": 1.6289,
+      "step": 5802
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000996110117044215,
+      "loss": 1.5078,
+      "step": 5803
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009958219791847802,
+      "loss": 1.6895,
+      "step": 5804
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000995533841672225,
+      "loss": 1.582,
+      "step": 5805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009952457045304718,
+      "loss": 1.5566,
+      "step": 5806
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009949575677834427,
+      "loss": 1.6211,
+      "step": 5807
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009946694314550604,
+      "loss": 1.6758,
+      "step": 5808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009943812955692474,
+      "loss": 1.6289,
+      "step": 5809
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009940931601499262,
+      "loss": 1.6465,
+      "step": 5810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000993805025221019,
+      "loss": 1.6562,
+      "step": 5811
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000993516890806448,
+      "loss": 1.5879,
+      "step": 5812
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009932287569301358,
+      "loss": 1.5273,
+      "step": 5813
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009929406236160049,
+      "loss": 1.6875,
+      "step": 5814
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009926524908879769,
+      "loss": 1.6152,
+      "step": 5815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009923643587699737,
+      "loss": 1.4688,
+      "step": 5816
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009920762272859185,
+      "loss": 1.543,
+      "step": 5817
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009917880964597327,
+      "loss": 1.6836,
+      "step": 5818
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009914999663153384,
+      "loss": 1.5625,
+      "step": 5819
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009912118368766573,
+      "loss": 1.625,
+      "step": 5820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009909237081676117,
+      "loss": 1.6582,
+      "step": 5821
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009906355802121233,
+      "loss": 1.7148,
+      "step": 5822
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009903474530341134,
+      "loss": 1.6406,
+      "step": 5823
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000990059326657504,
+      "loss": 1.4941,
+      "step": 5824
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009897712011062171,
+      "loss": 1.6934,
+      "step": 5825
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000989483076404174,
+      "loss": 1.5527,
+      "step": 5826
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000989194952575296,
+      "loss": 1.5977,
+      "step": 5827
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009889068296435044,
+      "loss": 1.7461,
+      "step": 5828
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000988618707632721,
+      "loss": 1.5566,
+      "step": 5829
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000988330586566867,
+      "loss": 1.6504,
+      "step": 5830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009880424664698636,
+      "loss": 1.6172,
+      "step": 5831
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000987754347365631,
+      "loss": 1.6484,
+      "step": 5832
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009874662292780917,
+      "loss": 1.7188,
+      "step": 5833
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009871781122311658,
+      "loss": 1.5879,
+      "step": 5834
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009868899962487741,
+      "loss": 1.5547,
+      "step": 5835
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009866018813548373,
+      "loss": 1.5312,
+      "step": 5836
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009863137675732768,
+      "loss": 1.6133,
+      "step": 5837
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009860256549280128,
+      "loss": 1.5586,
+      "step": 5838
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000985737543442965,
+      "loss": 1.6113,
+      "step": 5839
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009854494331420552,
+      "loss": 1.7031,
+      "step": 5840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009851613240492029,
+      "loss": 1.666,
+      "step": 5841
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009848732161883281,
+      "loss": 1.582,
+      "step": 5842
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009845851095833507,
+      "loss": 1.5918,
+      "step": 5843
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009842970042581918,
+      "loss": 1.6484,
+      "step": 5844
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009840089002367705,
+      "loss": 1.6328,
+      "step": 5845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009837207975430069,
+      "loss": 1.5195,
+      "step": 5846
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009834326962008197,
+      "loss": 1.4902,
+      "step": 5847
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009831445962341298,
+      "loss": 1.5996,
+      "step": 5848
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009828564976668561,
+      "loss": 1.6992,
+      "step": 5849
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009825684005229176,
+      "loss": 1.6465,
+      "step": 5850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009822803048262336,
+      "loss": 1.6738,
+      "step": 5851
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009819922106007238,
+      "loss": 1.793,
+      "step": 5852
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009817041178703065,
+      "loss": 1.5508,
+      "step": 5853
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000981416026658901,
+      "loss": 1.623,
+      "step": 5854
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000981127936990425,
+      "loss": 1.5156,
+      "step": 5855
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009808398488887985,
+      "loss": 1.5449,
+      "step": 5856
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000980551762377939,
+      "loss": 1.459,
+      "step": 5857
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009802636774817654,
+      "loss": 1.543,
+      "step": 5858
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009799755942241951,
+      "loss": 1.6035,
+      "step": 5859
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000979687512629147,
+      "loss": 1.582,
+      "step": 5860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009793994327205386,
+      "loss": 1.6934,
+      "step": 5861
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009791113545222878,
+      "loss": 1.7031,
+      "step": 5862
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009788232780583116,
+      "loss": 1.5039,
+      "step": 5863
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009785352033525284,
+      "loss": 1.5918,
+      "step": 5864
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009782471304288552,
+      "loss": 1.6816,
+      "step": 5865
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009779590593112092,
+      "loss": 1.625,
+      "step": 5866
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000977670990023507,
+      "loss": 1.6113,
+      "step": 5867
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009773829225896663,
+      "loss": 1.5078,
+      "step": 5868
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009770948570336033,
+      "loss": 1.5898,
+      "step": 5869
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009768067933792347,
+      "loss": 1.7148,
+      "step": 5870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009765187316504764,
+      "loss": 1.5566,
+      "step": 5871
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009762306718712457,
+      "loss": 1.6523,
+      "step": 5872
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009759426140654582,
+      "loss": 1.6055,
+      "step": 5873
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009756545582570297,
+      "loss": 1.6738,
+      "step": 5874
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009753665044698759,
+      "loss": 1.541,
+      "step": 5875
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009750784527279128,
+      "loss": 1.6348,
+      "step": 5876
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009747904030550556,
+      "loss": 1.5801,
+      "step": 5877
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009745023554752196,
+      "loss": 1.5918,
+      "step": 5878
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009742143100123197,
+      "loss": 1.752,
+      "step": 5879
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009739262666902711,
+      "loss": 1.5312,
+      "step": 5880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009736382255329884,
+      "loss": 1.7617,
+      "step": 5881
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009733501865643859,
+      "loss": 1.6738,
+      "step": 5882
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009730621498083783,
+      "loss": 1.5586,
+      "step": 5883
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009727741152888799,
+      "loss": 1.623,
+      "step": 5884
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009724860830298044,
+      "loss": 1.7344,
+      "step": 5885
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000972198053055066,
+      "loss": 1.6016,
+      "step": 5886
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009719100253885773,
+      "loss": 1.666,
+      "step": 5887
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000971622000054253,
+      "loss": 1.6035,
+      "step": 5888
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000971333977076006,
+      "loss": 1.6035,
+      "step": 5889
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009710459564777489,
+      "loss": 1.6777,
+      "step": 5890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009707579382833946,
+      "loss": 1.5801,
+      "step": 5891
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009704699225168563,
+      "loss": 1.5273,
+      "step": 5892
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009701819092020461,
+      "loss": 1.4531,
+      "step": 5893
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009698938983628763,
+      "loss": 1.5781,
+      "step": 5894
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009696058900232583,
+      "loss": 1.6953,
+      "step": 5895
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009693178842071051,
+      "loss": 1.5625,
+      "step": 5896
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009690298809383275,
+      "loss": 1.6074,
+      "step": 5897
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009687418802408374,
+      "loss": 1.6406,
+      "step": 5898
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009684538821385453,
+      "loss": 1.5391,
+      "step": 5899
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000968165886655363,
+      "loss": 1.5293,
+      "step": 5900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009678778938152008,
+      "loss": 1.582,
+      "step": 5901
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000967589903641969,
+      "loss": 1.5195,
+      "step": 5902
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009673019161595783,
+      "loss": 1.5469,
+      "step": 5903
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009670139313919389,
+      "loss": 1.5918,
+      "step": 5904
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009667259493629605,
+      "loss": 1.6016,
+      "step": 5905
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009664379700965528,
+      "loss": 1.6484,
+      "step": 5906
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009661499936166247,
+      "loss": 1.584,
+      "step": 5907
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009658620199470862,
+      "loss": 1.498,
+      "step": 5908
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009655740491118459,
+      "loss": 1.5977,
+      "step": 5909
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009652860811348125,
+      "loss": 1.6641,
+      "step": 5910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009649981160398941,
+      "loss": 1.5918,
+      "step": 5911
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009647101538509997,
+      "loss": 1.4707,
+      "step": 5912
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000964422194592037,
+      "loss": 1.7246,
+      "step": 5913
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009641342382869135,
+      "loss": 1.6973,
+      "step": 5914
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009638462849595365,
+      "loss": 1.6484,
+      "step": 5915
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009635583346338141,
+      "loss": 1.5645,
+      "step": 5916
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009632703873336529,
+      "loss": 1.5898,
+      "step": 5917
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009629824430829592,
+      "loss": 1.543,
+      "step": 5918
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009626945019056404,
+      "loss": 1.5723,
+      "step": 5919
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009624065638256019,
+      "loss": 1.6191,
+      "step": 5920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009621186288667504,
+      "loss": 1.5078,
+      "step": 5921
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009618306970529907,
+      "loss": 1.7188,
+      "step": 5922
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009615427684082293,
+      "loss": 1.6074,
+      "step": 5923
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000961254842956371,
+      "loss": 1.4473,
+      "step": 5924
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009609669207213207,
+      "loss": 1.5801,
+      "step": 5925
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009606790017269828,
+      "loss": 1.5957,
+      "step": 5926
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009603910859972624,
+      "loss": 1.6855,
+      "step": 5927
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009601031735560631,
+      "loss": 1.5078,
+      "step": 5928
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000959815264427289,
+      "loss": 1.5625,
+      "step": 5929
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009595273586348433,
+      "loss": 1.5605,
+      "step": 5930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00095923945620263,
+      "loss": 1.5938,
+      "step": 5931
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009589515571545518,
+      "loss": 1.6445,
+      "step": 5932
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009586636615145113,
+      "loss": 1.5918,
+      "step": 5933
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009583757693064107,
+      "loss": 1.6738,
+      "step": 5934
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000958087880554153,
+      "loss": 1.5684,
+      "step": 5935
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009577999952816397,
+      "loss": 1.625,
+      "step": 5936
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009575121135127723,
+      "loss": 1.5801,
+      "step": 5937
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009572242352714522,
+      "loss": 1.6484,
+      "step": 5938
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009569363605815804,
+      "loss": 1.6602,
+      "step": 5939
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009566484894670579,
+      "loss": 1.6641,
+      "step": 5940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009563606219517847,
+      "loss": 1.5488,
+      "step": 5941
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009560727580596608,
+      "loss": 1.7129,
+      "step": 5942
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009557848978145868,
+      "loss": 1.5312,
+      "step": 5943
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009554970412404618,
+      "loss": 1.5684,
+      "step": 5944
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000955209188361185,
+      "loss": 1.6367,
+      "step": 5945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009549213392006548,
+      "loss": 1.7324,
+      "step": 5946
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009546334937827709,
+      "loss": 1.5156,
+      "step": 5947
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009543456521314312,
+      "loss": 1.5645,
+      "step": 5948
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009540578142705334,
+      "loss": 1.5547,
+      "step": 5949
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009537699802239749,
+      "loss": 1.5957,
+      "step": 5950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000953482150015654,
+      "loss": 1.6387,
+      "step": 5951
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009531943236694672,
+      "loss": 1.5352,
+      "step": 5952
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009529065012093113,
+      "loss": 1.5332,
+      "step": 5953
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009526186826590821,
+      "loss": 1.5762,
+      "step": 5954
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009523308680426766,
+      "loss": 1.5625,
+      "step": 5955
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009520430573839903,
+      "loss": 1.6367,
+      "step": 5956
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009517552507069184,
+      "loss": 1.6387,
+      "step": 5957
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009514674480353558,
+      "loss": 1.7227,
+      "step": 5958
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009511796493931978,
+      "loss": 1.4629,
+      "step": 5959
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009508918548043385,
+      "loss": 1.6426,
+      "step": 5960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009506040642926717,
+      "loss": 1.668,
+      "step": 5961
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009503162778820917,
+      "loss": 1.5977,
+      "step": 5962
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009500284955964916,
+      "loss": 1.6582,
+      "step": 5963
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009497407174597647,
+      "loss": 1.5898,
+      "step": 5964
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009494529434958033,
+      "loss": 1.5117,
+      "step": 5965
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009491651737284998,
+      "loss": 1.5977,
+      "step": 5966
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009488774081817467,
+      "loss": 1.6387,
+      "step": 5967
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009485896468794356,
+      "loss": 1.5059,
+      "step": 5968
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009483018898454574,
+      "loss": 1.582,
+      "step": 5969
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009480141371037029,
+      "loss": 1.709,
+      "step": 5970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009477263886780634,
+      "loss": 1.6172,
+      "step": 5971
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000947438644592429,
+      "loss": 1.5898,
+      "step": 5972
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009471509048706895,
+      "loss": 1.6367,
+      "step": 5973
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009468631695367337,
+      "loss": 1.5605,
+      "step": 5974
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009465754386144518,
+      "loss": 1.623,
+      "step": 5975
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009462877121277323,
+      "loss": 1.5762,
+      "step": 5976
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009459999901004635,
+      "loss": 1.5762,
+      "step": 5977
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009457122725565334,
+      "loss": 1.6855,
+      "step": 5978
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009454245595198298,
+      "loss": 1.5059,
+      "step": 5979
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009451368510142402,
+      "loss": 1.623,
+      "step": 5980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009448491470636509,
+      "loss": 1.748,
+      "step": 5981
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000944561447691949,
+      "loss": 1.584,
+      "step": 5982
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009442737529230206,
+      "loss": 1.6172,
+      "step": 5983
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009439860627807516,
+      "loss": 1.5215,
+      "step": 5984
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009436983772890271,
+      "loss": 1.541,
+      "step": 5985
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000943410696471732,
+      "loss": 1.5801,
+      "step": 5986
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009431230203527515,
+      "loss": 1.6875,
+      "step": 5987
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009428353489559696,
+      "loss": 1.7871,
+      "step": 5988
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009425476823052702,
+      "loss": 1.6094,
+      "step": 5989
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009422600204245362,
+      "loss": 1.6543,
+      "step": 5990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009419723633376515,
+      "loss": 1.5352,
+      "step": 5991
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009416847110684985,
+      "loss": 1.5488,
+      "step": 5992
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009413970636409593,
+      "loss": 1.6367,
+      "step": 5993
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009411094210789154,
+      "loss": 1.6836,
+      "step": 5994
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009408217834062492,
+      "loss": 1.5957,
+      "step": 5995
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009405341506468411,
+      "loss": 1.5195,
+      "step": 5996
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009402465228245719,
+      "loss": 1.5938,
+      "step": 5997
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009399588999633216,
+      "loss": 1.582,
+      "step": 5998
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009396712820869706,
+      "loss": 1.6152,
+      "step": 5999
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009393836692193979,
+      "loss": 1.5137,
+      "step": 6000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000939096061384482,
+      "loss": 1.6602,
+      "step": 6001
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009388084586061026,
+      "loss": 1.4844,
+      "step": 6002
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009385208609081373,
+      "loss": 1.5273,
+      "step": 6003
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009382332683144637,
+      "loss": 1.5254,
+      "step": 6004
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009379456808489588,
+      "loss": 1.5195,
+      "step": 6005
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009376580985355005,
+      "loss": 1.6934,
+      "step": 6006
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009373705213979645,
+      "loss": 1.6074,
+      "step": 6007
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009370829494602271,
+      "loss": 1.5625,
+      "step": 6008
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009367953827461632,
+      "loss": 1.5449,
+      "step": 6009
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009365078212796492,
+      "loss": 1.6348,
+      "step": 6010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009362202650845591,
+      "loss": 1.4473,
+      "step": 6011
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009359327141847673,
+      "loss": 1.5332,
+      "step": 6012
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009356451686041472,
+      "loss": 1.4941,
+      "step": 6013
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000935357628366573,
+      "loss": 1.5781,
+      "step": 6014
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009350700934959174,
+      "loss": 1.6816,
+      "step": 6015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009347825640160526,
+      "loss": 1.541,
+      "step": 6016
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009344950399508508,
+      "loss": 1.5312,
+      "step": 6017
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000934207521324184,
+      "loss": 1.6328,
+      "step": 6018
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000933920008159923,
+      "loss": 1.6953,
+      "step": 6019
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009336325004819383,
+      "loss": 1.5938,
+      "step": 6020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009333449983141007,
+      "loss": 1.4961,
+      "step": 6021
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009330575016802797,
+      "loss": 1.5938,
+      "step": 6022
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009327700106043449,
+      "loss": 1.5605,
+      "step": 6023
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009324825251101649,
+      "loss": 1.5078,
+      "step": 6024
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009321950452216077,
+      "loss": 1.6328,
+      "step": 6025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009319075709625422,
+      "loss": 1.6406,
+      "step": 6026
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009316201023568355,
+      "loss": 1.5391,
+      "step": 6027
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009313326394283546,
+      "loss": 1.4707,
+      "step": 6028
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009310451822009656,
+      "loss": 1.5566,
+      "step": 6029
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009307577306985354,
+      "loss": 1.625,
+      "step": 6030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000930470284944929,
+      "loss": 1.5762,
+      "step": 6031
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009301828449640117,
+      "loss": 1.4375,
+      "step": 6032
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009298954107796477,
+      "loss": 1.5117,
+      "step": 6033
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009296079824157022,
+      "loss": 1.5957,
+      "step": 6034
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009293205598960383,
+      "loss": 1.7051,
+      "step": 6035
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000929033143244519,
+      "loss": 1.6973,
+      "step": 6036
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000928745732485007,
+      "loss": 1.4766,
+      "step": 6037
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000928458327641365,
+      "loss": 1.7344,
+      "step": 6038
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009281709287374542,
+      "loss": 1.6191,
+      "step": 6039
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009278835357971359,
+      "loss": 1.6738,
+      "step": 6040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000927596148844271,
+      "loss": 1.5703,
+      "step": 6041
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009273087679027199,
+      "loss": 1.6113,
+      "step": 6042
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000927021392996342,
+      "loss": 1.5352,
+      "step": 6043
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009267340241489968,
+      "loss": 1.6074,
+      "step": 6044
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009264466613845426,
+      "loss": 1.7109,
+      "step": 6045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009261593047268381,
+      "loss": 1.666,
+      "step": 6046
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009258719541997411,
+      "loss": 1.5996,
+      "step": 6047
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009255846098271085,
+      "loss": 1.5645,
+      "step": 6048
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009252972716327967,
+      "loss": 1.6836,
+      "step": 6049
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009250099396406626,
+      "loss": 1.6875,
+      "step": 6050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009247226138745616,
+      "loss": 1.6074,
+      "step": 6051
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009244352943583488,
+      "loss": 1.5918,
+      "step": 6052
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009241479811158785,
+      "loss": 1.7422,
+      "step": 6053
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009238606741710055,
+      "loss": 1.7852,
+      "step": 6054
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000923573373547583,
+      "loss": 1.418,
+      "step": 6055
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009232860792694642,
+      "loss": 1.7559,
+      "step": 6056
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009229987913605014,
+      "loss": 1.6445,
+      "step": 6057
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009227115098445469,
+      "loss": 1.5742,
+      "step": 6058
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009224242347454521,
+      "loss": 1.5527,
+      "step": 6059
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009221369660870676,
+      "loss": 1.6895,
+      "step": 6060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009218497038932443,
+      "loss": 1.6855,
+      "step": 6061
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009215624481878321,
+      "loss": 1.5547,
+      "step": 6062
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00092127519899468,
+      "loss": 1.6211,
+      "step": 6063
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009209879563376371,
+      "loss": 1.6309,
+      "step": 6064
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009207007202405511,
+      "loss": 1.5684,
+      "step": 6065
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009204134907272705,
+      "loss": 1.6172,
+      "step": 6066
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009201262678216421,
+      "loss": 1.6484,
+      "step": 6067
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009198390515475124,
+      "loss": 1.5957,
+      "step": 6068
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009195518419287274,
+      "loss": 1.5938,
+      "step": 6069
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000919264638989133,
+      "loss": 1.5312,
+      "step": 6070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000918977442752574,
+      "loss": 1.6602,
+      "step": 6071
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009186902532428948,
+      "loss": 1.5957,
+      "step": 6072
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009184030704839387,
+      "loss": 1.4512,
+      "step": 6073
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00091811589449955,
+      "loss": 1.5293,
+      "step": 6074
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009178287253135709,
+      "loss": 1.6973,
+      "step": 6075
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009175415629498434,
+      "loss": 1.5469,
+      "step": 6076
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009172544074322092,
+      "loss": 1.5918,
+      "step": 6077
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009169672587845096,
+      "loss": 1.3848,
+      "step": 6078
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009166801170305848,
+      "loss": 1.5117,
+      "step": 6079
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009163929821942745,
+      "loss": 1.6621,
+      "step": 6080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009161058542994182,
+      "loss": 1.6797,
+      "step": 6081
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009158187333698549,
+      "loss": 1.584,
+      "step": 6082
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009155316194294224,
+      "loss": 1.6758,
+      "step": 6083
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009152445125019584,
+      "loss": 1.6113,
+      "step": 6084
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009149574126112995,
+      "loss": 1.5859,
+      "step": 6085
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009146703197812827,
+      "loss": 1.5918,
+      "step": 6086
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009143832340357439,
+      "loss": 1.6816,
+      "step": 6087
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009140961553985173,
+      "loss": 1.5898,
+      "step": 6088
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009138090838934388,
+      "loss": 1.6426,
+      "step": 6089
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009135220195443419,
+      "loss": 1.5078,
+      "step": 6090
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00091323496237506,
+      "loss": 1.5664,
+      "step": 6091
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009129479124094256,
+      "loss": 1.5801,
+      "step": 6092
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009126608696712718,
+      "loss": 1.6191,
+      "step": 6093
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009123738341844298,
+      "loss": 1.627,
+      "step": 6094
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009120868059727307,
+      "loss": 1.7266,
+      "step": 6095
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009117997850600047,
+      "loss": 1.6562,
+      "step": 6096
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009115127714700821,
+      "loss": 1.6035,
+      "step": 6097
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000911225765226792,
+      "loss": 1.6465,
+      "step": 6098
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009109387663539627,
+      "loss": 1.541,
+      "step": 6099
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009106517748754225,
+      "loss": 1.6406,
+      "step": 6100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009103647908149989,
+      "loss": 1.582,
+      "step": 6101
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009100778141965187,
+      "loss": 1.7305,
+      "step": 6102
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009097908450438077,
+      "loss": 1.5312,
+      "step": 6103
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009095038833806915,
+      "loss": 1.6543,
+      "step": 6104
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009092169292309956,
+      "loss": 1.7246,
+      "step": 6105
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009089299826185438,
+      "loss": 1.6133,
+      "step": 6106
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009086430435671599,
+      "loss": 1.6133,
+      "step": 6107
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009083561121006665,
+      "loss": 1.5312,
+      "step": 6108
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009080691882428869,
+      "loss": 1.5137,
+      "step": 6109
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009077822720176426,
+      "loss": 1.498,
+      "step": 6110
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009074953634487545,
+      "loss": 1.5195,
+      "step": 6111
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009072084625600427,
+      "loss": 1.4961,
+      "step": 6112
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009069215693753282,
+      "loss": 1.5586,
+      "step": 6113
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009066346839184296,
+      "loss": 1.6836,
+      "step": 6114
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009063478062131656,
+      "loss": 1.5352,
+      "step": 6115
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009060609362833539,
+      "loss": 1.7188,
+      "step": 6116
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009057740741528121,
+      "loss": 1.6484,
+      "step": 6117
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009054872198453568,
+      "loss": 1.5488,
+      "step": 6118
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000905200373384804,
+      "loss": 1.623,
+      "step": 6119
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009049135347949689,
+      "loss": 1.7109,
+      "step": 6120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009046267040996667,
+      "loss": 1.6641,
+      "step": 6121
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000904339881322711,
+      "loss": 1.6211,
+      "step": 6122
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009040530664879153,
+      "loss": 1.5547,
+      "step": 6123
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009037662596190921,
+      "loss": 1.6035,
+      "step": 6124
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009034794607400541,
+      "loss": 1.6094,
+      "step": 6125
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009031926698746123,
+      "loss": 1.6094,
+      "step": 6126
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009029058870465776,
+      "loss": 1.6523,
+      "step": 6127
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009026191122797595,
+      "loss": 1.584,
+      "step": 6128
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009023323455979683,
+      "loss": 1.6172,
+      "step": 6129
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009020455870250123,
+      "loss": 1.5645,
+      "step": 6130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009017588365846999,
+      "loss": 1.5215,
+      "step": 6131
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009014720943008374,
+      "loss": 1.6504,
+      "step": 6132
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009011853601972328,
+      "loss": 1.543,
+      "step": 6133
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009008986342976919,
+      "loss": 1.5527,
+      "step": 6134
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009006119166260195,
+      "loss": 1.6348,
+      "step": 6135
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009003252072060204,
+      "loss": 1.627,
+      "step": 6136
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000900038506061499,
+      "loss": 1.5078,
+      "step": 6137
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008997518132162584,
+      "loss": 1.6016,
+      "step": 6138
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008994651286941011,
+      "loss": 1.5938,
+      "step": 6139
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008991784525188288,
+      "loss": 1.5664,
+      "step": 6140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008988917847142433,
+      "loss": 1.6055,
+      "step": 6141
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008986051253041448,
+      "loss": 1.5859,
+      "step": 6142
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008983184743123333,
+      "loss": 1.5586,
+      "step": 6143
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008980318317626072,
+      "loss": 1.6191,
+      "step": 6144
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008977451976787662,
+      "loss": 1.5059,
+      "step": 6145
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008974585720846073,
+      "loss": 1.6895,
+      "step": 6146
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008971719550039275,
+      "loss": 1.5488,
+      "step": 6147
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008968853464605227,
+      "loss": 1.6328,
+      "step": 6148
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008965987464781897,
+      "loss": 1.5938,
+      "step": 6149
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008963121550807225,
+      "loss": 1.5586,
+      "step": 6150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008960255722919156,
+      "loss": 1.623,
+      "step": 6151
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008957389981355619,
+      "loss": 1.5605,
+      "step": 6152
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008954524326354552,
+      "loss": 1.5605,
+      "step": 6153
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008951658758153869,
+      "loss": 1.623,
+      "step": 6154
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008948793276991484,
+      "loss": 1.623,
+      "step": 6155
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008945927883105301,
+      "loss": 1.584,
+      "step": 6156
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008943062576733222,
+      "loss": 1.6836,
+      "step": 6157
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008940197358113135,
+      "loss": 1.5195,
+      "step": 6158
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008937332227482929,
+      "loss": 1.6504,
+      "step": 6159
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008934467185080474,
+      "loss": 1.582,
+      "step": 6160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008931602231143647,
+      "loss": 1.5547,
+      "step": 6161
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008928737365910307,
+      "loss": 1.4883,
+      "step": 6162
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008925872589618307,
+      "loss": 1.5742,
+      "step": 6163
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008923007902505492,
+      "loss": 1.4707,
+      "step": 6164
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000892014330480971,
+      "loss": 1.668,
+      "step": 6165
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008917278796768791,
+      "loss": 1.625,
+      "step": 6166
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008914414378620556,
+      "loss": 1.6895,
+      "step": 6167
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008911550050602822,
+      "loss": 1.5215,
+      "step": 6168
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008908685812953408,
+      "loss": 1.6406,
+      "step": 6169
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000890582166591011,
+      "loss": 1.5371,
+      "step": 6170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008902957609710719,
+      "loss": 1.7207,
+      "step": 6171
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008900093644593033,
+      "loss": 1.6289,
+      "step": 6172
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008897229770794828,
+      "loss": 1.7109,
+      "step": 6173
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008894365988553874,
+      "loss": 1.582,
+      "step": 6174
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008891502298107937,
+      "loss": 1.668,
+      "step": 6175
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008888638699694775,
+      "loss": 1.4941,
+      "step": 6176
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008885775193552137,
+      "loss": 1.5352,
+      "step": 6177
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008882911779917764,
+      "loss": 1.6621,
+      "step": 6178
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000888004845902939,
+      "loss": 1.5645,
+      "step": 6179
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008877185231124747,
+      "loss": 1.6797,
+      "step": 6180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000887432209644155,
+      "loss": 1.4434,
+      "step": 6181
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008871459055217509,
+      "loss": 1.5859,
+      "step": 6182
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008868596107690324,
+      "loss": 1.5723,
+      "step": 6183
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00088657332540977,
+      "loss": 1.7012,
+      "step": 6184
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000886287049467732,
+      "loss": 1.5234,
+      "step": 6185
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008860007829666862,
+      "loss": 1.6641,
+      "step": 6186
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008857145259303997,
+      "loss": 1.5039,
+      "step": 6187
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008854282783826396,
+      "loss": 1.6973,
+      "step": 6188
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008851420403471711,
+      "loss": 1.6367,
+      "step": 6189
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008848558118477592,
+      "loss": 1.4746,
+      "step": 6190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008845695929081673,
+      "loss": 1.5977,
+      "step": 6191
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008842833835521598,
+      "loss": 1.5352,
+      "step": 6192
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008839971838034984,
+      "loss": 1.6484,
+      "step": 6193
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008837109936859451,
+      "loss": 1.4746,
+      "step": 6194
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008834248132232603,
+      "loss": 1.5762,
+      "step": 6195
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008831386424392047,
+      "loss": 1.7109,
+      "step": 6196
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008828524813575372,
+      "loss": 1.4785,
+      "step": 6197
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008825663300020162,
+      "loss": 1.5801,
+      "step": 6198
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008822801883963993,
+      "loss": 1.7168,
+      "step": 6199
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000881994056564444,
+      "loss": 1.6543,
+      "step": 6200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008817079345299056,
+      "loss": 1.6367,
+      "step": 6201
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008814218223165397,
+      "loss": 1.6406,
+      "step": 6202
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008811357199481001,
+      "loss": 1.6172,
+      "step": 6203
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008808496274483413,
+      "loss": 1.6133,
+      "step": 6204
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008805635448410155,
+      "loss": 1.6602,
+      "step": 6205
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008802774721498749,
+      "loss": 1.7227,
+      "step": 6206
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00087999140939867,
+      "loss": 1.6621,
+      "step": 6207
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008797053566111521,
+      "loss": 1.4785,
+      "step": 6208
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008794193138110701,
+      "loss": 1.6426,
+      "step": 6209
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008791332810221726,
+      "loss": 1.5938,
+      "step": 6210
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008788472582682073,
+      "loss": 1.4512,
+      "step": 6211
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008785612455729215,
+      "loss": 1.5469,
+      "step": 6212
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008782752429600614,
+      "loss": 1.5137,
+      "step": 6213
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000877989250453372,
+      "loss": 1.5137,
+      "step": 6214
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008777032680765979,
+      "loss": 1.7031,
+      "step": 6215
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008774172958534827,
+      "loss": 1.5938,
+      "step": 6216
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008771313338077694,
+      "loss": 1.6484,
+      "step": 6217
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008768453819631993,
+      "loss": 1.5859,
+      "step": 6218
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000876559440343514,
+      "loss": 1.6289,
+      "step": 6219
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008762735089724538,
+      "loss": 1.5684,
+      "step": 6220
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000875987587873758,
+      "loss": 1.6562,
+      "step": 6221
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000875701677071165,
+      "loss": 1.541,
+      "step": 6222
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008754157765884122,
+      "loss": 1.5762,
+      "step": 6223
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008751298864492371,
+      "loss": 1.6309,
+      "step": 6224
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008748440066773755,
+      "loss": 1.5117,
+      "step": 6225
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008745581372965624,
+      "loss": 1.6836,
+      "step": 6226
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008742722783305312,
+      "loss": 1.625,
+      "step": 6227
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008739864298030168,
+      "loss": 1.5371,
+      "step": 6228
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000873700591737751,
+      "loss": 1.6426,
+      "step": 6229
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008734147641584652,
+      "loss": 1.623,
+      "step": 6230
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008731289470888901,
+      "loss": 1.6523,
+      "step": 6231
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008728431405527563,
+      "loss": 1.6621,
+      "step": 6232
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008725573445737923,
+      "loss": 1.5371,
+      "step": 6233
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008722715591757265,
+      "loss": 1.5156,
+      "step": 6234
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008719857843822858,
+      "loss": 1.5684,
+      "step": 6235
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008717000202171969,
+      "loss": 1.627,
+      "step": 6236
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008714142667041851,
+      "loss": 1.582,
+      "step": 6237
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008711285238669754,
+      "loss": 1.6016,
+      "step": 6238
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000870842791729291,
+      "loss": 1.5117,
+      "step": 6239
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008705570703148553,
+      "loss": 1.6035,
+      "step": 6240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008702713596473899,
+      "loss": 1.5391,
+      "step": 6241
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008699856597506161,
+      "loss": 1.5781,
+      "step": 6242
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008696999706482533,
+      "loss": 1.7031,
+      "step": 6243
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008694142923640219,
+      "loss": 1.543,
+      "step": 6244
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008691286249216398,
+      "loss": 1.5938,
+      "step": 6245
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008688429683448245,
+      "loss": 1.5723,
+      "step": 6246
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008685573226572918,
+      "loss": 1.6816,
+      "step": 6247
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008682716878827588,
+      "loss": 1.4785,
+      "step": 6248
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008679860640449394,
+      "loss": 1.5996,
+      "step": 6249
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008677004511675477,
+      "loss": 1.584,
+      "step": 6250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008674148492742961,
+      "loss": 1.6973,
+      "step": 6251
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008671292583888974,
+      "loss": 1.7441,
+      "step": 6252
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008668436785350626,
+      "loss": 1.627,
+      "step": 6253
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008665581097365016,
+      "loss": 1.6699,
+      "step": 6254
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008662725520169236,
+      "loss": 1.4648,
+      "step": 6255
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008659870054000374,
+      "loss": 1.5234,
+      "step": 6256
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008657014699095499,
+      "loss": 1.6953,
+      "step": 6257
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008654159455691679,
+      "loss": 1.748,
+      "step": 6258
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008651304324025973,
+      "loss": 1.6504,
+      "step": 6259
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008648449304335425,
+      "loss": 1.6113,
+      "step": 6260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008645594396857073,
+      "loss": 1.6367,
+      "step": 6261
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008642739601827938,
+      "loss": 1.5859,
+      "step": 6262
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000863988491948505,
+      "loss": 1.6328,
+      "step": 6263
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008637030350065415,
+      "loss": 1.5762,
+      "step": 6264
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008634175893806029,
+      "loss": 1.7266,
+      "step": 6265
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008631321550943882,
+      "loss": 1.5938,
+      "step": 6266
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008628467321715963,
+      "loss": 1.5664,
+      "step": 6267
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008625613206359237,
+      "loss": 1.6562,
+      "step": 6268
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008622759205110669,
+      "loss": 1.5742,
+      "step": 6269
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008619905318207207,
+      "loss": 1.5039,
+      "step": 6270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008617051545885804,
+      "loss": 1.5723,
+      "step": 6271
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008614197888383386,
+      "loss": 1.5352,
+      "step": 6272
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000861134434593688,
+      "loss": 1.7227,
+      "step": 6273
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008608490918783197,
+      "loss": 1.6289,
+      "step": 6274
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008605637607159249,
+      "loss": 1.7578,
+      "step": 6275
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008602784411301927,
+      "loss": 1.5723,
+      "step": 6276
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008599931331448117,
+      "loss": 1.6465,
+      "step": 6277
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008597078367834695,
+      "loss": 1.5391,
+      "step": 6278
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000859422552069853,
+      "loss": 1.584,
+      "step": 6279
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008591372790276478,
+      "loss": 1.5508,
+      "step": 6280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008588520176805388,
+      "loss": 1.5859,
+      "step": 6281
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008585667680522089,
+      "loss": 1.5469,
+      "step": 6282
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008582815301663422,
+      "loss": 1.5098,
+      "step": 6283
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008579963040466197,
+      "loss": 1.5195,
+      "step": 6284
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008577110897167225,
+      "loss": 1.582,
+      "step": 6285
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008574258872003299,
+      "loss": 1.5547,
+      "step": 6286
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008571406965211215,
+      "loss": 1.498,
+      "step": 6287
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008568555177027753,
+      "loss": 1.5391,
+      "step": 6288
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008565703507689675,
+      "loss": 1.6074,
+      "step": 6289
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000856285195743374,
+      "loss": 1.4629,
+      "step": 6290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008560000526496706,
+      "loss": 1.6152,
+      "step": 6291
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008557149215115308,
+      "loss": 1.6074,
+      "step": 6292
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008554298023526273,
+      "loss": 1.6094,
+      "step": 6293
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008551446951966321,
+      "loss": 1.5508,
+      "step": 6294
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008548596000672164,
+      "loss": 1.6328,
+      "step": 6295
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00085457451698805,
+      "loss": 1.7188,
+      "step": 6296
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008542894459828021,
+      "loss": 1.5625,
+      "step": 6297
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008540043870751402,
+      "loss": 1.5508,
+      "step": 6298
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008537193402887316,
+      "loss": 1.6641,
+      "step": 6299
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008534343056472424,
+      "loss": 1.502,
+      "step": 6300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008531492831743372,
+      "loss": 1.6348,
+      "step": 6301
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008528642728936795,
+      "loss": 1.8203,
+      "step": 6302
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008525792748289333,
+      "loss": 1.6289,
+      "step": 6303
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008522942890037599,
+      "loss": 1.5039,
+      "step": 6304
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008520093154418203,
+      "loss": 1.7344,
+      "step": 6305
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008517243541667737,
+      "loss": 1.6699,
+      "step": 6306
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008514394052022801,
+      "loss": 1.5859,
+      "step": 6307
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008511544685719968,
+      "loss": 1.6211,
+      "step": 6308
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008508695442995805,
+      "loss": 1.5117,
+      "step": 6309
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008505846324086865,
+      "loss": 1.668,
+      "step": 6310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008502997329229707,
+      "loss": 1.5195,
+      "step": 6311
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008500148458660861,
+      "loss": 1.5957,
+      "step": 6312
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008497299712616856,
+      "loss": 1.5938,
+      "step": 6313
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008494451091334204,
+      "loss": 1.6328,
+      "step": 6314
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008491602595049418,
+      "loss": 1.6582,
+      "step": 6315
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008488754223998987,
+      "loss": 1.6641,
+      "step": 6316
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008485905978419402,
+      "loss": 1.4863,
+      "step": 6317
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008483057858547134,
+      "loss": 1.4844,
+      "step": 6318
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008480209864618651,
+      "loss": 1.6504,
+      "step": 6319
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008477361996870405,
+      "loss": 1.5625,
+      "step": 6320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000847451425553884,
+      "loss": 1.5645,
+      "step": 6321
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008471666640860384,
+      "loss": 1.6074,
+      "step": 6322
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000846881915307147,
+      "loss": 1.6152,
+      "step": 6323
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008465971792408503,
+      "loss": 1.6816,
+      "step": 6324
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008463124559107887,
+      "loss": 1.5762,
+      "step": 6325
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008460277453406008,
+      "loss": 1.6367,
+      "step": 6326
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008457430475539254,
+      "loss": 1.5371,
+      "step": 6327
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000845458362574399,
+      "loss": 1.457,
+      "step": 6328
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008451736904256577,
+      "loss": 1.707,
+      "step": 6329
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008448890311313357,
+      "loss": 1.582,
+      "step": 6330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000844604384715068,
+      "loss": 1.5664,
+      "step": 6331
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008443197512004863,
+      "loss": 1.7148,
+      "step": 6332
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008440351306112227,
+      "loss": 1.6406,
+      "step": 6333
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008437505229709073,
+      "loss": 1.6992,
+      "step": 6334
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008434659283031703,
+      "loss": 1.6172,
+      "step": 6335
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008431813466316393,
+      "loss": 1.5156,
+      "step": 6336
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008428967779799425,
+      "loss": 1.6465,
+      "step": 6337
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008426122223717053,
+      "loss": 1.4766,
+      "step": 6338
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008423276798305534,
+      "loss": 1.6719,
+      "step": 6339
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008420431503801109,
+      "loss": 1.6914,
+      "step": 6340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008417586340440001,
+      "loss": 1.7012,
+      "step": 6341
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008414741308458439,
+      "loss": 1.5605,
+      "step": 6342
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008411896408092626,
+      "loss": 1.5547,
+      "step": 6343
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008409051639578761,
+      "loss": 1.6328,
+      "step": 6344
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008406207003153023,
+      "loss": 1.4746,
+      "step": 6345
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008403362499051598,
+      "loss": 1.623,
+      "step": 6346
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008400518127510647,
+      "loss": 1.6133,
+      "step": 6347
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000839767388876632,
+      "loss": 1.5977,
+      "step": 6348
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008394829783054758,
+      "loss": 1.5566,
+      "step": 6349
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00083919858106121,
+      "loss": 1.6191,
+      "step": 6350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008389141971674461,
+      "loss": 1.4961,
+      "step": 6351
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000838629826647795,
+      "loss": 1.5879,
+      "step": 6352
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008383454695258665,
+      "loss": 1.6914,
+      "step": 6353
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008380611258252694,
+      "loss": 1.5117,
+      "step": 6354
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008377767955696114,
+      "loss": 1.5723,
+      "step": 6355
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008374924787824986,
+      "loss": 1.4941,
+      "step": 6356
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008372081754875364,
+      "loss": 1.6523,
+      "step": 6357
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008369238857083293,
+      "loss": 1.5684,
+      "step": 6358
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008366396094684804,
+      "loss": 1.6895,
+      "step": 6359
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008363553467915914,
+      "loss": 1.6094,
+      "step": 6360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008360710977012628,
+      "loss": 1.4824,
+      "step": 6361
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008357868622210953,
+      "loss": 1.6348,
+      "step": 6362
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008355026403746869,
+      "loss": 1.6699,
+      "step": 6363
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008352184321856352,
+      "loss": 1.6758,
+      "step": 6364
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008349342376775359,
+      "loss": 1.6406,
+      "step": 6365
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008346500568739852,
+      "loss": 1.6621,
+      "step": 6366
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008343658897985766,
+      "loss": 1.543,
+      "step": 6367
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008340817364749032,
+      "loss": 1.6328,
+      "step": 6368
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008337975969265561,
+      "loss": 1.5859,
+      "step": 6369
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000833513471177127,
+      "loss": 1.6797,
+      "step": 6370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008332293592502047,
+      "loss": 1.5859,
+      "step": 6371
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008329452611693778,
+      "loss": 1.5801,
+      "step": 6372
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008326611769582331,
+      "loss": 1.5703,
+      "step": 6373
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008323771066403571,
+      "loss": 1.7305,
+      "step": 6374
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008320930502393341,
+      "loss": 1.5586,
+      "step": 6375
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008318090077787485,
+      "loss": 1.5918,
+      "step": 6376
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008315249792821824,
+      "loss": 1.627,
+      "step": 6377
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008312409647732175,
+      "loss": 1.5176,
+      "step": 6378
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008309569642754339,
+      "loss": 1.6426,
+      "step": 6379
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008306729778124105,
+      "loss": 1.6953,
+      "step": 6380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008303890054077252,
+      "loss": 1.6016,
+      "step": 6381
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008301050470849551,
+      "loss": 1.6758,
+      "step": 6382
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008298211028676758,
+      "loss": 1.6602,
+      "step": 6383
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008295371727794614,
+      "loss": 1.5605,
+      "step": 6384
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008292532568438849,
+      "loss": 1.4355,
+      "step": 6385
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000828969355084519,
+      "loss": 1.6562,
+      "step": 6386
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008286854675249343,
+      "loss": 1.5684,
+      "step": 6387
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008284015941887004,
+      "loss": 1.6543,
+      "step": 6388
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008281177350993855,
+      "loss": 1.5625,
+      "step": 6389
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008278338902805578,
+      "loss": 1.6211,
+      "step": 6390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000827550059755783,
+      "loss": 1.5156,
+      "step": 6391
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000827266243548626,
+      "loss": 1.6621,
+      "step": 6392
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008269824416826504,
+      "loss": 1.6562,
+      "step": 6393
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008266986541814192,
+      "loss": 1.6367,
+      "step": 6394
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008264148810684935,
+      "loss": 1.5098,
+      "step": 6395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008261311223674337,
+      "loss": 1.6543,
+      "step": 6396
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008258473781017986,
+      "loss": 1.5938,
+      "step": 6397
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008255636482951463,
+      "loss": 1.6914,
+      "step": 6398
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008252799329710331,
+      "loss": 1.5742,
+      "step": 6399
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008249962321530146,
+      "loss": 1.7109,
+      "step": 6400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008247125458646445,
+      "loss": 1.6484,
+      "step": 6401
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008244288741294767,
+      "loss": 1.5469,
+      "step": 6402
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008241452169710624,
+      "loss": 1.6699,
+      "step": 6403
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008238615744129523,
+      "loss": 1.5352,
+      "step": 6404
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008235779464786953,
+      "loss": 1.5898,
+      "step": 6405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008232943331918402,
+      "loss": 1.6191,
+      "step": 6406
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008230107345759339,
+      "loss": 1.543,
+      "step": 6407
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008227271506545218,
+      "loss": 1.5859,
+      "step": 6408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008224435814511481,
+      "loss": 1.6328,
+      "step": 6409
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000822160026989357,
+      "loss": 1.584,
+      "step": 6410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00082187648729269,
+      "loss": 1.5215,
+      "step": 6411
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008215929623846878,
+      "loss": 1.6836,
+      "step": 6412
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008213094522888899,
+      "loss": 1.5625,
+      "step": 6413
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008210259570288353,
+      "loss": 1.6094,
+      "step": 6414
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008207424766280604,
+      "loss": 1.5137,
+      "step": 6415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008204590111101017,
+      "loss": 1.6016,
+      "step": 6416
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008201755604984932,
+      "loss": 1.582,
+      "step": 6417
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008198921248167691,
+      "loss": 1.498,
+      "step": 6418
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008196087040884611,
+      "loss": 1.5625,
+      "step": 6419
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008193252983371002,
+      "loss": 1.627,
+      "step": 6420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008190419075862158,
+      "loss": 1.6562,
+      "step": 6421
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000818758531859337,
+      "loss": 1.5879,
+      "step": 6422
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008184751711799907,
+      "loss": 1.4902,
+      "step": 6423
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008181918255717025,
+      "loss": 1.5645,
+      "step": 6424
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000817908495057998,
+      "loss": 1.6875,
+      "step": 6425
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008176251796623998,
+      "loss": 1.6094,
+      "step": 6426
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008173418794084306,
+      "loss": 1.6797,
+      "step": 6427
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008170585943196107,
+      "loss": 1.6133,
+      "step": 6428
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008167753244194605,
+      "loss": 1.6387,
+      "step": 6429
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008164920697314984,
+      "loss": 1.6777,
+      "step": 6430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008162088302792412,
+      "loss": 1.5215,
+      "step": 6431
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008159256060862046,
+      "loss": 1.5098,
+      "step": 6432
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008156423971759037,
+      "loss": 1.6074,
+      "step": 6433
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008153592035718516,
+      "loss": 1.6309,
+      "step": 6434
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008150760252975607,
+      "loss": 1.543,
+      "step": 6435
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008147928623765412,
+      "loss": 1.623,
+      "step": 6436
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008145097148323034,
+      "loss": 1.6465,
+      "step": 6437
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008142265826883551,
+      "loss": 1.5664,
+      "step": 6438
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008139434659682034,
+      "loss": 1.6074,
+      "step": 6439
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008136603646953535,
+      "loss": 1.752,
+      "step": 6440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008133772788933107,
+      "loss": 1.7012,
+      "step": 6441
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008130942085855777,
+      "loss": 1.6035,
+      "step": 6442
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008128111537956564,
+      "loss": 1.582,
+      "step": 6443
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008125281145470468,
+      "loss": 1.5605,
+      "step": 6444
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008122450908632493,
+      "loss": 1.6191,
+      "step": 6445
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008119620827677611,
+      "loss": 1.5918,
+      "step": 6446
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008116790902840793,
+      "loss": 1.6016,
+      "step": 6447
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008113961134356984,
+      "loss": 1.7344,
+      "step": 6448
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008111131522461137,
+      "loss": 1.6543,
+      "step": 6449
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008108302067388174,
+      "loss": 1.6309,
+      "step": 6450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000810547276937301,
+      "loss": 1.6387,
+      "step": 6451
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008102643628650544,
+      "loss": 1.5977,
+      "step": 6452
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008099814645455671,
+      "loss": 1.6348,
+      "step": 6453
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008096985820023262,
+      "loss": 1.6113,
+      "step": 6454
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008094157152588182,
+      "loss": 1.6777,
+      "step": 6455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008091328643385277,
+      "loss": 1.5801,
+      "step": 6456
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008088500292649388,
+      "loss": 1.6309,
+      "step": 6457
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008085672100615338,
+      "loss": 1.6641,
+      "step": 6458
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008082844067517934,
+      "loss": 1.6406,
+      "step": 6459
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008080016193591969,
+      "loss": 1.5781,
+      "step": 6460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008077188479072237,
+      "loss": 1.5859,
+      "step": 6461
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008074360924193502,
+      "loss": 1.6309,
+      "step": 6462
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008071533529190522,
+      "loss": 1.6777,
+      "step": 6463
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008068706294298037,
+      "loss": 1.582,
+      "step": 6464
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008065879219750785,
+      "loss": 1.6348,
+      "step": 6465
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008063052305783481,
+      "loss": 1.5879,
+      "step": 6466
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008060225552630827,
+      "loss": 1.6855,
+      "step": 6467
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008057398960527509,
+      "loss": 1.6367,
+      "step": 6468
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008054572529708214,
+      "loss": 1.6094,
+      "step": 6469
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008051746260407601,
+      "loss": 1.5547,
+      "step": 6470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008048920152860321,
+      "loss": 1.5273,
+      "step": 6471
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008046094207301007,
+      "loss": 1.543,
+      "step": 6472
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008043268423964289,
+      "loss": 1.5527,
+      "step": 6473
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008040442803084772,
+      "loss": 1.5879,
+      "step": 6474
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008037617344897056,
+      "loss": 1.6113,
+      "step": 6475
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008034792049635721,
+      "loss": 1.6113,
+      "step": 6476
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000803196691753534,
+      "loss": 1.5371,
+      "step": 6477
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008029141948830465,
+      "loss": 1.6289,
+      "step": 6478
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008026317143755643,
+      "loss": 1.6484,
+      "step": 6479
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008023492502545394,
+      "loss": 1.7617,
+      "step": 6480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008020668025434246,
+      "loss": 1.5781,
+      "step": 6481
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008017843712656692,
+      "loss": 1.6699,
+      "step": 6482
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008015019564447223,
+      "loss": 1.6035,
+      "step": 6483
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008012195581040306,
+      "loss": 1.5352,
+      "step": 6484
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008009371762670414,
+      "loss": 1.5391,
+      "step": 6485
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008006548109571986,
+      "loss": 1.5371,
+      "step": 6486
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008003724621979458,
+      "loss": 1.5586,
+      "step": 6487
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008000901300127244,
+      "loss": 1.627,
+      "step": 6488
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007998078144249758,
+      "loss": 1.5352,
+      "step": 6489
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007995255154581388,
+      "loss": 1.5527,
+      "step": 6490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007992432331356511,
+      "loss": 1.5781,
+      "step": 6491
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007989609674809491,
+      "loss": 1.5781,
+      "step": 6492
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007986787185174682,
+      "loss": 1.5527,
+      "step": 6493
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007983964862686415,
+      "loss": 1.5352,
+      "step": 6494
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007981142707579019,
+      "loss": 1.6016,
+      "step": 6495
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007978320720086797,
+      "loss": 1.6445,
+      "step": 6496
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007975498900444049,
+      "loss": 1.7559,
+      "step": 6497
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007972677248885053,
+      "loss": 1.584,
+      "step": 6498
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007969855765644076,
+      "loss": 1.5078,
+      "step": 6499
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007967034450955369,
+      "loss": 1.5254,
+      "step": 6500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007964213305053176,
+      "loss": 1.7676,
+      "step": 6501
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007961392328171721,
+      "loss": 1.5645,
+      "step": 6502
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007958571520545213,
+      "loss": 1.5723,
+      "step": 6503
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007955750882407844,
+      "loss": 1.5723,
+      "step": 6504
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007952930413993808,
+      "loss": 1.6426,
+      "step": 6505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007950110115537267,
+      "loss": 1.5293,
+      "step": 6506
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007947289987272377,
+      "loss": 1.5918,
+      "step": 6507
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007944470029433274,
+      "loss": 1.6602,
+      "step": 6508
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007941650242254092,
+      "loss": 1.8164,
+      "step": 6509
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007938830625968942,
+      "loss": 1.6855,
+      "step": 6510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007936011180811916,
+      "loss": 1.5508,
+      "step": 6511
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007933191907017105,
+      "loss": 1.5234,
+      "step": 6512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007930372804818572,
+      "loss": 1.6816,
+      "step": 6513
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007927553874450379,
+      "loss": 1.4609,
+      "step": 6514
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007924735116146559,
+      "loss": 1.4492,
+      "step": 6515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007921916530141148,
+      "loss": 1.6426,
+      "step": 6516
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007919098116668152,
+      "loss": 1.5254,
+      "step": 6517
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007916279875961571,
+      "loss": 1.5938,
+      "step": 6518
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007913461808255386,
+      "loss": 1.6055,
+      "step": 6519
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007910643913783573,
+      "loss": 1.6074,
+      "step": 6520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007907826192780083,
+      "loss": 1.625,
+      "step": 6521
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007905008645478857,
+      "loss": 1.6777,
+      "step": 6522
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007902191272113817,
+      "loss": 1.5645,
+      "step": 6523
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007899374072918884,
+      "loss": 1.4746,
+      "step": 6524
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007896557048127949,
+      "loss": 1.6172,
+      "step": 6525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007893740197974897,
+      "loss": 1.5723,
+      "step": 6526
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007890923522693591,
+      "loss": 1.4668,
+      "step": 6527
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007888107022517894,
+      "loss": 1.6133,
+      "step": 6528
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000788529069768164,
+      "loss": 1.6367,
+      "step": 6529
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007882474548418654,
+      "loss": 1.6152,
+      "step": 6530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007879658574962746,
+      "loss": 1.5234,
+      "step": 6531
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007876842777547714,
+      "loss": 1.5078,
+      "step": 6532
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007874027156407334,
+      "loss": 1.6523,
+      "step": 6533
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007871211711775379,
+      "loss": 1.5957,
+      "step": 6534
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007868396443885593,
+      "loss": 1.6289,
+      "step": 6535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000786558135297172,
+      "loss": 1.498,
+      "step": 6536
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000786276643926748,
+      "loss": 1.707,
+      "step": 6537
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007859951703006581,
+      "loss": 1.6738,
+      "step": 6538
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007857137144422708,
+      "loss": 1.5977,
+      "step": 6539
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007854322763749551,
+      "loss": 1.5977,
+      "step": 6540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007851508561220769,
+      "loss": 1.5547,
+      "step": 6541
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007848694537070009,
+      "loss": 1.6445,
+      "step": 6542
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007845880691530901,
+      "loss": 1.4844,
+      "step": 6543
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007843067024837074,
+      "loss": 1.6641,
+      "step": 6544
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007840253537222125,
+      "loss": 1.6133,
+      "step": 6545
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007837440228919645,
+      "loss": 1.5176,
+      "step": 6546
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007834627100163204,
+      "loss": 1.5078,
+      "step": 6547
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007831814151186369,
+      "loss": 1.5156,
+      "step": 6548
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007829001382222682,
+      "loss": 1.5586,
+      "step": 6549
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007826188793505671,
+      "loss": 1.5566,
+      "step": 6550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007823376385268848,
+      "loss": 1.5684,
+      "step": 6551
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007820564157745717,
+      "loss": 1.6953,
+      "step": 6552
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007817752111169761,
+      "loss": 1.5664,
+      "step": 6553
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007814940245774449,
+      "loss": 1.6113,
+      "step": 6554
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007812128561793235,
+      "loss": 1.5898,
+      "step": 6555
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007809317059459562,
+      "loss": 1.666,
+      "step": 6556
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000780650573900685,
+      "loss": 1.707,
+      "step": 6557
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007803694600668513,
+      "loss": 1.5996,
+      "step": 6558
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007800883644677936,
+      "loss": 1.5859,
+      "step": 6559
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007798072871268509,
+      "loss": 1.668,
+      "step": 6560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007795262280673591,
+      "loss": 1.6074,
+      "step": 6561
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000779245187312653,
+      "loss": 1.6543,
+      "step": 6562
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007789641648860657,
+      "loss": 1.4141,
+      "step": 6563
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007786831608109297,
+      "loss": 1.6562,
+      "step": 6564
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000778402175110575,
+      "loss": 1.5449,
+      "step": 6565
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007781212078083302,
+      "loss": 1.4883,
+      "step": 6566
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007778402589275223,
+      "loss": 1.582,
+      "step": 6567
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007775593284914778,
+      "loss": 1.6328,
+      "step": 6568
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007772784165235205,
+      "loss": 1.4941,
+      "step": 6569
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000776997523046973,
+      "loss": 1.6328,
+      "step": 6570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007767166480851562,
+      "loss": 1.5938,
+      "step": 6571
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007764357916613899,
+      "loss": 1.6113,
+      "step": 6572
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007761549537989924,
+      "loss": 1.4844,
+      "step": 6573
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007758741345212799,
+      "loss": 1.5488,
+      "step": 6574
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0007755933338515673,
+      "loss": 1.6191,
+      "step": 6575
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007753125518131683,
+      "loss": 1.6406,
+      "step": 6576
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007750317884293947,
+      "loss": 1.6914,
+      "step": 6577
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007747510437235565,
+      "loss": 1.6621,
+      "step": 6578
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007744703177189624,
+      "loss": 1.6348,
+      "step": 6579
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007741896104389203,
+      "loss": 1.6719,
+      "step": 6580
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007739089219067355,
+      "loss": 1.5293,
+      "step": 6581
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007736282521457121,
+      "loss": 1.6699,
+      "step": 6582
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000773347601179152,
+      "loss": 1.5098,
+      "step": 6583
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007730669690303575,
+      "loss": 1.6445,
+      "step": 6584
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000772786355722627,
+      "loss": 1.6113,
+      "step": 6585
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007725057612792589,
+      "loss": 1.5449,
+      "step": 6586
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007722251857235487,
+      "loss": 1.6055,
+      "step": 6587
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007719446290787922,
+      "loss": 1.6777,
+      "step": 6588
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007716640913682819,
+      "loss": 1.4824,
+      "step": 6589
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007713835726153096,
+      "loss": 1.6699,
+      "step": 6590
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000771103072843165,
+      "loss": 1.752,
+      "step": 6591
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007708225920751366,
+      "loss": 1.5508,
+      "step": 6592
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007705421303345118,
+      "loss": 1.7344,
+      "step": 6593
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007702616876445752,
+      "loss": 1.5254,
+      "step": 6594
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000769981264028611,
+      "loss": 1.6504,
+      "step": 6595
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007697008595099011,
+      "loss": 1.6562,
+      "step": 6596
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000769420474111726,
+      "loss": 1.6523,
+      "step": 6597
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007691401078573642,
+      "loss": 1.5312,
+      "step": 6598
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007688597607700939,
+      "loss": 1.7168,
+      "step": 6599
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007685794328731904,
+      "loss": 1.5859,
+      "step": 6600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007682991241899279,
+      "loss": 1.5625,
+      "step": 6601
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007680188347435787,
+      "loss": 1.5684,
+      "step": 6602
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007677385645574143,
+      "loss": 1.6562,
+      "step": 6603
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007674583136547038,
+      "loss": 1.623,
+      "step": 6604
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000767178082058715,
+      "loss": 1.6094,
+      "step": 6605
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007668978697927137,
+      "loss": 1.7148,
+      "step": 6606
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007666176768799651,
+      "loss": 1.5137,
+      "step": 6607
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007663375033437319,
+      "loss": 1.502,
+      "step": 6608
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007660573492072754,
+      "loss": 1.4668,
+      "step": 6609
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007657772144938551,
+      "loss": 1.6719,
+      "step": 6610
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007654970992267297,
+      "loss": 1.5645,
+      "step": 6611
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007652170034291549,
+      "loss": 1.6602,
+      "step": 6612
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007649369271243864,
+      "loss": 1.5781,
+      "step": 6613
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007646568703356769,
+      "loss": 1.6621,
+      "step": 6614
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007643768330862786,
+      "loss": 1.6094,
+      "step": 6615
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007640968153994411,
+      "loss": 1.5977,
+      "step": 6616
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007638168172984129,
+      "loss": 1.502,
+      "step": 6617
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007635368388064403,
+      "loss": 1.6484,
+      "step": 6618
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007632568799467696,
+      "loss": 1.623,
+      "step": 6619
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007629769407426435,
+      "loss": 1.5703,
+      "step": 6620
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007626970212173041,
+      "loss": 1.5605,
+      "step": 6621
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007624171213939911,
+      "loss": 1.5605,
+      "step": 6622
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007621372412959442,
+      "loss": 1.6074,
+      "step": 6623
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007618573809463998,
+      "loss": 1.5762,
+      "step": 6624
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007615775403685931,
+      "loss": 1.5684,
+      "step": 6625
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007612977195857577,
+      "loss": 1.5977,
+      "step": 6626
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007610179186211263,
+      "loss": 1.4355,
+      "step": 6627
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000760738137497929,
+      "loss": 1.4941,
+      "step": 6628
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007604583762393944,
+      "loss": 1.666,
+      "step": 6629
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007601786348687495,
+      "loss": 1.625,
+      "step": 6630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007598989134092203,
+      "loss": 1.5996,
+      "step": 6631
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00075961921188403,
+      "loss": 1.6543,
+      "step": 6632
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007593395303164012,
+      "loss": 1.5996,
+      "step": 6633
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007590598687295541,
+      "loss": 1.5508,
+      "step": 6634
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007587802271467079,
+      "loss": 1.541,
+      "step": 6635
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007585006055910796,
+      "loss": 1.6387,
+      "step": 6636
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007582210040858845,
+      "loss": 1.623,
+      "step": 6637
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007579414226543365,
+      "loss": 1.5918,
+      "step": 6638
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007576618613196482,
+      "loss": 1.5137,
+      "step": 6639
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007573823201050299,
+      "loss": 1.4922,
+      "step": 6640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007571027990336903,
+      "loss": 1.6836,
+      "step": 6641
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007568232981288363,
+      "loss": 1.5742,
+      "step": 6642
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007565438174136742,
+      "loss": 1.6152,
+      "step": 6643
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007562643569114074,
+      "loss": 1.6348,
+      "step": 6644
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007559849166452379,
+      "loss": 1.6426,
+      "step": 6645
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000755705496638366,
+      "loss": 1.6504,
+      "step": 6646
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007554260969139912,
+      "loss": 1.7012,
+      "step": 6647
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007551467174953102,
+      "loss": 1.6367,
+      "step": 6648
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007548673584055184,
+      "loss": 1.7227,
+      "step": 6649
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007545880196678093,
+      "loss": 1.457,
+      "step": 6650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007543087013053751,
+      "loss": 1.543,
+      "step": 6651
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007540294033414064,
+      "loss": 1.6895,
+      "step": 6652
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007537501257990917,
+      "loss": 1.5977,
+      "step": 6653
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007534708687016178,
+      "loss": 1.5762,
+      "step": 6654
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007531916320721703,
+      "loss": 1.5215,
+      "step": 6655
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007529124159339324,
+      "loss": 1.4668,
+      "step": 6656
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007526332203100861,
+      "loss": 1.6348,
+      "step": 6657
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007523540452238111,
+      "loss": 1.5742,
+      "step": 6658
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007520748906982868,
+      "loss": 1.5293,
+      "step": 6659
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007517957567566893,
+      "loss": 1.5059,
+      "step": 6660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007515166434221938,
+      "loss": 1.623,
+      "step": 6661
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007512375507179733,
+      "loss": 1.6445,
+      "step": 6662
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007509584786672,
+      "loss": 1.4355,
+      "step": 6663
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007506794272930435,
+      "loss": 1.5957,
+      "step": 6664
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007504003966186721,
+      "loss": 1.7168,
+      "step": 6665
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007501213866672516,
+      "loss": 1.6074,
+      "step": 6666
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007498423974619477,
+      "loss": 1.5996,
+      "step": 6667
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000749563429025923,
+      "loss": 1.5859,
+      "step": 6668
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007492844813823387,
+      "loss": 1.6602,
+      "step": 6669
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007490055545543543,
+      "loss": 1.5078,
+      "step": 6670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007487266485651278,
+      "loss": 1.709,
+      "step": 6671
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007484477634378155,
+      "loss": 1.4707,
+      "step": 6672
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007481688991955715,
+      "loss": 1.5469,
+      "step": 6673
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007478900558615483,
+      "loss": 1.541,
+      "step": 6674
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007476112334588974,
+      "loss": 1.6387,
+      "step": 6675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007473324320107673,
+      "loss": 1.6777,
+      "step": 6676
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007470536515403054,
+      "loss": 1.5547,
+      "step": 6677
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007467748920706583,
+      "loss": 1.7363,
+      "step": 6678
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000746496153624969,
+      "loss": 1.627,
+      "step": 6679
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007462174362263803,
+      "loss": 1.6113,
+      "step": 6680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000745938739898032,
+      "loss": 1.4277,
+      "step": 6681
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007456600646630633,
+      "loss": 1.6973,
+      "step": 6682
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007453814105446113,
+      "loss": 1.6367,
+      "step": 6683
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007451027775658108,
+      "loss": 1.5039,
+      "step": 6684
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000744824165749795,
+      "loss": 1.5332,
+      "step": 6685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007445455751196964,
+      "loss": 1.5801,
+      "step": 6686
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007442670056986445,
+      "loss": 1.7109,
+      "step": 6687
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0007439884575097674,
+      "loss": 1.6387,
+      "step": 6688
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007437099305761914,
+      "loss": 1.623,
+      "step": 6689
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007434314249210414,
+      "loss": 1.668,
+      "step": 6690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00074315294056744,
+      "loss": 1.6172,
+      "step": 6691
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007428744775385085,
+      "loss": 1.6562,
+      "step": 6692
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007425960358573662,
+      "loss": 1.6133,
+      "step": 6693
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007423176155471308,
+      "loss": 1.541,
+      "step": 6694
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007420392166309178,
+      "loss": 1.5664,
+      "step": 6695
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007417608391318415,
+      "loss": 1.5879,
+      "step": 6696
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007414824830730135,
+      "loss": 1.5234,
+      "step": 6697
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007412041484775451,
+      "loss": 1.6797,
+      "step": 6698
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007409258353685447,
+      "loss": 1.7012,
+      "step": 6699
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007406475437691188,
+      "loss": 1.6113,
+      "step": 6700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007403692737023725,
+      "loss": 1.5645,
+      "step": 6701
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007400910251914098,
+      "loss": 1.6152,
+      "step": 6702
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007398127982593319,
+      "loss": 1.5801,
+      "step": 6703
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007395345929292383,
+      "loss": 1.6289,
+      "step": 6704
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007392564092242266,
+      "loss": 1.7617,
+      "step": 6705
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007389782471673938,
+      "loss": 1.5527,
+      "step": 6706
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007387001067818339,
+      "loss": 1.5801,
+      "step": 6707
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007384219880906395,
+      "loss": 1.7109,
+      "step": 6708
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000738143891116901,
+      "loss": 1.541,
+      "step": 6709
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007378658158837074,
+      "loss": 1.5273,
+      "step": 6710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007375877624141463,
+      "loss": 1.5605,
+      "step": 6711
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007373097307313028,
+      "loss": 1.6445,
+      "step": 6712
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00073703172085826,
+      "loss": 1.5957,
+      "step": 6713
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007367537328181002,
+      "loss": 1.5879,
+      "step": 6714
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007364757666339032,
+      "loss": 1.5176,
+      "step": 6715
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007361978223287466,
+      "loss": 1.6152,
+      "step": 6716
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007359198999257069,
+      "loss": 1.6934,
+      "step": 6717
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007356419994478589,
+      "loss": 1.6348,
+      "step": 6718
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007353641209182751,
+      "loss": 1.6211,
+      "step": 6719
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000735086264360026,
+      "loss": 1.5078,
+      "step": 6720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007348084297961805,
+      "loss": 1.6973,
+      "step": 6721
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007345306172498063,
+      "loss": 1.6328,
+      "step": 6722
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007342528267439685,
+      "loss": 1.5469,
+      "step": 6723
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007339750583017304,
+      "loss": 1.7539,
+      "step": 6724
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007336973119461535,
+      "loss": 1.6602,
+      "step": 6725
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007334195877002983,
+      "loss": 1.6699,
+      "step": 6726
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007331418855872222,
+      "loss": 1.6074,
+      "step": 6727
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007328642056299819,
+      "loss": 1.5117,
+      "step": 6728
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007325865478516309,
+      "loss": 1.5352,
+      "step": 6729
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007323089122752225,
+      "loss": 1.4961,
+      "step": 6730
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007320312989238068,
+      "loss": 1.5762,
+      "step": 6731
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007317537078204329,
+      "loss": 1.543,
+      "step": 6732
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007314761389881475,
+      "loss": 1.5977,
+      "step": 6733
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000731198592449996,
+      "loss": 1.6211,
+      "step": 6734
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007309210682290212,
+      "loss": 1.5078,
+      "step": 6735
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007306435663482651,
+      "loss": 1.6973,
+      "step": 6736
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007303660868307662,
+      "loss": 1.7148,
+      "step": 6737
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007300886296995633,
+      "loss": 1.5293,
+      "step": 6738
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007298111949776918,
+      "loss": 1.6445,
+      "step": 6739
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007295337826881855,
+      "loss": 1.627,
+      "step": 6740
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007292563928540763,
+      "loss": 1.6211,
+      "step": 6741
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000728979025498395,
+      "loss": 1.625,
+      "step": 6742
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007287016806441698,
+      "loss": 1.3457,
+      "step": 6743
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007284243583144272,
+      "loss": 1.5547,
+      "step": 6744
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007281470585321911,
+      "loss": 1.6016,
+      "step": 6745
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007278697813204853,
+      "loss": 1.6621,
+      "step": 6746
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007275925267023303,
+      "loss": 1.5293,
+      "step": 6747
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007273152947007451,
+      "loss": 1.5645,
+      "step": 6748
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007270380853387465,
+      "loss": 1.4922,
+      "step": 6749
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00072676089863935,
+      "loss": 1.6328,
+      "step": 6750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007264837346255692,
+      "loss": 1.6426,
+      "step": 6751
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007262065933204153,
+      "loss": 1.5957,
+      "step": 6752
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007259294747468979,
+      "loss": 1.5703,
+      "step": 6753
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007256523789280249,
+      "loss": 1.6543,
+      "step": 6754
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000725375305886802,
+      "loss": 1.5742,
+      "step": 6755
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007250982556462331,
+      "loss": 1.6738,
+      "step": 6756
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007248212282293197,
+      "loss": 1.5918,
+      "step": 6757
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007245442236590631,
+      "loss": 1.6055,
+      "step": 6758
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007242672419584609,
+      "loss": 1.5684,
+      "step": 6759
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007239902831505093,
+      "loss": 1.5371,
+      "step": 6760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007237133472582026,
+      "loss": 1.6133,
+      "step": 6761
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007234364343045341,
+      "loss": 1.6113,
+      "step": 6762
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000723159544312494,
+      "loss": 1.5898,
+      "step": 6763
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007228826773050705,
+      "loss": 1.5449,
+      "step": 6764
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007226058333052514,
+      "loss": 1.5195,
+      "step": 6765
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007223290123360214,
+      "loss": 1.5527,
+      "step": 6766
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007220522144203629,
+      "loss": 1.6035,
+      "step": 6767
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007217754395812573,
+      "loss": 1.5938,
+      "step": 6768
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007214986878416839,
+      "loss": 1.5762,
+      "step": 6769
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007212219592246198,
+      "loss": 1.6973,
+      "step": 6770
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007209452537530406,
+      "loss": 1.6094,
+      "step": 6771
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007206685714499191,
+      "loss": 1.7344,
+      "step": 6772
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007203919123382277,
+      "loss": 1.6113,
+      "step": 6773
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007201152764409352,
+      "loss": 1.6465,
+      "step": 6774
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007198386637810095,
+      "loss": 1.5527,
+      "step": 6775
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007195620743814159,
+      "loss": 1.5215,
+      "step": 6776
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007192855082651191,
+      "loss": 1.5605,
+      "step": 6777
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007190089654550802,
+      "loss": 1.6016,
+      "step": 6778
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007187324459742595,
+      "loss": 1.5742,
+      "step": 6779
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007184559498456143,
+      "loss": 1.5059,
+      "step": 6780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007181794770921013,
+      "loss": 1.7207,
+      "step": 6781
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007179030277366747,
+      "loss": 1.6445,
+      "step": 6782
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007176266018022861,
+      "loss": 1.6523,
+      "step": 6783
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007173501993118857,
+      "loss": 1.627,
+      "step": 6784
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007170738202884224,
+      "loss": 1.5488,
+      "step": 6785
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000716797464754842,
+      "loss": 1.7285,
+      "step": 6786
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000716521132734089,
+      "loss": 1.6504,
+      "step": 6787
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007162448242491055,
+      "loss": 1.5605,
+      "step": 6788
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007159685393228324,
+      "loss": 1.666,
+      "step": 6789
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000715692277978208,
+      "loss": 1.668,
+      "step": 6790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000715416040238169,
+      "loss": 1.6445,
+      "step": 6791
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007151398261256494,
+      "loss": 1.6719,
+      "step": 6792
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007148636356635827,
+      "loss": 1.6855,
+      "step": 6793
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000714587468874899,
+      "loss": 1.6367,
+      "step": 6794
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000714311325782527,
+      "loss": 1.5332,
+      "step": 6795
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007140352064093932,
+      "loss": 1.5801,
+      "step": 6796
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007137591107784229,
+      "loss": 1.6523,
+      "step": 6797
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007134830389125387,
+      "loss": 1.5039,
+      "step": 6798
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007132069908346614,
+      "loss": 1.4941,
+      "step": 6799
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0007129309665677092,
+      "loss": 1.6113,
+      "step": 6800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007126549661346,
+      "loss": 1.5586,
+      "step": 6801
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007123789895582482,
+      "loss": 1.4824,
+      "step": 6802
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007121030368615666,
+      "loss": 1.5059,
+      "step": 6803
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007118271080674658,
+      "loss": 1.5332,
+      "step": 6804
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007115512031988556,
+      "loss": 1.7539,
+      "step": 6805
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007112753222786423,
+      "loss": 1.582,
+      "step": 6806
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007109994653297312,
+      "loss": 1.5293,
+      "step": 6807
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007107236323750248,
+      "loss": 1.7109,
+      "step": 6808
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007104478234374243,
+      "loss": 1.5898,
+      "step": 6809
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007101720385398289,
+      "loss": 1.5352,
+      "step": 6810
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007098962777051355,
+      "loss": 1.6406,
+      "step": 6811
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007096205409562387,
+      "loss": 1.7129,
+      "step": 6812
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007093448283160317,
+      "loss": 1.5938,
+      "step": 6813
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007090691398074059,
+      "loss": 1.584,
+      "step": 6814
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007087934754532498,
+      "loss": 1.5039,
+      "step": 6815
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007085178352764501,
+      "loss": 1.6191,
+      "step": 6816
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007082422192998925,
+      "loss": 1.625,
+      "step": 6817
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007079666275464595,
+      "loss": 1.5586,
+      "step": 6818
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007076910600390325,
+      "loss": 1.5859,
+      "step": 6819
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007074155168004894,
+      "loss": 1.7168,
+      "step": 6820
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007071399978537083,
+      "loss": 1.7207,
+      "step": 6821
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007068645032215635,
+      "loss": 1.6797,
+      "step": 6822
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000706589032926928,
+      "loss": 1.625,
+      "step": 6823
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007063135869926722,
+      "loss": 1.4648,
+      "step": 6824
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007060381654416659,
+      "loss": 1.5488,
+      "step": 6825
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007057627682967753,
+      "loss": 1.5996,
+      "step": 6826
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007054873955808653,
+      "loss": 1.6465,
+      "step": 6827
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007052120473167985,
+      "loss": 1.5098,
+      "step": 6828
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007049367235274355,
+      "loss": 1.5938,
+      "step": 6829
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007046614242356358,
+      "loss": 1.5879,
+      "step": 6830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007043861494642554,
+      "loss": 1.4902,
+      "step": 6831
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007041108992361486,
+      "loss": 1.6172,
+      "step": 6832
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007038356735741689,
+      "loss": 1.5977,
+      "step": 6833
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007035604725011664,
+      "loss": 1.4648,
+      "step": 6834
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007032852960399895,
+      "loss": 1.6504,
+      "step": 6835
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007030101442134843,
+      "loss": 1.6484,
+      "step": 6836
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007027350170444961,
+      "loss": 1.5098,
+      "step": 6837
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007024599145558669,
+      "loss": 1.543,
+      "step": 6838
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007021848367704369,
+      "loss": 1.6504,
+      "step": 6839
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007019097837110441,
+      "loss": 1.5488,
+      "step": 6840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007016347554005254,
+      "loss": 1.6113,
+      "step": 6841
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007013597518617146,
+      "loss": 1.5195,
+      "step": 6842
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007010847731174439,
+      "loss": 1.5508,
+      "step": 6843
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007008098191905429,
+      "loss": 1.5703,
+      "step": 6844
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007005348901038403,
+      "loss": 1.584,
+      "step": 6845
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0007002599858801617,
+      "loss": 1.5605,
+      "step": 6846
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006999851065423308,
+      "loss": 1.5859,
+      "step": 6847
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006997102521131697,
+      "loss": 1.6816,
+      "step": 6848
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006994354226154978,
+      "loss": 1.75,
+      "step": 6849
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006991606180721332,
+      "loss": 1.6191,
+      "step": 6850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006988858385058909,
+      "loss": 1.5762,
+      "step": 6851
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000698611083939585,
+      "loss": 1.4355,
+      "step": 6852
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006983363543960268,
+      "loss": 1.6406,
+      "step": 6853
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006980616498980253,
+      "loss": 1.752,
+      "step": 6854
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006977869704683877,
+      "loss": 1.5996,
+      "step": 6855
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006975123161299199,
+      "loss": 1.6855,
+      "step": 6856
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006972376869054246,
+      "loss": 1.5566,
+      "step": 6857
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006969630828177029,
+      "loss": 1.7012,
+      "step": 6858
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006966885038895534,
+      "loss": 1.4062,
+      "step": 6859
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006964139501437736,
+      "loss": 1.5215,
+      "step": 6860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006961394216031578,
+      "loss": 1.5039,
+      "step": 6861
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006958649182904988,
+      "loss": 1.7383,
+      "step": 6862
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006955904402285868,
+      "loss": 1.6172,
+      "step": 6863
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000695315987440211,
+      "loss": 1.5586,
+      "step": 6864
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006950415599481575,
+      "loss": 1.6699,
+      "step": 6865
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006947671577752105,
+      "loss": 1.666,
+      "step": 6866
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006944927809441519,
+      "loss": 1.5742,
+      "step": 6867
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006942184294777622,
+      "loss": 1.6133,
+      "step": 6868
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006939441033988193,
+      "loss": 1.4258,
+      "step": 6869
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000693669802730099,
+      "loss": 1.543,
+      "step": 6870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006933955274943749,
+      "loss": 1.7617,
+      "step": 6871
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000693121277714419,
+      "loss": 1.6094,
+      "step": 6872
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006928470534130005,
+      "loss": 1.5664,
+      "step": 6873
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006925728546128871,
+      "loss": 1.6895,
+      "step": 6874
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006922986813368432,
+      "loss": 1.6367,
+      "step": 6875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006920245336076334,
+      "loss": 1.6953,
+      "step": 6876
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000691750411448018,
+      "loss": 1.5254,
+      "step": 6877
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006914763148807559,
+      "loss": 1.4355,
+      "step": 6878
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006912022439286036,
+      "loss": 1.5859,
+      "step": 6879
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006909281986143166,
+      "loss": 1.4785,
+      "step": 6880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006906541789606468,
+      "loss": 1.6191,
+      "step": 6881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006903801849903451,
+      "loss": 1.6113,
+      "step": 6882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000690106216726159,
+      "loss": 1.7051,
+      "step": 6883
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006898322741908356,
+      "loss": 1.6406,
+      "step": 6884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006895583574071185,
+      "loss": 1.5742,
+      "step": 6885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006892844663977496,
+      "loss": 1.498,
+      "step": 6886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006890106011854684,
+      "loss": 1.5527,
+      "step": 6887
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006887367617930127,
+      "loss": 1.6152,
+      "step": 6888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006884629482431181,
+      "loss": 1.5469,
+      "step": 6889
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006881891605585177,
+      "loss": 1.5781,
+      "step": 6890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006879153987619427,
+      "loss": 1.6758,
+      "step": 6891
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006876416628761224,
+      "loss": 1.582,
+      "step": 6892
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006873679529237833,
+      "loss": 1.6973,
+      "step": 6893
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006870942689276503,
+      "loss": 1.6523,
+      "step": 6894
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006868206109104454,
+      "loss": 1.6562,
+      "step": 6895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00068654697889489,
+      "loss": 1.6777,
+      "step": 6896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006862733729037017,
+      "loss": 1.582,
+      "step": 6897
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006859997929595965,
+      "loss": 1.5469,
+      "step": 6898
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006857262390852884,
+      "loss": 1.5488,
+      "step": 6899
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006854527113034894,
+      "loss": 1.5449,
+      "step": 6900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006851792096369089,
+      "loss": 1.625,
+      "step": 6901
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006849057341082545,
+      "loss": 1.6484,
+      "step": 6902
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006846322847402309,
+      "loss": 1.5117,
+      "step": 6903
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006843588615555418,
+      "loss": 1.7148,
+      "step": 6904
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006840854645768881,
+      "loss": 1.6797,
+      "step": 6905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006838120938269681,
+      "loss": 1.5371,
+      "step": 6906
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006835387493284784,
+      "loss": 1.6289,
+      "step": 6907
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006832654311041134,
+      "loss": 1.6699,
+      "step": 6908
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006829921391765657,
+      "loss": 1.5352,
+      "step": 6909
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006827188735685249,
+      "loss": 1.6777,
+      "step": 6910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006824456343026787,
+      "loss": 1.6875,
+      "step": 6911
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006821724214017131,
+      "loss": 1.4297,
+      "step": 6912
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0006818992348883115,
+      "loss": 1.6055,
+      "step": 6913
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006816260747851548,
+      "loss": 1.5664,
+      "step": 6914
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006813529411149221,
+      "loss": 1.6738,
+      "step": 6915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006810798339002906,
+      "loss": 1.6016,
+      "step": 6916
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000680806753163935,
+      "loss": 1.541,
+      "step": 6917
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006805336989285275,
+      "loss": 1.5859,
+      "step": 6918
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006802606712167379,
+      "loss": 1.5859,
+      "step": 6919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006799876700512354,
+      "loss": 1.6953,
+      "step": 6920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006797146954546852,
+      "loss": 1.5977,
+      "step": 6921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000679441747449751,
+      "loss": 1.6016,
+      "step": 6922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006791688260590938,
+      "loss": 1.666,
+      "step": 6923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006788959313053737,
+      "loss": 1.582,
+      "step": 6924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006786230632112474,
+      "loss": 1.6836,
+      "step": 6925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006783502217993695,
+      "loss": 1.584,
+      "step": 6926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006780774070923925,
+      "loss": 1.5645,
+      "step": 6927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006778046191129673,
+      "loss": 1.7266,
+      "step": 6928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006775318578837418,
+      "loss": 1.6602,
+      "step": 6929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006772591234273619,
+      "loss": 1.6836,
+      "step": 6930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006769864157664712,
+      "loss": 1.5176,
+      "step": 6931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006767137349237115,
+      "loss": 1.502,
+      "step": 6932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006764410809217219,
+      "loss": 1.5273,
+      "step": 6933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000676168453783139,
+      "loss": 1.4668,
+      "step": 6934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006758958535305984,
+      "loss": 1.5176,
+      "step": 6935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006756232801867325,
+      "loss": 1.4531,
+      "step": 6936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006753507337741713,
+      "loss": 1.6875,
+      "step": 6937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006750782143155426,
+      "loss": 1.5566,
+      "step": 6938
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006748057218334732,
+      "loss": 1.6172,
+      "step": 6939
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006745332563505862,
+      "loss": 1.5391,
+      "step": 6940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006742608178895031,
+      "loss": 1.5742,
+      "step": 6941
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006739884064728426,
+      "loss": 1.666,
+      "step": 6942
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006737160221232224,
+      "loss": 1.5371,
+      "step": 6943
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006734436648632567,
+      "loss": 1.543,
+      "step": 6944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006731713347155579,
+      "loss": 1.5586,
+      "step": 6945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006728990317027359,
+      "loss": 1.5898,
+      "step": 6946
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006726267558473991,
+      "loss": 1.459,
+      "step": 6947
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006723545071721528,
+      "loss": 1.5801,
+      "step": 6948
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006720822856996009,
+      "loss": 1.668,
+      "step": 6949
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006718100914523436,
+      "loss": 1.707,
+      "step": 6950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006715379244529807,
+      "loss": 1.5605,
+      "step": 6951
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006712657847241083,
+      "loss": 1.6055,
+      "step": 6952
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006709936722883209,
+      "loss": 1.4902,
+      "step": 6953
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00067072158716821,
+      "loss": 1.6836,
+      "step": 6954
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006704495293863664,
+      "loss": 1.625,
+      "step": 6955
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006701774989653771,
+      "loss": 1.6387,
+      "step": 6956
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006699054959278276,
+      "loss": 1.5957,
+      "step": 6957
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006696335202963002,
+      "loss": 1.5762,
+      "step": 6958
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006693615720933765,
+      "loss": 1.5938,
+      "step": 6959
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006690896513416347,
+      "loss": 1.7656,
+      "step": 6960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006688177580636507,
+      "loss": 1.4902,
+      "step": 6961
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006685458922819982,
+      "loss": 1.5645,
+      "step": 6962
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006682740540192493,
+      "loss": 1.459,
+      "step": 6963
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006680022432979734,
+      "loss": 1.6289,
+      "step": 6964
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000667730460140737,
+      "loss": 1.8066,
+      "step": 6965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006674587045701051,
+      "loss": 1.541,
+      "step": 6966
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00066718697660864,
+      "loss": 1.7539,
+      "step": 6967
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006669152762789021,
+      "loss": 1.625,
+      "step": 6968
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006666436036034493,
+      "loss": 1.6562,
+      "step": 6969
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006663719586048367,
+      "loss": 1.4238,
+      "step": 6970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006661003413056182,
+      "loss": 1.7441,
+      "step": 6971
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006658287517283443,
+      "loss": 1.6621,
+      "step": 6972
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006655571898955638,
+      "loss": 1.5664,
+      "step": 6973
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006652856558298227,
+      "loss": 1.5605,
+      "step": 6974
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006650141495536658,
+      "loss": 1.5762,
+      "step": 6975
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006647426710896345,
+      "loss": 1.5664,
+      "step": 6976
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006644712204602681,
+      "loss": 1.5742,
+      "step": 6977
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006641997976881035,
+      "loss": 1.5781,
+      "step": 6978
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006639284027956762,
+      "loss": 1.502,
+      "step": 6979
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006636570358055183,
+      "loss": 1.5957,
+      "step": 6980
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006633856967401601,
+      "loss": 1.6016,
+      "step": 6981
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006631143856221289,
+      "loss": 1.541,
+      "step": 6982
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006628431024739513,
+      "loss": 1.5293,
+      "step": 6983
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006625718473181499,
+      "loss": 1.6016,
+      "step": 6984
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006623006201772456,
+      "loss": 1.7539,
+      "step": 6985
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006620294210737569,
+      "loss": 1.5293,
+      "step": 6986
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006617582500302,
+      "loss": 1.5898,
+      "step": 6987
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006614871070690894,
+      "loss": 1.5938,
+      "step": 6988
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006612159922129362,
+      "loss": 1.4766,
+      "step": 6989
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006609449054842495,
+      "loss": 1.5684,
+      "step": 6990
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006606738469055367,
+      "loss": 1.7344,
+      "step": 6991
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006604028164993021,
+      "loss": 1.6387,
+      "step": 6992
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006601318142880479,
+      "loss": 1.5898,
+      "step": 6993
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006598608402942736,
+      "loss": 1.7051,
+      "step": 6994
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006595898945404777,
+      "loss": 1.6738,
+      "step": 6995
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000659318977049155,
+      "loss": 1.5527,
+      "step": 6996
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006590480878427982,
+      "loss": 1.6465,
+      "step": 6997
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006587772269438976,
+      "loss": 1.5742,
+      "step": 6998
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000658506394374942,
+      "loss": 1.7188,
+      "step": 6999
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006582355901584168,
+      "loss": 1.6445,
+      "step": 7000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006579648143168056,
+      "loss": 1.5039,
+      "step": 7001
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006576940668725891,
+      "loss": 1.627,
+      "step": 7002
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006574233478482467,
+      "loss": 1.5859,
+      "step": 7003
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006571526572662546,
+      "loss": 1.748,
+      "step": 7004
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006568819951490866,
+      "loss": 1.5566,
+      "step": 7005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006566113615192142,
+      "loss": 1.5742,
+      "step": 7006
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006563407563991071,
+      "loss": 1.7012,
+      "step": 7007
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006560701798112323,
+      "loss": 1.4922,
+      "step": 7008
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006557996317780541,
+      "loss": 1.6289,
+      "step": 7009
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006555291123220346,
+      "loss": 1.5957,
+      "step": 7010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000655258621465634,
+      "loss": 1.5957,
+      "step": 7011
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006549881592313094,
+      "loss": 1.5508,
+      "step": 7012
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000654717725641516,
+      "loss": 1.6914,
+      "step": 7013
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000654447320718706,
+      "loss": 1.5684,
+      "step": 7014
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006541769444853305,
+      "loss": 1.6426,
+      "step": 7015
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006539065969638374,
+      "loss": 1.6035,
+      "step": 7016
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006536362781766711,
+      "loss": 1.7051,
+      "step": 7017
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006533659881462763,
+      "loss": 1.7109,
+      "step": 7018
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000653095726895093,
+      "loss": 1.6602,
+      "step": 7019
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006528254944455597,
+      "loss": 1.6191,
+      "step": 7020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006525552908201119,
+      "loss": 1.5879,
+      "step": 7021
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000652285116041184,
+      "loss": 1.5605,
+      "step": 7022
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006520149701312068,
+      "loss": 1.6523,
+      "step": 7023
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006517448531126091,
+      "loss": 1.6289,
+      "step": 7024
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0006514747650078173,
+      "loss": 1.6152,
+      "step": 7025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006512047058392554,
+      "loss": 1.6562,
+      "step": 7026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000650934675629345,
+      "loss": 1.6172,
+      "step": 7027
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006506646744005056,
+      "loss": 1.7891,
+      "step": 7028
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006503947021751535,
+      "loss": 1.5273,
+      "step": 7029
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006501247589757035,
+      "loss": 1.4961,
+      "step": 7030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006498548448245673,
+      "loss": 1.582,
+      "step": 7031
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006495849597441547,
+      "loss": 1.6445,
+      "step": 7032
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006493151037568723,
+      "loss": 1.6562,
+      "step": 7033
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006490452768851256,
+      "loss": 1.5273,
+      "step": 7034
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006487754791513165,
+      "loss": 1.5938,
+      "step": 7035
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006485057105778449,
+      "loss": 1.6562,
+      "step": 7036
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006482359711871081,
+      "loss": 1.5996,
+      "step": 7037
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006479662610015016,
+      "loss": 1.5039,
+      "step": 7038
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006476965800434179,
+      "loss": 1.5586,
+      "step": 7039
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006474269283352472,
+      "loss": 1.4531,
+      "step": 7040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006471573058993766,
+      "loss": 1.5762,
+      "step": 7041
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006468877127581925,
+      "loss": 1.5449,
+      "step": 7042
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006466181489340774,
+      "loss": 1.5254,
+      "step": 7043
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006463486144494115,
+      "loss": 1.5957,
+      "step": 7044
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000646079109326573,
+      "loss": 1.6973,
+      "step": 7045
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006458096335879376,
+      "loss": 1.709,
+      "step": 7046
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006455401872558786,
+      "loss": 1.6562,
+      "step": 7047
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006452707703527665,
+      "loss": 1.623,
+      "step": 7048
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006450013829009694,
+      "loss": 1.5977,
+      "step": 7049
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006447320249228536,
+      "loss": 1.543,
+      "step": 7050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006444626964407822,
+      "loss": 1.584,
+      "step": 7051
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006441933974771162,
+      "loss": 1.5254,
+      "step": 7052
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006439241280542135,
+      "loss": 1.6582,
+      "step": 7053
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006436548881944313,
+      "loss": 1.5781,
+      "step": 7054
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006433856779201225,
+      "loss": 1.5879,
+      "step": 7055
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006431164972536383,
+      "loss": 1.5645,
+      "step": 7056
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006428473462173268,
+      "loss": 1.5918,
+      "step": 7057
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006425782248335355,
+      "loss": 1.6816,
+      "step": 7058
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006423091331246071,
+      "loss": 1.5566,
+      "step": 7059
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006420400711128833,
+      "loss": 1.4609,
+      "step": 7060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006417710388207026,
+      "loss": 1.5859,
+      "step": 7061
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006415020362704016,
+      "loss": 1.5898,
+      "step": 7062
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006412330634843143,
+      "loss": 1.5801,
+      "step": 7063
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000640964120484772,
+      "loss": 1.6543,
+      "step": 7064
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000640695207294103,
+      "loss": 1.5391,
+      "step": 7065
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006404263239346347,
+      "loss": 1.5547,
+      "step": 7066
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006401574704286907,
+      "loss": 1.5371,
+      "step": 7067
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006398886467985927,
+      "loss": 1.6133,
+      "step": 7068
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006396198530666592,
+      "loss": 1.5371,
+      "step": 7069
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006393510892552073,
+      "loss": 1.6152,
+      "step": 7070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006390823553865508,
+      "loss": 1.627,
+      "step": 7071
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006388136514830013,
+      "loss": 1.5996,
+      "step": 7072
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006385449775668675,
+      "loss": 1.5996,
+      "step": 7073
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006382763336604567,
+      "loss": 1.4512,
+      "step": 7074
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006380077197860728,
+      "loss": 1.543,
+      "step": 7075
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000637739135966017,
+      "loss": 1.6816,
+      "step": 7076
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006374705822225885,
+      "loss": 1.5938,
+      "step": 7077
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006372020585780845,
+      "loss": 1.6191,
+      "step": 7078
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006369335650547985,
+      "loss": 1.5645,
+      "step": 7079
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006366651016750225,
+      "loss": 1.6191,
+      "step": 7080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006363966684610451,
+      "loss": 1.7051,
+      "step": 7081
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006361282654351535,
+      "loss": 1.5176,
+      "step": 7082
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006358598926196317,
+      "loss": 1.6113,
+      "step": 7083
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000635591550036761,
+      "loss": 1.582,
+      "step": 7084
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006353232377088203,
+      "loss": 1.6387,
+      "step": 7085
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006350549556580868,
+      "loss": 1.5469,
+      "step": 7086
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006347867039068343,
+      "loss": 1.6094,
+      "step": 7087
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006345184824773344,
+      "loss": 1.5996,
+      "step": 7088
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006342502913918558,
+      "loss": 1.7148,
+      "step": 7089
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006339821306726652,
+      "loss": 1.5215,
+      "step": 7090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006337140003420268,
+      "loss": 1.4453,
+      "step": 7091
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006334459004222017,
+      "loss": 1.6348,
+      "step": 7092
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000633177830935449,
+      "loss": 1.6426,
+      "step": 7093
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006329097919040252,
+      "loss": 1.4434,
+      "step": 7094
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006326417833501842,
+      "loss": 1.6406,
+      "step": 7095
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006323738052961772,
+      "loss": 1.4688,
+      "step": 7096
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006321058577642526,
+      "loss": 1.6211,
+      "step": 7097
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006318379407766576,
+      "loss": 1.7246,
+      "step": 7098
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006315700543556356,
+      "loss": 1.6074,
+      "step": 7099
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006313021985234273,
+      "loss": 1.6113,
+      "step": 7100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006310343733022721,
+      "loss": 1.5938,
+      "step": 7101
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006307665787144057,
+      "loss": 1.5859,
+      "step": 7102
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000630498814782062,
+      "loss": 1.6738,
+      "step": 7103
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006302310815274715,
+      "loss": 1.5859,
+      "step": 7104
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006299633789728632,
+      "loss": 1.5898,
+      "step": 7105
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006296957071404627,
+      "loss": 1.6113,
+      "step": 7106
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006294280660524939,
+      "loss": 1.498,
+      "step": 7107
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006291604557311771,
+      "loss": 1.6133,
+      "step": 7108
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006288928761987309,
+      "loss": 1.6465,
+      "step": 7109
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000628625327477371,
+      "loss": 1.6328,
+      "step": 7110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006283578095893104,
+      "loss": 1.4941,
+      "step": 7111
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006280903225567594,
+      "loss": 1.6367,
+      "step": 7112
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000627822866401927,
+      "loss": 1.5605,
+      "step": 7113
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000627555441147018,
+      "loss": 1.4785,
+      "step": 7114
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006272880468142356,
+      "loss": 1.625,
+      "step": 7115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006270206834257796,
+      "loss": 1.6367,
+      "step": 7116
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006267533510038486,
+      "loss": 1.625,
+      "step": 7117
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006264860495706374,
+      "loss": 1.6504,
+      "step": 7118
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006262187791483387,
+      "loss": 1.7148,
+      "step": 7119
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006259515397591421,
+      "loss": 1.6426,
+      "step": 7120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006256843314252359,
+      "loss": 1.6836,
+      "step": 7121
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006254171541688047,
+      "loss": 1.6055,
+      "step": 7122
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006251500080120309,
+      "loss": 1.5215,
+      "step": 7123
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006248828929770937,
+      "loss": 1.6211,
+      "step": 7124
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006246158090861709,
+      "loss": 1.7031,
+      "step": 7125
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000624348756361437,
+      "loss": 1.543,
+      "step": 7126
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006240817348250638,
+      "loss": 1.5586,
+      "step": 7127
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006238147444992206,
+      "loss": 1.7559,
+      "step": 7128
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006235477854060745,
+      "loss": 1.5938,
+      "step": 7129
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006232808575677897,
+      "loss": 1.6797,
+      "step": 7130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006230139610065278,
+      "loss": 1.582,
+      "step": 7131
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006227470957444471,
+      "loss": 1.6504,
+      "step": 7132
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006224802618037054,
+      "loss": 1.5234,
+      "step": 7133
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006222134592064555,
+      "loss": 1.4453,
+      "step": 7134
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000621946687974849,
+      "loss": 1.4336,
+      "step": 7135
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006216799481310343,
+      "loss": 1.6113,
+      "step": 7136
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006214132396971577,
+      "loss": 1.627,
+      "step": 7137
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0006211465626953625,
+      "loss": 1.6094,
+      "step": 7138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006208799171477896,
+      "loss": 1.7324,
+      "step": 7139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006206133030765766,
+      "loss": 1.5312,
+      "step": 7140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006203467205038599,
+      "loss": 1.5254,
+      "step": 7141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006200801694517722,
+      "loss": 1.5508,
+      "step": 7142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006198136499424438,
+      "loss": 1.4844,
+      "step": 7143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006195471619980017,
+      "loss": 1.502,
+      "step": 7144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006192807056405724,
+      "loss": 1.7988,
+      "step": 7145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006190142808922774,
+      "loss": 1.5137,
+      "step": 7146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006187478877752369,
+      "loss": 1.6035,
+      "step": 7147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006184815263115678,
+      "loss": 1.4707,
+      "step": 7148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006182151965233851,
+      "loss": 1.6641,
+      "step": 7149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006179488984328008,
+      "loss": 1.5547,
+      "step": 7150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006176826320619238,
+      "loss": 1.5605,
+      "step": 7151
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006174163974328607,
+      "loss": 1.5918,
+      "step": 7152
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006171501945677163,
+      "loss": 1.6055,
+      "step": 7153
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006168840234885918,
+      "loss": 1.5508,
+      "step": 7154
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006166178842175858,
+      "loss": 1.6426,
+      "step": 7155
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006163517767767938,
+      "loss": 1.5801,
+      "step": 7156
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006160857011883108,
+      "loss": 1.5723,
+      "step": 7157
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006158196574742267,
+      "loss": 1.5664,
+      "step": 7158
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00061555364565663,
+      "loss": 1.627,
+      "step": 7159
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006152876657576057,
+      "loss": 1.6797,
+      "step": 7160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006150217177992377,
+      "loss": 1.5527,
+      "step": 7161
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006147558018036054,
+      "loss": 1.6504,
+      "step": 7162
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006144899177927872,
+      "loss": 1.623,
+      "step": 7163
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000614224065788857,
+      "loss": 1.5938,
+      "step": 7164
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006139582458138883,
+      "loss": 1.623,
+      "step": 7165
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006136924578899501,
+      "loss": 1.6309,
+      "step": 7166
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006134267020391095,
+      "loss": 1.6172,
+      "step": 7167
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006131609782834305,
+      "loss": 1.5703,
+      "step": 7168
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006128952866449752,
+      "loss": 1.5137,
+      "step": 7169
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006126296271458026,
+      "loss": 1.5605,
+      "step": 7170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006123639998079686,
+      "loss": 1.5527,
+      "step": 7171
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006120984046535272,
+      "loss": 1.5918,
+      "step": 7172
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006118328417045294,
+      "loss": 1.4629,
+      "step": 7173
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006115673109830233,
+      "loss": 1.4102,
+      "step": 7174
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006113018125110546,
+      "loss": 1.5566,
+      "step": 7175
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006110363463106659,
+      "loss": 1.623,
+      "step": 7176
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006107709124038984,
+      "loss": 1.5508,
+      "step": 7177
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000610505510812789,
+      "loss": 1.541,
+      "step": 7178
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006102401415593728,
+      "loss": 1.7383,
+      "step": 7179
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006099748046656814,
+      "loss": 1.6133,
+      "step": 7180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006097095001537455,
+      "loss": 1.6777,
+      "step": 7181
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006094442280455913,
+      "loss": 1.5996,
+      "step": 7182
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006091789883632431,
+      "loss": 1.6426,
+      "step": 7183
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006089137811287215,
+      "loss": 1.498,
+      "step": 7184
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006086486063640468,
+      "loss": 1.5742,
+      "step": 7185
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006083834640912343,
+      "loss": 1.5664,
+      "step": 7186
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006081183543322971,
+      "loss": 1.7207,
+      "step": 7187
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006078532771092464,
+      "loss": 1.6836,
+      "step": 7188
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006075882324440897,
+      "loss": 1.5547,
+      "step": 7189
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006073232203588329,
+      "loss": 1.6074,
+      "step": 7190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006070582408754777,
+      "loss": 1.5117,
+      "step": 7191
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006067932940160247,
+      "loss": 1.6641,
+      "step": 7192
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000606528379802471,
+      "loss": 1.6738,
+      "step": 7193
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006062634982568107,
+      "loss": 1.5957,
+      "step": 7194
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006059986494010354,
+      "loss": 1.7168,
+      "step": 7195
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006057338332571347,
+      "loss": 1.5488,
+      "step": 7196
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006054690498470947,
+      "loss": 1.5488,
+      "step": 7197
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006052042991928989,
+      "loss": 1.5469,
+      "step": 7198
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006049395813165277,
+      "loss": 1.543,
+      "step": 7199
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006046748962399601,
+      "loss": 1.5254,
+      "step": 7200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000604410243985171,
+      "loss": 1.6484,
+      "step": 7201
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006041456245741335,
+      "loss": 1.5605,
+      "step": 7202
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006038810380288167,
+      "loss": 1.6699,
+      "step": 7203
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006036164843711887,
+      "loss": 1.6367,
+      "step": 7204
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006033519636232139,
+      "loss": 1.4824,
+      "step": 7205
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006030874758068539,
+      "loss": 1.6699,
+      "step": 7206
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006028230209440675,
+      "loss": 1.5195,
+      "step": 7207
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006025585990568114,
+      "loss": 1.4277,
+      "step": 7208
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000602294210167039,
+      "loss": 1.5625,
+      "step": 7209
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006020298542967013,
+      "loss": 1.5527,
+      "step": 7210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006017655314677457,
+      "loss": 1.6152,
+      "step": 7211
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006015012417021186,
+      "loss": 1.4902,
+      "step": 7212
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006012369850217619,
+      "loss": 1.623,
+      "step": 7213
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006009727614486157,
+      "loss": 1.5,
+      "step": 7214
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006007085710046167,
+      "loss": 1.5723,
+      "step": 7215
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006004444137116998,
+      "loss": 1.543,
+      "step": 7216
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0006001802895917965,
+      "loss": 1.6191,
+      "step": 7217
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005999161986668355,
+      "loss": 1.582,
+      "step": 7218
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005996521409587426,
+      "loss": 1.5684,
+      "step": 7219
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000599388116489442,
+      "loss": 1.5039,
+      "step": 7220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005991241252808534,
+      "loss": 1.5352,
+      "step": 7221
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000598860167354895,
+      "loss": 1.5254,
+      "step": 7222
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005985962427334815,
+      "loss": 1.5859,
+      "step": 7223
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005983323514385259,
+      "loss": 1.5312,
+      "step": 7224
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005980684934919372,
+      "loss": 1.6191,
+      "step": 7225
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005978046689156223,
+      "loss": 1.6602,
+      "step": 7226
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005975408777314849,
+      "loss": 1.623,
+      "step": 7227
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005972771199614264,
+      "loss": 1.5859,
+      "step": 7228
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005970133956273453,
+      "loss": 1.6348,
+      "step": 7229
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005967497047511375,
+      "loss": 1.6797,
+      "step": 7230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000596486047354695,
+      "loss": 1.5293,
+      "step": 7231
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005962224234599087,
+      "loss": 1.4941,
+      "step": 7232
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005959588330886658,
+      "loss": 1.5684,
+      "step": 7233
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005956952762628507,
+      "loss": 1.4727,
+      "step": 7234
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005954317530043449,
+      "loss": 1.6309,
+      "step": 7235
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005951682633350278,
+      "loss": 1.7148,
+      "step": 7236
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005949048072767755,
+      "loss": 1.625,
+      "step": 7237
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005946413848514614,
+      "loss": 1.5,
+      "step": 7238
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005943779960809553,
+      "loss": 1.5703,
+      "step": 7239
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005941146409871263,
+      "loss": 1.5078,
+      "step": 7240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005938513195918386,
+      "loss": 1.6133,
+      "step": 7241
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005935880319169547,
+      "loss": 1.5723,
+      "step": 7242
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005933247779843333,
+      "loss": 1.5293,
+      "step": 7243
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005930615578158319,
+      "loss": 1.543,
+      "step": 7244
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005927983714333041,
+      "loss": 1.4727,
+      "step": 7245
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005925352188586007,
+      "loss": 1.5664,
+      "step": 7246
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005922721001135697,
+      "loss": 1.5723,
+      "step": 7247
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005920090152200569,
+      "loss": 1.5195,
+      "step": 7248
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005917459641999048,
+      "loss": 1.6035,
+      "step": 7249
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0005914829470749526,
+      "loss": 1.6289,
+      "step": 7250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005912199638670376,
+      "loss": 1.6289,
+      "step": 7251
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005909570145979945,
+      "loss": 1.5605,
+      "step": 7252
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005906940992896539,
+      "loss": 1.668,
+      "step": 7253
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005904312179638444,
+      "loss": 1.75,
+      "step": 7254
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005901683706423915,
+      "loss": 1.5469,
+      "step": 7255
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005899055573471185,
+      "loss": 1.6152,
+      "step": 7256
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005896427780998452,
+      "loss": 1.6133,
+      "step": 7257
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005893800329223888,
+      "loss": 1.6309,
+      "step": 7258
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005891173218365633,
+      "loss": 1.6465,
+      "step": 7259
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005888546448641809,
+      "loss": 1.7188,
+      "step": 7260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005885920020270501,
+      "loss": 1.6113,
+      "step": 7261
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005883293933469765,
+      "loss": 1.5723,
+      "step": 7262
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005880668188457627,
+      "loss": 1.6836,
+      "step": 7263
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00058780427854521,
+      "loss": 1.7109,
+      "step": 7264
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005875417724671152,
+      "loss": 1.4805,
+      "step": 7265
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005872793006332728,
+      "loss": 1.5137,
+      "step": 7266
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005870168630654742,
+      "loss": 1.5332,
+      "step": 7267
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005867544597855088,
+      "loss": 1.6895,
+      "step": 7268
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005864920908151621,
+      "loss": 1.4395,
+      "step": 7269
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005862297561762171,
+      "loss": 1.5742,
+      "step": 7270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005859674558904547,
+      "loss": 1.5527,
+      "step": 7271
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000585705189979652,
+      "loss": 1.5586,
+      "step": 7272
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005854429584655835,
+      "loss": 1.5449,
+      "step": 7273
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005851807613700207,
+      "loss": 1.5742,
+      "step": 7274
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005849185987147329,
+      "loss": 1.625,
+      "step": 7275
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000584656470521486,
+      "loss": 1.582,
+      "step": 7276
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005843943768120431,
+      "loss": 1.5645,
+      "step": 7277
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000584132317608164,
+      "loss": 1.5234,
+      "step": 7278
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005838702929316068,
+      "loss": 1.7051,
+      "step": 7279
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000583608302804126,
+      "loss": 1.6426,
+      "step": 7280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000583346347247473,
+      "loss": 1.5254,
+      "step": 7281
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000583084426283396,
+      "loss": 1.6094,
+      "step": 7282
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005828225399336422,
+      "loss": 1.6582,
+      "step": 7283
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005825606882199539,
+      "loss": 1.5742,
+      "step": 7284
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005822988711640715,
+      "loss": 1.5781,
+      "step": 7285
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005820370887877317,
+      "loss": 1.5332,
+      "step": 7286
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005817753411126701,
+      "loss": 1.6543,
+      "step": 7287
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005815136281606172,
+      "loss": 1.6406,
+      "step": 7288
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005812519499533023,
+      "loss": 1.7051,
+      "step": 7289
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005809903065124505,
+      "loss": 1.5137,
+      "step": 7290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005807286978597854,
+      "loss": 1.5938,
+      "step": 7291
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005804671240170268,
+      "loss": 1.4609,
+      "step": 7292
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005802055850058915,
+      "loss": 1.6309,
+      "step": 7293
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000579944080848094,
+      "loss": 1.7012,
+      "step": 7294
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005796826115653458,
+      "loss": 1.5742,
+      "step": 7295
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005794211771793543,
+      "loss": 1.6309,
+      "step": 7296
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005791597777118265,
+      "loss": 1.4434,
+      "step": 7297
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005788984131844641,
+      "loss": 1.6855,
+      "step": 7298
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005786370836189672,
+      "loss": 1.541,
+      "step": 7299
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005783757890370325,
+      "loss": 1.6367,
+      "step": 7300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000578114529460354,
+      "loss": 1.8008,
+      "step": 7301
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005778533049106223,
+      "loss": 1.6602,
+      "step": 7302
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000577592115409526,
+      "loss": 1.6523,
+      "step": 7303
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005773309609787504,
+      "loss": 1.5234,
+      "step": 7304
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005770698416399774,
+      "loss": 1.4453,
+      "step": 7305
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005768087574148862,
+      "loss": 1.6367,
+      "step": 7306
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005765477083251539,
+      "loss": 1.5703,
+      "step": 7307
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005762866943924538,
+      "loss": 1.5039,
+      "step": 7308
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005760257156384563,
+      "loss": 1.5527,
+      "step": 7309
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000575764772084829,
+      "loss": 1.4766,
+      "step": 7310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005755038637532372,
+      "loss": 1.5645,
+      "step": 7311
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005752429906653427,
+      "loss": 1.5996,
+      "step": 7312
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005749821528428041,
+      "loss": 1.543,
+      "step": 7313
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000574721350307277,
+      "loss": 1.5312,
+      "step": 7314
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005744605830804155,
+      "loss": 1.6133,
+      "step": 7315
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005741998511838693,
+      "loss": 1.666,
+      "step": 7316
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005739391546392854,
+      "loss": 1.4902,
+      "step": 7317
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005736784934683079,
+      "loss": 1.5664,
+      "step": 7318
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005734178676925787,
+      "loss": 1.4258,
+      "step": 7319
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005731572773337361,
+      "loss": 1.498,
+      "step": 7320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005728967224134154,
+      "loss": 1.584,
+      "step": 7321
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005726362029532486,
+      "loss": 1.4785,
+      "step": 7322
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005723757189748662,
+      "loss": 1.5488,
+      "step": 7323
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005721152704998944,
+      "loss": 1.582,
+      "step": 7324
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000571854857549957,
+      "loss": 1.5469,
+      "step": 7325
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005715944801466742,
+      "loss": 1.6895,
+      "step": 7326
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005713341383116646,
+      "loss": 1.5781,
+      "step": 7327
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005710738320665425,
+      "loss": 1.6367,
+      "step": 7328
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005708135614329199,
+      "loss": 1.5723,
+      "step": 7329
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005705533264324052,
+      "loss": 1.6113,
+      "step": 7330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005702931270866054,
+      "loss": 1.6602,
+      "step": 7331
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005700329634171228,
+      "loss": 1.5527,
+      "step": 7332
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005697728354455578,
+      "loss": 1.6191,
+      "step": 7333
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005695127431935069,
+      "loss": 1.4648,
+      "step": 7334
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005692526866825645,
+      "loss": 1.4512,
+      "step": 7335
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005689926659343218,
+      "loss": 1.4863,
+      "step": 7336
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005687326809703672,
+      "loss": 1.7461,
+      "step": 7337
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005684727318122856,
+      "loss": 1.5859,
+      "step": 7338
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005682128184816592,
+      "loss": 1.6816,
+      "step": 7339
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005679529410000672,
+      "loss": 1.6152,
+      "step": 7340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005676930993890861,
+      "loss": 1.6992,
+      "step": 7341
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005674332936702886,
+      "loss": 1.5488,
+      "step": 7342
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005671735238652458,
+      "loss": 1.5879,
+      "step": 7343
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005669137899955248,
+      "loss": 1.4531,
+      "step": 7344
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005666540920826898,
+      "loss": 1.6582,
+      "step": 7345
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005663944301483018,
+      "loss": 1.5449,
+      "step": 7346
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005661348042139199,
+      "loss": 1.5078,
+      "step": 7347
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005658752143010993,
+      "loss": 1.6602,
+      "step": 7348
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005656156604313922,
+      "loss": 1.6387,
+      "step": 7349
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005653561426263476,
+      "loss": 1.623,
+      "step": 7350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000565096660907513,
+      "loss": 1.5742,
+      "step": 7351
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000564837215296431,
+      "loss": 1.7109,
+      "step": 7352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005645778058146421,
+      "loss": 1.5938,
+      "step": 7353
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005643184324836836,
+      "loss": 1.502,
+      "step": 7354
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005640590953250904,
+      "loss": 1.5586,
+      "step": 7355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005637997943603937,
+      "loss": 1.6309,
+      "step": 7356
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005635405296111213,
+      "loss": 1.4121,
+      "step": 7357
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005632813010987998,
+      "loss": 1.5449,
+      "step": 7358
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005630221088449508,
+      "loss": 1.5332,
+      "step": 7359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005627629528710936,
+      "loss": 1.5859,
+      "step": 7360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005625038331987447,
+      "loss": 1.6719,
+      "step": 7361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0005622447498494176,
+      "loss": 1.4688,
+      "step": 7362
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005619857028446228,
+      "loss": 1.5527,
+      "step": 7363
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005617266922058673,
+      "loss": 1.623,
+      "step": 7364
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000561467717954655,
+      "loss": 1.6602,
+      "step": 7365
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005612087801124882,
+      "loss": 1.6914,
+      "step": 7366
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005609498787008646,
+      "loss": 1.5723,
+      "step": 7367
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005606910137412794,
+      "loss": 1.5527,
+      "step": 7368
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005604321852552244,
+      "loss": 1.5684,
+      "step": 7369
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005601733932641897,
+      "loss": 1.6172,
+      "step": 7370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000559914637789661,
+      "loss": 1.5742,
+      "step": 7371
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005596559188531214,
+      "loss": 1.5977,
+      "step": 7372
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005593972364760509,
+      "loss": 1.6191,
+      "step": 7373
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005591385906799267,
+      "loss": 1.4785,
+      "step": 7374
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005588799814862223,
+      "loss": 1.5137,
+      "step": 7375
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005586214089164094,
+      "loss": 1.6445,
+      "step": 7376
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005583628729919559,
+      "loss": 1.6152,
+      "step": 7377
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005581043737343262,
+      "loss": 1.6758,
+      "step": 7378
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005578459111649825,
+      "loss": 1.6035,
+      "step": 7379
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005575874853053832,
+      "loss": 1.6328,
+      "step": 7380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005573290961769842,
+      "loss": 1.582,
+      "step": 7381
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005570707438012387,
+      "loss": 1.582,
+      "step": 7382
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005568124281995959,
+      "loss": 1.5703,
+      "step": 7383
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005565541493935024,
+      "loss": 1.6465,
+      "step": 7384
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005562959074044013,
+      "loss": 1.6855,
+      "step": 7385
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005560377022537342,
+      "loss": 1.4629,
+      "step": 7386
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005557795339629378,
+      "loss": 1.5879,
+      "step": 7387
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005555214025534465,
+      "loss": 1.5801,
+      "step": 7388
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005552633080466913,
+      "loss": 1.5215,
+      "step": 7389
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005550052504641013,
+      "loss": 1.5117,
+      "step": 7390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005547472298271011,
+      "loss": 1.4922,
+      "step": 7391
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005544892461571129,
+      "loss": 1.7559,
+      "step": 7392
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005542312994755554,
+      "loss": 1.6035,
+      "step": 7393
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005539733898038453,
+      "loss": 1.5859,
+      "step": 7394
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000553715517163395,
+      "loss": 1.4941,
+      "step": 7395
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005534576815756143,
+      "loss": 1.5859,
+      "step": 7396
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005531998830619098,
+      "loss": 1.6816,
+      "step": 7397
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005529421216436859,
+      "loss": 1.6211,
+      "step": 7398
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005526843973423425,
+      "loss": 1.5527,
+      "step": 7399
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005524267101792773,
+      "loss": 1.5898,
+      "step": 7400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005521690601758844,
+      "loss": 1.6699,
+      "step": 7401
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005519114473535559,
+      "loss": 1.5176,
+      "step": 7402
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005516538717336795,
+      "loss": 1.6055,
+      "step": 7403
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005513963333376406,
+      "loss": 1.5859,
+      "step": 7404
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005511388321868205,
+      "loss": 1.5137,
+      "step": 7405
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005508813683025993,
+      "loss": 1.6621,
+      "step": 7406
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005506239417063524,
+      "loss": 1.6777,
+      "step": 7407
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005503665524194526,
+      "loss": 1.6758,
+      "step": 7408
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000550109200463269,
+      "loss": 1.543,
+      "step": 7409
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005498518858591693,
+      "loss": 1.4629,
+      "step": 7410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005495946086285164,
+      "loss": 1.4805,
+      "step": 7411
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005493373687926707,
+      "loss": 1.668,
+      "step": 7412
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005490801663729896,
+      "loss": 1.75,
+      "step": 7413
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005488230013908266,
+      "loss": 1.6719,
+      "step": 7414
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005485658738675338,
+      "loss": 1.584,
+      "step": 7415
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005483087838244587,
+      "loss": 1.6309,
+      "step": 7416
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000548051731282946,
+      "loss": 1.5723,
+      "step": 7417
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005477947162643378,
+      "loss": 1.541,
+      "step": 7418
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005475377387899723,
+      "loss": 1.5293,
+      "step": 7419
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005472807988811852,
+      "loss": 1.4824,
+      "step": 7420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005470238965593086,
+      "loss": 1.6113,
+      "step": 7421
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005467670318456722,
+      "loss": 1.4941,
+      "step": 7422
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000546510204761602,
+      "loss": 1.5527,
+      "step": 7423
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005462534153284211,
+      "loss": 1.5879,
+      "step": 7424
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005459966635674487,
+      "loss": 1.4941,
+      "step": 7425
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005457399495000027,
+      "loss": 1.5801,
+      "step": 7426
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005454832731473959,
+      "loss": 1.6074,
+      "step": 7427
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005452266345309391,
+      "loss": 1.7285,
+      "step": 7428
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005449700336719392,
+      "loss": 1.7051,
+      "step": 7429
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005447134705917012,
+      "loss": 1.6777,
+      "step": 7430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005444569453115258,
+      "loss": 1.5645,
+      "step": 7431
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000544200457852711,
+      "loss": 1.4668,
+      "step": 7432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005439440082365512,
+      "loss": 1.5488,
+      "step": 7433
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005436875964843389,
+      "loss": 1.666,
+      "step": 7434
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005434312226173621,
+      "loss": 1.5605,
+      "step": 7435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005431748866569062,
+      "loss": 1.7051,
+      "step": 7436
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005429185886242532,
+      "loss": 1.5547,
+      "step": 7437
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005426623285406829,
+      "loss": 1.6543,
+      "step": 7438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005424061064274708,
+      "loss": 1.5801,
+      "step": 7439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005421499223058892,
+      "loss": 1.5762,
+      "step": 7440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005418937761972088,
+      "loss": 1.5723,
+      "step": 7441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005416376681226956,
+      "loss": 1.498,
+      "step": 7442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005413815981036125,
+      "loss": 1.5977,
+      "step": 7443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005411255661612197,
+      "loss": 1.4941,
+      "step": 7444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005408695723167751,
+      "loss": 1.6465,
+      "step": 7445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005406136165915316,
+      "loss": 1.6289,
+      "step": 7446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005403576990067405,
+      "loss": 1.5508,
+      "step": 7447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005401018195836483,
+      "loss": 1.541,
+      "step": 7448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005398459783435006,
+      "loss": 1.5703,
+      "step": 7449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005395901753075379,
+      "loss": 1.6113,
+      "step": 7450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005393344104969982,
+      "loss": 1.5488,
+      "step": 7451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005390786839331164,
+      "loss": 1.5156,
+      "step": 7452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005388229956371242,
+      "loss": 1.5762,
+      "step": 7453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005385673456302496,
+      "loss": 1.4805,
+      "step": 7454
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005383117339337185,
+      "loss": 1.7168,
+      "step": 7455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005380561605687528,
+      "loss": 1.6074,
+      "step": 7456
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005378006255565714,
+      "loss": 1.6562,
+      "step": 7457
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005375451289183901,
+      "loss": 1.5664,
+      "step": 7458
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005372896706754215,
+      "loss": 1.6152,
+      "step": 7459
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005370342508488744,
+      "loss": 1.5469,
+      "step": 7460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005367788694599555,
+      "loss": 1.6719,
+      "step": 7461
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000536523526529868,
+      "loss": 1.6992,
+      "step": 7462
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005362682220798114,
+      "loss": 1.4824,
+      "step": 7463
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005360129561309819,
+      "loss": 1.666,
+      "step": 7464
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005357577287045736,
+      "loss": 1.5996,
+      "step": 7465
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005355025398217763,
+      "loss": 1.5332,
+      "step": 7466
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005352473895037773,
+      "loss": 1.5195,
+      "step": 7467
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005349922777717596,
+      "loss": 1.7422,
+      "step": 7468
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005347372046469048,
+      "loss": 1.5527,
+      "step": 7469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00053448217015039,
+      "loss": 1.5859,
+      "step": 7470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005342271743033891,
+      "loss": 1.6465,
+      "step": 7471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005339722171270729,
+      "loss": 1.5352,
+      "step": 7472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005337172986426098,
+      "loss": 1.5781,
+      "step": 7473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005334624188711642,
+      "loss": 1.6055,
+      "step": 7474
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0005332075778338972,
+      "loss": 1.666,
+      "step": 7475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005329527755519666,
+      "loss": 1.5762,
+      "step": 7476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005326980120465281,
+      "loss": 1.6309,
+      "step": 7477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000532443287338733,
+      "loss": 1.5723,
+      "step": 7478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005321886014497296,
+      "loss": 1.5469,
+      "step": 7479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005319339544006629,
+      "loss": 1.4863,
+      "step": 7480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005316793462126759,
+      "loss": 1.5176,
+      "step": 7481
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005314247769069067,
+      "loss": 1.5215,
+      "step": 7482
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005311702465044908,
+      "loss": 1.7051,
+      "step": 7483
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005309157550265604,
+      "loss": 1.6055,
+      "step": 7484
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005306613024942452,
+      "loss": 1.5312,
+      "step": 7485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005304068889286709,
+      "loss": 1.7148,
+      "step": 7486
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005301525143509598,
+      "loss": 1.6816,
+      "step": 7487
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005298981787822311,
+      "loss": 1.5527,
+      "step": 7488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005296438822436018,
+      "loss": 1.5254,
+      "step": 7489
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005293896247561842,
+      "loss": 1.6133,
+      "step": 7490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005291354063410882,
+      "loss": 1.582,
+      "step": 7491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005288812270194201,
+      "loss": 1.6797,
+      "step": 7492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005286270868122827,
+      "loss": 1.7324,
+      "step": 7493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005283729857407766,
+      "loss": 1.5332,
+      "step": 7494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005281189238259985,
+      "loss": 1.6523,
+      "step": 7495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005278649010890412,
+      "loss": 1.5527,
+      "step": 7496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005276109175509955,
+      "loss": 1.5879,
+      "step": 7497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005273569732329481,
+      "loss": 1.582,
+      "step": 7498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005271030681559825,
+      "loss": 1.6465,
+      "step": 7499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005268492023411789,
+      "loss": 1.457,
+      "step": 7500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005265953758096151,
+      "loss": 1.6543,
+      "step": 7501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005263415885823648,
+      "loss": 1.6172,
+      "step": 7502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005260878406804986,
+      "loss": 1.4785,
+      "step": 7503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005258341321250832,
+      "loss": 1.498,
+      "step": 7504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005255804629371839,
+      "loss": 1.4922,
+      "step": 7505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005253268331378607,
+      "loss": 1.6621,
+      "step": 7506
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005250732427481714,
+      "loss": 1.5566,
+      "step": 7507
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00052481969178917,
+      "loss": 1.5156,
+      "step": 7508
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005245661802819081,
+      "loss": 1.6621,
+      "step": 7509
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000524312708247433,
+      "loss": 1.5449,
+      "step": 7510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005240592757067893,
+      "loss": 1.625,
+      "step": 7511
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005238058826810178,
+      "loss": 1.6211,
+      "step": 7512
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005235525291911572,
+      "loss": 1.6016,
+      "step": 7513
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005232992152582414,
+      "loss": 1.5352,
+      "step": 7514
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005230459409033022,
+      "loss": 1.5996,
+      "step": 7515
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005227927061473669,
+      "loss": 1.6562,
+      "step": 7516
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005225395110114615,
+      "loss": 1.6543,
+      "step": 7517
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005222863555166064,
+      "loss": 1.5312,
+      "step": 7518
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005220332396838203,
+      "loss": 1.7363,
+      "step": 7519
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005217801635341175,
+      "loss": 1.5156,
+      "step": 7520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005215271270885104,
+      "loss": 1.4766,
+      "step": 7521
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005212741303680069,
+      "loss": 1.6074,
+      "step": 7522
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005210211733936116,
+      "loss": 1.6367,
+      "step": 7523
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005207682561863272,
+      "loss": 1.5469,
+      "step": 7524
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005205153787671513,
+      "loss": 1.5332,
+      "step": 7525
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005202625411570793,
+      "loss": 1.5586,
+      "step": 7526
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005200097433771025,
+      "loss": 1.6973,
+      "step": 7527
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00051975698544821,
+      "loss": 1.5332,
+      "step": 7528
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005195042673913868,
+      "loss": 1.4785,
+      "step": 7529
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005192515892276147,
+      "loss": 1.6309,
+      "step": 7530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005189989509778723,
+      "loss": 1.6309,
+      "step": 7531
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005187463526631343,
+      "loss": 1.5254,
+      "step": 7532
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005184937943043734,
+      "loss": 1.666,
+      "step": 7533
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005182412759225581,
+      "loss": 1.543,
+      "step": 7534
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005179887975386533,
+      "loss": 1.5449,
+      "step": 7535
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005177363591736213,
+      "loss": 1.6445,
+      "step": 7536
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005174839608484205,
+      "loss": 1.5762,
+      "step": 7537
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005172316025840063,
+      "loss": 1.5723,
+      "step": 7538
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005169792844013302,
+      "loss": 1.5781,
+      "step": 7539
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005167270063213417,
+      "loss": 1.584,
+      "step": 7540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005164747683649858,
+      "loss": 1.6309,
+      "step": 7541
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005162225705532046,
+      "loss": 1.5195,
+      "step": 7542
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005159704129069362,
+      "loss": 1.6309,
+      "step": 7543
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005157182954471167,
+      "loss": 1.5137,
+      "step": 7544
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005154662181946778,
+      "loss": 1.6074,
+      "step": 7545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005152141811705482,
+      "loss": 1.5156,
+      "step": 7546
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005149621843956526,
+      "loss": 1.5195,
+      "step": 7547
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000514710227890914,
+      "loss": 1.5391,
+      "step": 7548
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005144583116772505,
+      "loss": 1.5977,
+      "step": 7549
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005142064357755774,
+      "loss": 1.6641,
+      "step": 7550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005139546002068064,
+      "loss": 1.5156,
+      "step": 7551
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005137028049918467,
+      "loss": 1.7617,
+      "step": 7552
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005134510501516032,
+      "loss": 1.6582,
+      "step": 7553
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005131993357069778,
+      "loss": 1.5488,
+      "step": 7554
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005129476616788687,
+      "loss": 1.5215,
+      "step": 7555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005126960280881719,
+      "loss": 1.709,
+      "step": 7556
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005124444349557786,
+      "loss": 1.6055,
+      "step": 7557
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005121928823025773,
+      "loss": 1.5488,
+      "step": 7558
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000511941370149453,
+      "loss": 1.6406,
+      "step": 7559
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000511689898517288,
+      "loss": 1.5801,
+      "step": 7560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005114384674269603,
+      "loss": 1.5957,
+      "step": 7561
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005111870768993448,
+      "loss": 1.7246,
+      "step": 7562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005109357269553131,
+      "loss": 1.6602,
+      "step": 7563
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000510684417615734,
+      "loss": 1.6035,
+      "step": 7564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000510433148901472,
+      "loss": 1.4609,
+      "step": 7565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005101819208333887,
+      "loss": 1.5547,
+      "step": 7566
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005099307334323418,
+      "loss": 1.6777,
+      "step": 7567
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005096795867191868,
+      "loss": 1.5801,
+      "step": 7568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000509428480714775,
+      "loss": 1.5684,
+      "step": 7569
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005091774154399542,
+      "loss": 1.4492,
+      "step": 7570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000508926390915569,
+      "loss": 1.6562,
+      "step": 7571
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005086754071624604,
+      "loss": 1.5879,
+      "step": 7572
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005084244642014669,
+      "loss": 1.5312,
+      "step": 7573
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005081735620534228,
+      "loss": 1.5781,
+      "step": 7574
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000507922700739159,
+      "loss": 1.6133,
+      "step": 7575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005076718802795033,
+      "loss": 1.6465,
+      "step": 7576
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00050742110069528,
+      "loss": 1.4727,
+      "step": 7577
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00050717036200731,
+      "loss": 1.6367,
+      "step": 7578
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005069196642364104,
+      "loss": 1.5566,
+      "step": 7579
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005066690074033962,
+      "loss": 1.6934,
+      "step": 7580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005064183915290776,
+      "loss": 1.5078,
+      "step": 7581
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005061678166342621,
+      "loss": 1.5371,
+      "step": 7582
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005059172827397531,
+      "loss": 1.5195,
+      "step": 7583
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005056667898663521,
+      "loss": 1.5957,
+      "step": 7584
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005054163380348555,
+      "loss": 1.543,
+      "step": 7585
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005051659272660574,
+      "loss": 1.5293,
+      "step": 7586
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0005049155575807473,
+      "loss": 1.5957,
+      "step": 7587
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005046652289997131,
+      "loss": 1.5996,
+      "step": 7588
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005044149415437378,
+      "loss": 1.5742,
+      "step": 7589
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005041646952336016,
+      "loss": 1.6465,
+      "step": 7590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005039144900900806,
+      "loss": 1.5566,
+      "step": 7591
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005036643261339489,
+      "loss": 1.5586,
+      "step": 7592
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005034142033859758,
+      "loss": 1.6641,
+      "step": 7593
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005031641218669279,
+      "loss": 1.6855,
+      "step": 7594
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005029140815975675,
+      "loss": 1.4609,
+      "step": 7595
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000502664082598655,
+      "loss": 1.582,
+      "step": 7596
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005024141248909464,
+      "loss": 1.4336,
+      "step": 7597
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000502164208495194,
+      "loss": 1.5742,
+      "step": 7598
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005019143334321469,
+      "loss": 1.6016,
+      "step": 7599
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005016644997225516,
+      "loss": 1.6074,
+      "step": 7600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00050141470738715,
+      "loss": 1.5977,
+      "step": 7601
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005011649564466813,
+      "loss": 1.7188,
+      "step": 7602
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005009152469218805,
+      "loss": 1.5645,
+      "step": 7603
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005006655788334806,
+      "loss": 1.6641,
+      "step": 7604
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005004159522022095,
+      "loss": 1.5234,
+      "step": 7605
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0005001663670487926,
+      "loss": 1.625,
+      "step": 7606
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004999168233939514,
+      "loss": 1.5938,
+      "step": 7607
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004996673212584047,
+      "loss": 1.5781,
+      "step": 7608
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004994178606628673,
+      "loss": 1.623,
+      "step": 7609
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004991684416280503,
+      "loss": 1.5625,
+      "step": 7610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004989190641746615,
+      "loss": 1.6582,
+      "step": 7611
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004986697283234059,
+      "loss": 1.6133,
+      "step": 7612
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004984204340949845,
+      "loss": 1.6582,
+      "step": 7613
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004981711815100948,
+      "loss": 1.625,
+      "step": 7614
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004979219705894308,
+      "loss": 1.6094,
+      "step": 7615
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004976728013536832,
+      "loss": 1.6113,
+      "step": 7616
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004974236738235394,
+      "loss": 1.709,
+      "step": 7617
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004971745880196826,
+      "loss": 1.4629,
+      "step": 7618
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000496925543962794,
+      "loss": 1.5215,
+      "step": 7619
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004966765416735499,
+      "loss": 1.5293,
+      "step": 7620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004964275811726237,
+      "loss": 1.584,
+      "step": 7621
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000496178662480685,
+      "loss": 1.6875,
+      "step": 7622
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004959297856184008,
+      "loss": 1.5938,
+      "step": 7623
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000495680950606434,
+      "loss": 1.5898,
+      "step": 7624
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004954321574654436,
+      "loss": 1.5332,
+      "step": 7625
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004951834062160857,
+      "loss": 1.5215,
+      "step": 7626
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004949346968790133,
+      "loss": 1.6895,
+      "step": 7627
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004946860294748751,
+      "loss": 1.627,
+      "step": 7628
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004944374040243166,
+      "loss": 1.4883,
+      "step": 7629
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004941888205479795,
+      "loss": 1.582,
+      "step": 7630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004939402790665033,
+      "loss": 1.6172,
+      "step": 7631
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004936917796005226,
+      "loss": 1.6406,
+      "step": 7632
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000493443322170669,
+      "loss": 1.6113,
+      "step": 7633
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004931949067975703,
+      "loss": 1.5586,
+      "step": 7634
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004929465335018519,
+      "loss": 1.582,
+      "step": 7635
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004926982023041346,
+      "loss": 1.5625,
+      "step": 7636
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004924499132250357,
+      "loss": 1.6738,
+      "step": 7637
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004922016662851695,
+      "loss": 1.666,
+      "step": 7638
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000491953461505147,
+      "loss": 1.666,
+      "step": 7639
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004917052989055751,
+      "loss": 1.4102,
+      "step": 7640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004914571785070575,
+      "loss": 1.5059,
+      "step": 7641
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000491209100330194,
+      "loss": 1.5664,
+      "step": 7642
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004909610643955818,
+      "loss": 1.6562,
+      "step": 7643
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004907130707238137,
+      "loss": 1.5703,
+      "step": 7644
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004904651193354795,
+      "loss": 1.6875,
+      "step": 7645
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004902172102511646,
+      "loss": 1.5859,
+      "step": 7646
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004899693434914528,
+      "loss": 1.5566,
+      "step": 7647
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004897215190769224,
+      "loss": 1.5762,
+      "step": 7648
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004894737370281493,
+      "loss": 1.6387,
+      "step": 7649
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004892259973657051,
+      "loss": 1.627,
+      "step": 7650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004889783001101585,
+      "loss": 1.4785,
+      "step": 7651
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004887306452820749,
+      "loss": 1.5156,
+      "step": 7652
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004884830329020155,
+      "loss": 1.6641,
+      "step": 7653
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004882354629905382,
+      "loss": 1.6484,
+      "step": 7654
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004879879355681975,
+      "loss": 1.7188,
+      "step": 7655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00048774045065554453,
+      "loss": 1.5996,
+      "step": 7656
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00048749300827312625,
+      "loss": 1.7227,
+      "step": 7657
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00048724560844148635,
+      "loss": 1.6191,
+      "step": 7658
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004869982511811659,
+      "loss": 1.6113,
+      "step": 7659
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004867509365127013,
+      "loss": 1.5977,
+      "step": 7660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00048650366445662584,
+      "loss": 1.4902,
+      "step": 7661
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00048625643503346874,
+      "loss": 1.6953,
+      "step": 7662
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000486009248263757,
+      "loss": 1.541,
+      "step": 7663
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004857621041680128,
+      "loss": 1.6309,
+      "step": 7664
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004855150027667553,
+      "loss": 1.6172,
+      "step": 7665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004852679440804997,
+      "loss": 1.541,
+      "step": 7666
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004850209281297585,
+      "loss": 1.6621,
+      "step": 7667
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004847739549350401,
+      "loss": 1.748,
+      "step": 7668
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00048452702451684917,
+      "loss": 1.6465,
+      "step": 7669
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00048428013689568675,
+      "loss": 1.625,
+      "step": 7670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00048403329209205137,
+      "loss": 1.5996,
+      "step": 7671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00048378649012643693,
+      "loss": 1.6348,
+      "step": 7672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004835397310193341,
+      "loss": 1.4219,
+      "step": 7673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004832930147912296,
+      "loss": 1.6484,
+      "step": 7674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00048304634146260764,
+      "loss": 1.4805,
+      "step": 7675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00048279971105394805,
+      "loss": 1.6562,
+      "step": 7676
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00048255312358572714,
+      "loss": 1.6562,
+      "step": 7677
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004823065790784173,
+      "loss": 1.5488,
+      "step": 7678
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004820600775524888,
+      "loss": 1.6699,
+      "step": 7679
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004818136190284069,
+      "loss": 1.582,
+      "step": 7680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00048156720352663377,
+      "loss": 1.6621,
+      "step": 7681
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00048132083106762756,
+      "loss": 1.6367,
+      "step": 7682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004810745016718442,
+      "loss": 1.5762,
+      "step": 7683
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00048082821535973464,
+      "loss": 1.5,
+      "step": 7684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004805819721517467,
+      "loss": 1.6055,
+      "step": 7685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00048033577206832445,
+      "loss": 1.625,
+      "step": 7686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00048008961512990925,
+      "loss": 1.5703,
+      "step": 7687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00047984350135693777,
+      "loss": 1.5566,
+      "step": 7688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00047959743076984374,
+      "loss": 1.6055,
+      "step": 7689
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00047935140338905657,
+      "loss": 1.502,
+      "step": 7690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004791054192350035,
+      "loss": 1.6699,
+      "step": 7691
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004788594783281068,
+      "loss": 1.623,
+      "step": 7692
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004786135806887858,
+      "loss": 1.5078,
+      "step": 7693
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00047836772633745594,
+      "loss": 1.6562,
+      "step": 7694
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00047812191529452943,
+      "loss": 1.5195,
+      "step": 7695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00047787614758041454,
+      "loss": 1.5254,
+      "step": 7696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00047763042321551574,
+      "loss": 1.4902,
+      "step": 7697
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000477384742220235,
+      "loss": 1.668,
+      "step": 7698
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004771391046149696,
+      "loss": 1.5645,
+      "step": 7699
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0004768935104201133,
+      "loss": 1.623,
+      "step": 7700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047664795965605646,
+      "loss": 1.5605,
+      "step": 7701
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047640245234318636,
+      "loss": 1.5977,
+      "step": 7702
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047615698850188607,
+      "loss": 1.5117,
+      "step": 7703
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047591156815253487,
+      "loss": 1.4648,
+      "step": 7704
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004756661913155087,
+      "loss": 1.4004,
+      "step": 7705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004754208580111804,
+      "loss": 1.5547,
+      "step": 7706
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004751755682599184,
+      "loss": 1.6289,
+      "step": 7707
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047493032208208787,
+      "loss": 1.4941,
+      "step": 7708
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047468511949804994,
+      "loss": 1.4941,
+      "step": 7709
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004744399605281633,
+      "loss": 1.5605,
+      "step": 7710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047419484519278164,
+      "loss": 1.5625,
+      "step": 7711
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047394977351225586,
+      "loss": 1.6543,
+      "step": 7712
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004737047455069324,
+      "loss": 1.5977,
+      "step": 7713
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004734597611971556,
+      "loss": 1.5469,
+      "step": 7714
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004732148206032647,
+      "loss": 1.5605,
+      "step": 7715
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000472969923745596,
+      "loss": 1.7695,
+      "step": 7716
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004727250706444813,
+      "loss": 1.6738,
+      "step": 7717
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004724802613202507,
+      "loss": 1.5605,
+      "step": 7718
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004722354957932287,
+      "loss": 1.5059,
+      "step": 7719
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047199077408373693,
+      "loss": 1.6367,
+      "step": 7720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004717460962120931,
+      "loss": 1.6445,
+      "step": 7721
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004715014621986123,
+      "loss": 1.498,
+      "step": 7722
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004712568720636048,
+      "loss": 1.7188,
+      "step": 7723
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047101232582737754,
+      "loss": 1.4961,
+      "step": 7724
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004707678235102337,
+      "loss": 1.6133,
+      "step": 7725
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004705233651324737,
+      "loss": 1.5371,
+      "step": 7726
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004702789507143933,
+      "loss": 1.6465,
+      "step": 7727
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00047003458027628485,
+      "loss": 1.5312,
+      "step": 7728
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046979025383843734,
+      "loss": 1.5723,
+      "step": 7729
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004695459714211354,
+      "loss": 1.6172,
+      "step": 7730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004693017330446613,
+      "loss": 1.5996,
+      "step": 7731
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046905753872929256,
+      "loss": 1.6133,
+      "step": 7732
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004688133884953032,
+      "loss": 1.5078,
+      "step": 7733
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004685692823629639,
+      "loss": 1.5234,
+      "step": 7734
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046832522035254156,
+      "loss": 1.6719,
+      "step": 7735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004680812024842992,
+      "loss": 1.6543,
+      "step": 7736
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000467837228778496,
+      "loss": 1.6035,
+      "step": 7737
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046759329925538873,
+      "loss": 1.5742,
+      "step": 7738
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000467349413935229,
+      "loss": 1.582,
+      "step": 7739
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004671055728382656,
+      "loss": 1.7637,
+      "step": 7740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046686177598474266,
+      "loss": 1.6836,
+      "step": 7741
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046661802339490247,
+      "loss": 1.6113,
+      "step": 7742
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046637431508898185,
+      "loss": 1.4824,
+      "step": 7743
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004661306510872149,
+      "loss": 1.4922,
+      "step": 7744
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004658870314098311,
+      "loss": 1.6289,
+      "step": 7745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004656434560770579,
+      "loss": 1.5938,
+      "step": 7746
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046539992510911764,
+      "loss": 1.6973,
+      "step": 7747
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004651564385262295,
+      "loss": 1.6211,
+      "step": 7748
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004649129963486084,
+      "loss": 1.4766,
+      "step": 7749
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046466959859646685,
+      "loss": 1.5859,
+      "step": 7750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004644262452900125,
+      "loss": 1.5918,
+      "step": 7751
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004641829364494499,
+      "loss": 1.668,
+      "step": 7752
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000463939672094979,
+      "loss": 1.5605,
+      "step": 7753
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004636964522467978,
+      "loss": 1.5664,
+      "step": 7754
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046345327692509907,
+      "loss": 1.4766,
+      "step": 7755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004632101461500725,
+      "loss": 1.6973,
+      "step": 7756
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004629670599419035,
+      "loss": 1.5996,
+      "step": 7757
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046272401832077493,
+      "loss": 1.5332,
+      "step": 7758
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046248102130686496,
+      "loss": 1.6758,
+      "step": 7759
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046223806892034834,
+      "loss": 1.5469,
+      "step": 7760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046199516118139596,
+      "loss": 1.6387,
+      "step": 7761
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004617522981101756,
+      "loss": 1.6191,
+      "step": 7762
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004615094797268508,
+      "loss": 1.5508,
+      "step": 7763
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004612667060515814,
+      "loss": 1.4082,
+      "step": 7764
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046102397710452324,
+      "loss": 1.6094,
+      "step": 7765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004607812929058296,
+      "loss": 1.5801,
+      "step": 7766
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046053865347564895,
+      "loss": 1.6953,
+      "step": 7767
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004602960588341263,
+      "loss": 1.5957,
+      "step": 7768
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00046005350900140275,
+      "loss": 1.5938,
+      "step": 7769
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045981100399761655,
+      "loss": 1.6309,
+      "step": 7770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045956854384290136,
+      "loss": 1.7129,
+      "step": 7771
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004593261285573873,
+      "loss": 1.6367,
+      "step": 7772
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045908375816120097,
+      "loss": 1.6289,
+      "step": 7773
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000458841432674465,
+      "loss": 1.5293,
+      "step": 7774
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045859915211729843,
+      "loss": 1.6172,
+      "step": 7775
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045835691650981616,
+      "loss": 1.6289,
+      "step": 7776
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045811472587213066,
+      "loss": 1.375,
+      "step": 7777
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004578725802243491,
+      "loss": 1.6035,
+      "step": 7778
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004576304795865758,
+      "loss": 1.707,
+      "step": 7779
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045738842397891066,
+      "loss": 1.6055,
+      "step": 7780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000457146413421451,
+      "loss": 1.6836,
+      "step": 7781
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004569044479342893,
+      "loss": 1.6055,
+      "step": 7782
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004566625275375149,
+      "loss": 1.5898,
+      "step": 7783
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045642065225121254,
+      "loss": 1.541,
+      "step": 7784
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004561788220954648,
+      "loss": 1.4727,
+      "step": 7785
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004559370370903491,
+      "loss": 1.5391,
+      "step": 7786
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004556952972559395,
+      "loss": 1.5898,
+      "step": 7787
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004554536026123063,
+      "loss": 1.6113,
+      "step": 7788
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045521195317951667,
+      "loss": 1.6035,
+      "step": 7789
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004549703489776331,
+      "loss": 1.4688,
+      "step": 7790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045472879002671495,
+      "loss": 1.5547,
+      "step": 7791
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045448727634681696,
+      "loss": 1.5859,
+      "step": 7792
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045424580795799173,
+      "loss": 1.5352,
+      "step": 7793
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004540043848802866,
+      "loss": 1.4414,
+      "step": 7794
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004537630071337456,
+      "loss": 1.5449,
+      "step": 7795
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000453521674738409,
+      "loss": 1.5293,
+      "step": 7796
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004532803877143138,
+      "loss": 1.6562,
+      "step": 7797
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004530391460814925,
+      "loss": 1.5684,
+      "step": 7798
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045279794985997427,
+      "loss": 1.6582,
+      "step": 7799
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045255679906978377,
+      "loss": 1.502,
+      "step": 7800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004523156937309434,
+      "loss": 1.5859,
+      "step": 7801
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045207463386347046,
+      "loss": 1.627,
+      "step": 7802
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000451833619487379,
+      "loss": 1.6426,
+      "step": 7803
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004515926506226786,
+      "loss": 1.5605,
+      "step": 7804
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004513517272893766,
+      "loss": 1.5176,
+      "step": 7805
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045111084950747506,
+      "loss": 1.5195,
+      "step": 7806
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045087001729697296,
+      "loss": 1.6484,
+      "step": 7807
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045062923067786475,
+      "loss": 1.5098,
+      "step": 7808
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004503884896701428,
+      "loss": 1.4746,
+      "step": 7809
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00045014779429379385,
+      "loss": 1.6641,
+      "step": 7810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004499071445688017,
+      "loss": 1.5996,
+      "step": 7811
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0004496665405151462,
+      "loss": 1.5801,
+      "step": 7812
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004494259821528035,
+      "loss": 1.5684,
+      "step": 7813
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044918546950174597,
+      "loss": 1.5234,
+      "step": 7814
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044894500258194206,
+      "loss": 1.668,
+      "step": 7815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000448704581413356,
+      "loss": 1.5918,
+      "step": 7816
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004484642060159496,
+      "loss": 1.6387,
+      "step": 7817
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044822387640967956,
+      "loss": 1.5977,
+      "step": 7818
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004479835926144992,
+      "loss": 1.4512,
+      "step": 7819
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044774335465035754,
+      "loss": 1.5762,
+      "step": 7820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044750316253720123,
+      "loss": 1.7012,
+      "step": 7821
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044726301629497166,
+      "loss": 1.5293,
+      "step": 7822
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000447022915943607,
+      "loss": 1.5332,
+      "step": 7823
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044678286150304106,
+      "loss": 1.5234,
+      "step": 7824
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044654285299320495,
+      "loss": 1.418,
+      "step": 7825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044630289043402527,
+      "loss": 1.541,
+      "step": 7826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004460629738454246,
+      "loss": 1.5801,
+      "step": 7827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004458231032473218,
+      "loss": 1.6562,
+      "step": 7828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004455832786596325,
+      "loss": 1.5879,
+      "step": 7829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004453435001022681,
+      "loss": 1.6016,
+      "step": 7830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004451037675951358,
+      "loss": 1.5781,
+      "step": 7831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044486408115813924,
+      "loss": 1.6211,
+      "step": 7832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004446244408111789,
+      "loss": 1.5332,
+      "step": 7833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044438484657415055,
+      "loss": 1.627,
+      "step": 7834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044414529846694654,
+      "loss": 1.4668,
+      "step": 7835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004439057965094547,
+      "loss": 1.5332,
+      "step": 7836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044366634072156067,
+      "loss": 1.5586,
+      "step": 7837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004434269311231446,
+      "loss": 1.6406,
+      "step": 7838
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044318756773408343,
+      "loss": 1.5859,
+      "step": 7839
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044294825057425,
+      "loss": 1.584,
+      "step": 7840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004427089796635143,
+      "loss": 1.5352,
+      "step": 7841
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004424697550217414,
+      "loss": 1.5059,
+      "step": 7842
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044223057666879274,
+      "loss": 1.7168,
+      "step": 7843
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044199144462452565,
+      "loss": 1.5488,
+      "step": 7844
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044175235890879505,
+      "loss": 1.6758,
+      "step": 7845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044151331954145025,
+      "loss": 1.627,
+      "step": 7846
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004412743265423377,
+      "loss": 1.6582,
+      "step": 7847
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044103537993129914,
+      "loss": 1.4668,
+      "step": 7848
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044079647972817404,
+      "loss": 1.5781,
+      "step": 7849
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004405576259527966,
+      "loss": 1.5664,
+      "step": 7850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044031881862499747,
+      "loss": 1.5508,
+      "step": 7851
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00044008005776460357,
+      "loss": 1.5156,
+      "step": 7852
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043984134339143823,
+      "loss": 1.5273,
+      "step": 7853
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004396026755253204,
+      "loss": 1.5781,
+      "step": 7854
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043936405418606506,
+      "loss": 1.6777,
+      "step": 7855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043912547939348466,
+      "loss": 1.6289,
+      "step": 7856
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004388869511673863,
+      "loss": 1.6035,
+      "step": 7857
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043864846952757365,
+      "loss": 1.6738,
+      "step": 7858
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043841003449384677,
+      "loss": 1.6875,
+      "step": 7859
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004381716460860011,
+      "loss": 1.6621,
+      "step": 7860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004379333043238297,
+      "loss": 1.6777,
+      "step": 7861
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043769500922712047,
+      "loss": 1.6055,
+      "step": 7862
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043745676081565735,
+      "loss": 1.627,
+      "step": 7863
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004372185591092216,
+      "loss": 1.6406,
+      "step": 7864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043698040412758953,
+      "loss": 1.541,
+      "step": 7865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004367422958905339,
+      "loss": 1.4688,
+      "step": 7866
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043650423441782325,
+      "loss": 1.748,
+      "step": 7867
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004362662197292233,
+      "loss": 1.498,
+      "step": 7868
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004360282518444948,
+      "loss": 1.5977,
+      "step": 7869
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004357903307833949,
+      "loss": 1.707,
+      "step": 7870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043555245656567667,
+      "loss": 1.5938,
+      "step": 7871
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004353146292110903,
+      "loss": 1.6328,
+      "step": 7872
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043507684873938103,
+      "loss": 1.6152,
+      "step": 7873
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004348391151702904,
+      "loss": 1.5469,
+      "step": 7874
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004346014285235559,
+      "loss": 1.5547,
+      "step": 7875
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043436378881891205,
+      "loss": 1.6523,
+      "step": 7876
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004341261960760886,
+      "loss": 1.6719,
+      "step": 7877
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004338886503148115,
+      "loss": 1.4668,
+      "step": 7878
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043365115155480274,
+      "loss": 1.5605,
+      "step": 7879
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004334136998157813,
+      "loss": 1.5938,
+      "step": 7880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043317629511746103,
+      "loss": 1.6348,
+      "step": 7881
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004329389374795525,
+      "loss": 1.5137,
+      "step": 7882
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043270162692176197,
+      "loss": 1.541,
+      "step": 7883
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043246436346379267,
+      "loss": 1.7207,
+      "step": 7884
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004322271471253432,
+      "loss": 1.6523,
+      "step": 7885
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004319899779261082,
+      "loss": 1.5352,
+      "step": 7886
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004317528558857783,
+      "loss": 1.7148,
+      "step": 7887
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004315157810240412,
+      "loss": 1.584,
+      "step": 7888
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043127875336057976,
+      "loss": 1.5,
+      "step": 7889
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043104177291507294,
+      "loss": 1.6719,
+      "step": 7890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004308048397071961,
+      "loss": 1.4727,
+      "step": 7891
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00043056795375662064,
+      "loss": 1.5996,
+      "step": 7892
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004303311150830138,
+      "loss": 1.5742,
+      "step": 7893
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004300943237060392,
+      "loss": 1.498,
+      "step": 7894
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000429857579645356,
+      "loss": 1.5449,
+      "step": 7895
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00042962088292062043,
+      "loss": 1.6387,
+      "step": 7896
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004293842335514839,
+      "loss": 1.5156,
+      "step": 7897
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004291476315575943,
+      "loss": 1.5039,
+      "step": 7898
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004289110769585949,
+      "loss": 1.5938,
+      "step": 7899
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00042867456977412635,
+      "loss": 1.5938,
+      "step": 7900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00042843811002382437,
+      "loss": 1.6172,
+      "step": 7901
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00042820169772732076,
+      "loss": 1.5469,
+      "step": 7902
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00042796533290424355,
+      "loss": 1.6406,
+      "step": 7903
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004277290155742173,
+      "loss": 1.5703,
+      "step": 7904
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00042749274575686205,
+      "loss": 1.5156,
+      "step": 7905
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004272565234717939,
+      "loss": 1.6035,
+      "step": 7906
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00042702034873862485,
+      "loss": 1.5,
+      "step": 7907
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00042678422157696396,
+      "loss": 1.5234,
+      "step": 7908
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004265481420064153,
+      "loss": 1.4961,
+      "step": 7909
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004263121100465793,
+      "loss": 1.5117,
+      "step": 7910
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004260761257170522,
+      "loss": 1.7441,
+      "step": 7911
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004258401890374272,
+      "loss": 1.3809,
+      "step": 7912
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00042560430002729245,
+      "loss": 1.6055,
+      "step": 7913
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00042536845870623264,
+      "loss": 1.5879,
+      "step": 7914
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004251326650938281,
+      "loss": 1.6738,
+      "step": 7915
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004248969192096562,
+      "loss": 1.5566,
+      "step": 7916
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004246612210732895,
+      "loss": 1.6797,
+      "step": 7917
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00042442557070429657,
+      "loss": 1.6895,
+      "step": 7918
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000424189968122242,
+      "loss": 1.627,
+      "step": 7919
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004239544133466874,
+      "loss": 1.5996,
+      "step": 7920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00042371890639718914,
+      "loss": 1.582,
+      "step": 7921
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004234834472933002,
+      "loss": 1.6973,
+      "step": 7922
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0004232480360545692,
+      "loss": 1.6152,
+      "step": 7923
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00042301267270054176,
+      "loss": 1.6484,
+      "step": 7924
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004227773572507586,
+      "loss": 1.6074,
+      "step": 7925
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00042254208972475674,
+      "loss": 1.584,
+      "step": 7926
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004223068701420687,
+      "loss": 1.6621,
+      "step": 7927
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004220716985222244,
+      "loss": 1.5918,
+      "step": 7928
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00042183657488474834,
+      "loss": 1.5703,
+      "step": 7929
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00042160149924916205,
+      "loss": 1.5938,
+      "step": 7930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004213664716349821,
+      "loss": 1.4707,
+      "step": 7931
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000421131492061722,
+      "loss": 1.6504,
+      "step": 7932
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004208965605488907,
+      "loss": 1.7402,
+      "step": 7933
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004206616771159929,
+      "loss": 1.6543,
+      "step": 7934
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004204268417825308,
+      "loss": 1.6289,
+      "step": 7935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004201920545680007,
+      "loss": 1.5977,
+      "step": 7936
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004199573154918962,
+      "loss": 1.541,
+      "step": 7937
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041972262457370626,
+      "loss": 1.5273,
+      "step": 7938
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004194879818329156,
+      "loss": 1.5742,
+      "step": 7939
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041925338728900633,
+      "loss": 1.6641,
+      "step": 7940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041901884096145505,
+      "loss": 1.4863,
+      "step": 7941
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000418784342869735,
+      "loss": 1.6289,
+      "step": 7942
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041854989303331493,
+      "loss": 1.6523,
+      "step": 7943
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041831549147166084,
+      "loss": 1.5059,
+      "step": 7944
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004180811382042333,
+      "loss": 1.502,
+      "step": 7945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004178468332504892,
+      "loss": 1.4688,
+      "step": 7946
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004176125766298824,
+      "loss": 1.5762,
+      "step": 7947
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004173783683618615,
+      "loss": 1.666,
+      "step": 7948
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004171442084658718,
+      "loss": 1.4805,
+      "step": 7949
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041691009696135387,
+      "loss": 1.6367,
+      "step": 7950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041667603386774546,
+      "loss": 1.6309,
+      "step": 7951
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004164420192044794,
+      "loss": 1.5996,
+      "step": 7952
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041620805299098454,
+      "loss": 1.584,
+      "step": 7953
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004159741352466856,
+      "loss": 1.5801,
+      "step": 7954
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041574026599100423,
+      "loss": 1.6016,
+      "step": 7955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004155064452433571,
+      "loss": 1.7031,
+      "step": 7956
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041527267302315707,
+      "loss": 1.5566,
+      "step": 7957
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004150389493498126,
+      "loss": 1.5801,
+      "step": 7958
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004148052742427293,
+      "loss": 1.6855,
+      "step": 7959
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004145716477213077,
+      "loss": 1.5176,
+      "step": 7960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004143380698049445,
+      "loss": 1.623,
+      "step": 7961
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041410454051303227,
+      "loss": 1.5469,
+      "step": 7962
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004138710598649602,
+      "loss": 1.5703,
+      "step": 7963
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004136376278801127,
+      "loss": 1.6758,
+      "step": 7964
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004134042445778705,
+      "loss": 1.625,
+      "step": 7965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041317090997760973,
+      "loss": 1.4668,
+      "step": 7966
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004129376240987037,
+      "loss": 1.5195,
+      "step": 7967
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004127043869605206,
+      "loss": 1.5645,
+      "step": 7968
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004124711985824249,
+      "loss": 1.627,
+      "step": 7969
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041223805898377684,
+      "loss": 1.5879,
+      "step": 7970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004120049681839331,
+      "loss": 1.6641,
+      "step": 7971
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041177192620224566,
+      "loss": 1.4492,
+      "step": 7972
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004115389330580629,
+      "loss": 1.4902,
+      "step": 7973
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041130598877072875,
+      "loss": 1.5918,
+      "step": 7974
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000411073093359584,
+      "loss": 1.6758,
+      "step": 7975
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041084024684396437,
+      "loss": 1.6055,
+      "step": 7976
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041060744924320195,
+      "loss": 1.6465,
+      "step": 7977
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004103747005766243,
+      "loss": 1.5059,
+      "step": 7978
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00041014200086355603,
+      "loss": 1.498,
+      "step": 7979
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040990935012331665,
+      "loss": 1.5469,
+      "step": 7980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040967674837522197,
+      "loss": 1.459,
+      "step": 7981
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040944419563858326,
+      "loss": 1.4902,
+      "step": 7982
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004092116919327089,
+      "loss": 1.5293,
+      "step": 7983
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040897923727690224,
+      "loss": 1.5117,
+      "step": 7984
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040874683169046257,
+      "loss": 1.6406,
+      "step": 7985
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000408514475192685,
+      "loss": 1.6523,
+      "step": 7986
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040828216780286166,
+      "loss": 1.7402,
+      "step": 7987
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004080499095402794,
+      "loss": 1.5996,
+      "step": 7988
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004078177004242216,
+      "loss": 1.6094,
+      "step": 7989
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040758554047396666,
+      "loss": 1.5859,
+      "step": 7990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040735342970879063,
+      "loss": 1.582,
+      "step": 7991
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004071213681479641,
+      "loss": 1.4766,
+      "step": 7992
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004068893558107536,
+      "loss": 1.6621,
+      "step": 7993
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004066573927164219,
+      "loss": 1.582,
+      "step": 7994
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004064254788842282,
+      "loss": 1.584,
+      "step": 7995
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040619361433342685,
+      "loss": 1.4922,
+      "step": 7996
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040596179908326824,
+      "loss": 1.5918,
+      "step": 7997
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004057300331529985,
+      "loss": 1.6074,
+      "step": 7998
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040549831656186067,
+      "loss": 1.6172,
+      "step": 7999
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004052666493290925,
+      "loss": 1.5469,
+      "step": 8000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040503503147392816,
+      "loss": 1.4004,
+      "step": 8001
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040480346301559745,
+      "loss": 1.543,
+      "step": 8002
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040457194397332697,
+      "loss": 1.6992,
+      "step": 8003
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040434047436633803,
+      "loss": 1.7422,
+      "step": 8004
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040410905421384835,
+      "loss": 1.5215,
+      "step": 8005
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004038776835350714,
+      "loss": 1.6387,
+      "step": 8006
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040364636234921716,
+      "loss": 1.4805,
+      "step": 8007
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004034150906754907,
+      "loss": 1.582,
+      "step": 8008
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004031838685330934,
+      "loss": 1.5859,
+      "step": 8009
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004029526959412224,
+      "loss": 1.5859,
+      "step": 8010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004027215729190706,
+      "loss": 1.5762,
+      "step": 8011
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004024904994858273,
+      "loss": 1.584,
+      "step": 8012
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040225947566067656,
+      "loss": 1.5723,
+      "step": 8013
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004020285014628,
+      "loss": 1.6348,
+      "step": 8014
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004017975769113738,
+      "loss": 1.6055,
+      "step": 8015
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040156670202557044,
+      "loss": 1.5938,
+      "step": 8016
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040133587682455817,
+      "loss": 1.6504,
+      "step": 8017
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000401105101327501,
+      "loss": 1.5742,
+      "step": 8018
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040087437555355967,
+      "loss": 1.6191,
+      "step": 8019
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00040064369952188975,
+      "loss": 1.5723,
+      "step": 8020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000400413073251643,
+      "loss": 1.6504,
+      "step": 8021
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0004001824967619668,
+      "loss": 1.7207,
+      "step": 8022
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003999519700720056,
+      "loss": 1.6523,
+      "step": 8023
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003997214932008981,
+      "loss": 1.6582,
+      "step": 8024
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003994910661677801,
+      "loss": 1.5352,
+      "step": 8025
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003992606889917819,
+      "loss": 1.5684,
+      "step": 8026
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00039903036169203145,
+      "loss": 1.6094,
+      "step": 8027
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00039880008428765134,
+      "loss": 1.7012,
+      "step": 8028
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00039856985679776017,
+      "loss": 1.4941,
+      "step": 8029
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003983396792414722,
+      "loss": 1.5645,
+      "step": 8030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003981095516378986,
+      "loss": 1.6035,
+      "step": 8031
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00039787947400614533,
+      "loss": 1.5996,
+      "step": 8032
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003976494463653142,
+      "loss": 1.5449,
+      "step": 8033
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003974194687345037,
+      "loss": 1.5664,
+      "step": 8034
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003971895411328077,
+      "loss": 1.4492,
+      "step": 8035
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003969596635793156,
+      "loss": 1.6406,
+      "step": 8036
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0003967298360931125,
+      "loss": 1.5664,
+      "step": 8037
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00039650005869328075,
+      "loss": 1.5586,
+      "step": 8038
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00039627033139889715,
+      "loss": 1.5195,
+      "step": 8039
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00039604065422903456,
+      "loss": 1.5059,
+      "step": 8040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00039581102720276183,
+      "loss": 1.6562,
+      "step": 8041
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00039558145033914417,
+      "loss": 1.6289,
+      "step": 8042
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00039535192365724184,
+      "loss": 1.6543,
+      "step": 8043
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003951224471761112,
+      "loss": 1.6406,
+      "step": 8044
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003948930209148043,
+      "loss": 1.5762,
+      "step": 8045
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003946636448923697,
+      "loss": 1.543,
+      "step": 8046
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003944343191278511,
+      "loss": 1.6016,
+      "step": 8047
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00039420504364028817,
+      "loss": 1.5918,
+      "step": 8048
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00039397581844871654,
+      "loss": 1.668,
+      "step": 8049
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003937466435721674,
+      "loss": 1.5,
+      "step": 8050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00039351751902966806,
+      "loss": 1.6504,
+      "step": 8051
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003932884448402411,
+      "loss": 1.6016,
+      "step": 8052
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00039305942102290607,
+      "loss": 1.6504,
+      "step": 8053
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00039283044759667743,
+      "loss": 1.4824,
+      "step": 8054
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003926015245805654,
+      "loss": 1.5801,
+      "step": 8055
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003923726519935765,
+      "loss": 1.5566,
+      "step": 8056
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003921438298547123,
+      "loss": 1.6113,
+      "step": 8057
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003919150581829716,
+      "loss": 1.5156,
+      "step": 8058
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003916863369973475,
+      "loss": 1.5352,
+      "step": 8059
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003914576663168298,
+      "loss": 1.4785,
+      "step": 8060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00039122904616040324,
+      "loss": 1.5449,
+      "step": 8061
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003910004765470497,
+      "loss": 1.6094,
+      "step": 8062
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00039077195749574604,
+      "loss": 1.4277,
+      "step": 8063
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003905434890254647,
+      "loss": 1.5254,
+      "step": 8064
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00039031507115517395,
+      "loss": 1.5918,
+      "step": 8065
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003900867039038388,
+      "loss": 1.5293,
+      "step": 8066
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003898583872904192,
+      "loss": 1.6035,
+      "step": 8067
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000389630121333871,
+      "loss": 1.6094,
+      "step": 8068
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038940190605314554,
+      "loss": 1.5781,
+      "step": 8069
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038917374146719096,
+      "loss": 1.5859,
+      "step": 8070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003889456275949504,
+      "loss": 1.5645,
+      "step": 8071
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000388717564455363,
+      "loss": 1.5898,
+      "step": 8072
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000388489552067363,
+      "loss": 1.5664,
+      "step": 8073
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038826159044988217,
+      "loss": 1.5508,
+      "step": 8074
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038803367962184635,
+      "loss": 1.5098,
+      "step": 8075
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038780581960217787,
+      "loss": 1.6113,
+      "step": 8076
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038757801040979436,
+      "loss": 1.6289,
+      "step": 8077
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003873502520636104,
+      "loss": 1.6289,
+      "step": 8078
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038712254458253515,
+      "loss": 1.8047,
+      "step": 8079
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000386894887985474,
+      "loss": 1.5352,
+      "step": 8080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038666728229132784,
+      "loss": 1.5469,
+      "step": 8081
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038643972751899414,
+      "loss": 1.4902,
+      "step": 8082
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003862122236873653,
+      "loss": 1.5195,
+      "step": 8083
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003859847708153298,
+      "loss": 1.5859,
+      "step": 8084
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038575736892177136,
+      "loss": 1.5156,
+      "step": 8085
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000385530018025571,
+      "loss": 1.5137,
+      "step": 8086
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003853027181456037,
+      "loss": 1.5898,
+      "step": 8087
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038507546930074136,
+      "loss": 1.6055,
+      "step": 8088
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038484827150985103,
+      "loss": 1.4727,
+      "step": 8089
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003846211247917958,
+      "loss": 1.6504,
+      "step": 8090
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038439402916543453,
+      "loss": 1.5508,
+      "step": 8091
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003841669846496214,
+      "loss": 1.5547,
+      "step": 8092
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038393999126320745,
+      "loss": 1.4297,
+      "step": 8093
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038371304902503847,
+      "loss": 1.543,
+      "step": 8094
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000383486157953956,
+      "loss": 1.5898,
+      "step": 8095
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003832593180687981,
+      "loss": 1.623,
+      "step": 8096
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038303252938839727,
+      "loss": 1.4766,
+      "step": 8097
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038280579193158363,
+      "loss": 1.5586,
+      "step": 8098
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038257910571718145,
+      "loss": 1.4785,
+      "step": 8099
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038235247076401146,
+      "loss": 1.6777,
+      "step": 8100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003821258870908896,
+      "loss": 1.5762,
+      "step": 8101
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003818993547166285,
+      "loss": 1.6543,
+      "step": 8102
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038167287366003565,
+      "loss": 1.5234,
+      "step": 8103
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003814464439399147,
+      "loss": 1.5215,
+      "step": 8104
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003812200655750645,
+      "loss": 1.582,
+      "step": 8105
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00038099373858428064,
+      "loss": 1.4062,
+      "step": 8106
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003807674629863538,
+      "loss": 1.666,
+      "step": 8107
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003805412388000703,
+      "loss": 1.625,
+      "step": 8108
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000380315066044212,
+      "loss": 1.5508,
+      "step": 8109
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003800889447375575,
+      "loss": 1.5742,
+      "step": 8110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037986287489888014,
+      "loss": 1.5469,
+      "step": 8111
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037963685654694945,
+      "loss": 1.6191,
+      "step": 8112
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037941088970053005,
+      "loss": 1.5508,
+      "step": 8113
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003791849743783835,
+      "loss": 1.5957,
+      "step": 8114
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037895911059926613,
+      "loss": 1.541,
+      "step": 8115
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003787332983819298,
+      "loss": 1.5625,
+      "step": 8116
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037850753774512326,
+      "loss": 1.6992,
+      "step": 8117
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037828182870758986,
+      "loss": 1.5586,
+      "step": 8118
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000378056171288069,
+      "loss": 1.5449,
+      "step": 8119
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003778305655052955,
+      "loss": 1.6855,
+      "step": 8120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037760501137800096,
+      "loss": 1.6035,
+      "step": 8121
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037737950892491166,
+      "loss": 1.5801,
+      "step": 8122
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003771540581647498,
+      "loss": 1.5195,
+      "step": 8123
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037692865911623306,
+      "loss": 1.4941,
+      "step": 8124
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037670331179807595,
+      "loss": 1.6074,
+      "step": 8125
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037647801622898736,
+      "loss": 1.623,
+      "step": 8126
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037625277242767243,
+      "loss": 1.6348,
+      "step": 8127
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037602758041283213,
+      "loss": 1.4102,
+      "step": 8128
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037580244020316277,
+      "loss": 1.6445,
+      "step": 8129
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037557735181735675,
+      "loss": 1.6035,
+      "step": 8130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003753523152741015,
+      "loss": 1.5566,
+      "step": 8131
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003751273305920814,
+      "loss": 1.5078,
+      "step": 8132
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037490239778997537,
+      "loss": 1.5234,
+      "step": 8133
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037467751688645846,
+      "loss": 1.5625,
+      "step": 8134
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037445268790020126,
+      "loss": 1.6113,
+      "step": 8135
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037422791084986985,
+      "loss": 1.5215,
+      "step": 8136
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037400318575412706,
+      "loss": 1.6426,
+      "step": 8137
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003737785126316301,
+      "loss": 1.5488,
+      "step": 8138
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003735538915010327,
+      "loss": 1.4766,
+      "step": 8139
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003733293223809834,
+      "loss": 1.5898,
+      "step": 8140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003731048052901277,
+      "loss": 1.7754,
+      "step": 8141
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003728803402471058,
+      "loss": 1.5723,
+      "step": 8142
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037265592727055385,
+      "loss": 1.4043,
+      "step": 8143
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037243156637910345,
+      "loss": 1.498,
+      "step": 8144
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037220725759138263,
+      "loss": 1.4375,
+      "step": 8145
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037198300092601436,
+      "loss": 1.5039,
+      "step": 8146
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00037175879640161737,
+      "loss": 1.5254,
+      "step": 8147
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003715346440368059,
+      "loss": 1.5762,
+      "step": 8148
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0003713105438501909,
+      "loss": 1.6133,
+      "step": 8149
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003710864958603779,
+      "loss": 1.625,
+      "step": 8150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003708625000859683,
+      "loss": 1.6504,
+      "step": 8151
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003706385565455591,
+      "loss": 1.5664,
+      "step": 8152
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003704146652577437,
+      "loss": 1.5254,
+      "step": 8153
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003701908262411104,
+      "loss": 1.6484,
+      "step": 8154
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003699670395142435,
+      "loss": 1.6973,
+      "step": 8155
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003697433050957222,
+      "loss": 1.5684,
+      "step": 8156
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003695196230041229,
+      "loss": 1.7812,
+      "step": 8157
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003692959932580163,
+      "loss": 1.6602,
+      "step": 8158
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003690724158759694,
+      "loss": 1.709,
+      "step": 8159
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036884889087654405,
+      "loss": 1.627,
+      "step": 8160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036862541827829923,
+      "loss": 1.5703,
+      "step": 8161
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036840199809978834,
+      "loss": 1.5977,
+      "step": 8162
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036817863035956065,
+      "loss": 1.6016,
+      "step": 8163
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036795531507616107,
+      "loss": 1.5664,
+      "step": 8164
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000367732052268131,
+      "loss": 1.5957,
+      "step": 8165
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036750884195400625,
+      "loss": 1.541,
+      "step": 8166
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003672856841523189,
+      "loss": 1.6758,
+      "step": 8167
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036706257888159655,
+      "loss": 1.4629,
+      "step": 8168
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036683952616036256,
+      "loss": 1.666,
+      "step": 8169
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003666165260071357,
+      "loss": 1.7168,
+      "step": 8170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036639357844043033,
+      "loss": 1.5312,
+      "step": 8171
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003661706834787571,
+      "loss": 1.5645,
+      "step": 8172
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003659478411406216,
+      "loss": 1.4688,
+      "step": 8173
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003657250514445254,
+      "loss": 1.6191,
+      "step": 8174
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003655023144089652,
+      "loss": 1.752,
+      "step": 8175
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003652796300524337,
+      "loss": 1.5234,
+      "step": 8176
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036505699839341975,
+      "loss": 1.5195,
+      "step": 8177
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036483441945040694,
+      "loss": 1.5508,
+      "step": 8178
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003646118932418748,
+      "loss": 1.6797,
+      "step": 8179
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036438941978629834,
+      "loss": 1.5801,
+      "step": 8180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000364166999102149,
+      "loss": 1.5156,
+      "step": 8181
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003639446312078927,
+      "loss": 1.541,
+      "step": 8182
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003637223161219917,
+      "loss": 1.5586,
+      "step": 8183
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003635000538629032,
+      "loss": 1.625,
+      "step": 8184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003632778444490811,
+      "loss": 1.4434,
+      "step": 8185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036305568789897415,
+      "loss": 1.543,
+      "step": 8186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036283358423102664,
+      "loss": 1.5684,
+      "step": 8187
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003626115334636785,
+      "loss": 1.6113,
+      "step": 8188
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036238953561536593,
+      "loss": 1.6152,
+      "step": 8189
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003621675907045201,
+      "loss": 1.5391,
+      "step": 8190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036194569874956783,
+      "loss": 1.5605,
+      "step": 8191
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003617238597689314,
+      "loss": 1.6133,
+      "step": 8192
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003615020737810296,
+      "loss": 1.5312,
+      "step": 8193
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036128034080427576,
+      "loss": 1.4629,
+      "step": 8194
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036105866085707917,
+      "loss": 1.7285,
+      "step": 8195
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003608370339578446,
+      "loss": 1.4902,
+      "step": 8196
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036061546012497306,
+      "loss": 1.4375,
+      "step": 8197
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003603939393768605,
+      "loss": 1.5879,
+      "step": 8198
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00036017247173189814,
+      "loss": 1.7324,
+      "step": 8199
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035995105720847397,
+      "loss": 1.5996,
+      "step": 8200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003597296958249707,
+      "loss": 1.543,
+      "step": 8201
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035950838759976666,
+      "loss": 1.5488,
+      "step": 8202
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003592871325512357,
+      "loss": 1.623,
+      "step": 8203
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000359065930697748,
+      "loss": 1.5625,
+      "step": 8204
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035884478205766855,
+      "loss": 1.4863,
+      "step": 8205
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035862368664935806,
+      "loss": 1.6875,
+      "step": 8206
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003584026444911731,
+      "loss": 1.6016,
+      "step": 8207
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035818165560146554,
+      "loss": 1.5996,
+      "step": 8208
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035796071999858294,
+      "loss": 1.541,
+      "step": 8209
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035773983770086813,
+      "loss": 1.6406,
+      "step": 8210
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035751900872666044,
+      "loss": 1.5781,
+      "step": 8211
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003572982330942939,
+      "loss": 1.7305,
+      "step": 8212
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003570775108220982,
+      "loss": 1.541,
+      "step": 8213
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035685684192839886,
+      "loss": 1.5801,
+      "step": 8214
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003566362264315165,
+      "loss": 1.5039,
+      "step": 8215
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035641566434976846,
+      "loss": 1.6484,
+      "step": 8216
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035619515570146633,
+      "loss": 1.5332,
+      "step": 8217
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035597470050491786,
+      "loss": 1.6641,
+      "step": 8218
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000355754298778426,
+      "loss": 1.6523,
+      "step": 8219
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035553395054029025,
+      "loss": 1.5723,
+      "step": 8220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003553136558088046,
+      "loss": 1.5156,
+      "step": 8221
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035509341460225884,
+      "loss": 1.6406,
+      "step": 8222
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003548732269389384,
+      "loss": 1.5996,
+      "step": 8223
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003546530928371248,
+      "loss": 1.6934,
+      "step": 8224
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003544330123150942,
+      "loss": 1.6562,
+      "step": 8225
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003542129853911189,
+      "loss": 1.5215,
+      "step": 8226
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035399301208346626,
+      "loss": 1.5273,
+      "step": 8227
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003537730924104,
+      "loss": 1.6738,
+      "step": 8228
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035355322639017873,
+      "loss": 1.5879,
+      "step": 8229
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003533334140410568,
+      "loss": 1.6211,
+      "step": 8230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003531136553812836,
+      "loss": 1.709,
+      "step": 8231
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035289395042910533,
+      "loss": 1.5078,
+      "step": 8232
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003526742992027625,
+      "loss": 1.623,
+      "step": 8233
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035245470172049176,
+      "loss": 1.6602,
+      "step": 8234
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003522351580005247,
+      "loss": 1.543,
+      "step": 8235
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035201566806108964,
+      "loss": 1.498,
+      "step": 8236
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035179623192040923,
+      "loss": 1.6914,
+      "step": 8237
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003515768495967022,
+      "loss": 1.666,
+      "step": 8238
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003513575211081823,
+      "loss": 1.6152,
+      "step": 8239
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035113824647306003,
+      "loss": 1.5098,
+      "step": 8240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003509190257095402,
+      "loss": 1.6367,
+      "step": 8241
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003506998588358235,
+      "loss": 1.627,
+      "step": 8242
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035048074587010593,
+      "loss": 1.4863,
+      "step": 8243
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035026168683058,
+      "loss": 1.6562,
+      "step": 8244
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00035004268173543276,
+      "loss": 1.498,
+      "step": 8245
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003498237306028469,
+      "loss": 1.5762,
+      "step": 8246
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003496048334510009,
+      "loss": 1.5957,
+      "step": 8247
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003493859902980686,
+      "loss": 1.5898,
+      "step": 8248
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003491672011622195,
+      "loss": 1.6934,
+      "step": 8249
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00034894846606161803,
+      "loss": 1.4766,
+      "step": 8250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003487297850144254,
+      "loss": 1.5898,
+      "step": 8251
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003485111580387972,
+      "loss": 1.5234,
+      "step": 8252
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00034829258515288485,
+      "loss": 1.6387,
+      "step": 8253
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003480740663748353,
+      "loss": 1.5469,
+      "step": 8254
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00034785560172279073,
+      "loss": 1.6133,
+      "step": 8255
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00034763719121488993,
+      "loss": 1.5723,
+      "step": 8256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00034741883486926573,
+      "loss": 1.4434,
+      "step": 8257
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003472005327040474,
+      "loss": 1.457,
+      "step": 8258
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00034698228473735893,
+      "loss": 1.5586,
+      "step": 8259
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003467640909873211,
+      "loss": 1.6719,
+      "step": 8260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003465459514720489,
+      "loss": 1.5469,
+      "step": 8261
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0003463278662096535,
+      "loss": 1.6055,
+      "step": 8262
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003461098352182409,
+      "loss": 1.4355,
+      "step": 8263
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003458918585159139,
+      "loss": 1.6562,
+      "step": 8264
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00034567393612076945,
+      "loss": 1.5742,
+      "step": 8265
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003454560680509006,
+      "loss": 1.7188,
+      "step": 8266
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00034523825432439546,
+      "loss": 1.5293,
+      "step": 8267
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003450204949593386,
+      "loss": 1.5918,
+      "step": 8268
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003448027899738092,
+      "loss": 1.5312,
+      "step": 8269
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003445851393858821,
+      "loss": 1.6035,
+      "step": 8270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003443675432136274,
+      "loss": 1.5703,
+      "step": 8271
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003441500014751115,
+      "loss": 1.6406,
+      "step": 8272
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00034393251418839556,
+      "loss": 1.5508,
+      "step": 8273
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003437150813715364,
+      "loss": 1.4941,
+      "step": 8274
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00034349770304258596,
+      "loss": 1.5664,
+      "step": 8275
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00034328037921959275,
+      "loss": 1.6113,
+      "step": 8276
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00034306310992059975,
+      "loss": 1.5332,
+      "step": 8277
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003428458951636455,
+      "loss": 1.6465,
+      "step": 8278
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003426287349667639,
+      "loss": 1.5508,
+      "step": 8279
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00034241162934798555,
+      "loss": 1.5977,
+      "step": 8280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000342194578325335,
+      "loss": 1.6777,
+      "step": 8281
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003419775819168328,
+      "loss": 1.7012,
+      "step": 8282
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000341760640140495,
+      "loss": 1.5625,
+      "step": 8283
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003415437530143337,
+      "loss": 1.5098,
+      "step": 8284
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003413269205563554,
+      "loss": 1.6074,
+      "step": 8285
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00034111014278456276,
+      "loss": 1.5977,
+      "step": 8286
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003408934197169536,
+      "loss": 1.6484,
+      "step": 8287
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003406767513715214,
+      "loss": 1.459,
+      "step": 8288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003404601377662546,
+      "loss": 1.5918,
+      "step": 8289
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00034024357891913816,
+      "loss": 1.582,
+      "step": 8290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00034002707484815146,
+      "loss": 1.6387,
+      "step": 8291
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003398106255712699,
+      "loss": 1.5977,
+      "step": 8292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003395942311064638,
+      "loss": 1.5488,
+      "step": 8293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033937789147169917,
+      "loss": 1.5059,
+      "step": 8294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003391616066849381,
+      "loss": 1.6289,
+      "step": 8295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003389453767641375,
+      "loss": 1.6875,
+      "step": 8296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003387292017272495,
+      "loss": 1.5273,
+      "step": 8297
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033851308159222183,
+      "loss": 1.6484,
+      "step": 8298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003382970163769985,
+      "loss": 1.6211,
+      "step": 8299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033808100609951776,
+      "loss": 1.5566,
+      "step": 8300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033786505077771403,
+      "loss": 1.5801,
+      "step": 8301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003376491504295165,
+      "loss": 1.5117,
+      "step": 8302
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033743330507285087,
+      "loss": 1.5605,
+      "step": 8303
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033721751472563745,
+      "loss": 1.5508,
+      "step": 8304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033700177940579215,
+      "loss": 1.4609,
+      "step": 8305
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033678609913122596,
+      "loss": 1.4805,
+      "step": 8306
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003365704739198463,
+      "loss": 1.5996,
+      "step": 8307
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033635490378955527,
+      "loss": 1.625,
+      "step": 8308
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033613938875825037,
+      "loss": 1.6035,
+      "step": 8309
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003359239288438244,
+      "loss": 1.7031,
+      "step": 8310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003357085240641665,
+      "loss": 1.6016,
+      "step": 8311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003354931744371603,
+      "loss": 1.5977,
+      "step": 8312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033527787998068525,
+      "loss": 1.6074,
+      "step": 8313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003350626407126156,
+      "loss": 1.4922,
+      "step": 8314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003348474566508224,
+      "loss": 1.6523,
+      "step": 8315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003346323278131709,
+      "loss": 1.5645,
+      "step": 8316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003344172542175219,
+      "loss": 1.6348,
+      "step": 8317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003342022358817317,
+      "loss": 1.5625,
+      "step": 8318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033398727282365283,
+      "loss": 1.5781,
+      "step": 8319
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003337723650611322,
+      "loss": 1.4961,
+      "step": 8320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003335575126120124,
+      "loss": 1.6035,
+      "step": 8321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003333427154941312,
+      "loss": 1.6465,
+      "step": 8322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033312797372532277,
+      "loss": 1.5723,
+      "step": 8323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033291328732341566,
+      "loss": 1.543,
+      "step": 8324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003326986563062342,
+      "loss": 1.6953,
+      "step": 8325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033248408069159807,
+      "loss": 1.5098,
+      "step": 8326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003322695604973223,
+      "loss": 1.6973,
+      "step": 8327
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003320550957412171,
+      "loss": 1.7168,
+      "step": 8328
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000331840686441089,
+      "loss": 1.6895,
+      "step": 8329
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033162633261473893,
+      "loss": 1.6602,
+      "step": 8330
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003314120342799637,
+      "loss": 1.623,
+      "step": 8331
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003311977914545552,
+      "loss": 1.5996,
+      "step": 8332
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000330983604156301,
+      "loss": 1.6699,
+      "step": 8333
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003307694724029836,
+      "loss": 1.6934,
+      "step": 8334
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00033055539621238187,
+      "loss": 1.5605,
+      "step": 8335
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003303413756022692,
+      "loss": 1.6035,
+      "step": 8336
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003301274105904145,
+      "loss": 1.668,
+      "step": 8337
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032991350119458186,
+      "loss": 1.5312,
+      "step": 8338
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032969964743253177,
+      "loss": 1.6035,
+      "step": 8339
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003294858493220191,
+      "loss": 1.541,
+      "step": 8340
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032927210688079423,
+      "loss": 1.6523,
+      "step": 8341
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003290584201266028,
+      "loss": 1.668,
+      "step": 8342
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032884478907718685,
+      "loss": 1.4883,
+      "step": 8343
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003286312137502827,
+      "loss": 1.5391,
+      "step": 8344
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003284176941636224,
+      "loss": 1.4922,
+      "step": 8345
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032820423033493295,
+      "loss": 1.623,
+      "step": 8346
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032799082228193786,
+      "loss": 1.5527,
+      "step": 8347
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032777747002235493,
+      "loss": 1.5508,
+      "step": 8348
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003275641735738978,
+      "loss": 1.4766,
+      "step": 8349
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003273509329542748,
+      "loss": 1.582,
+      "step": 8350
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000327137748181191,
+      "loss": 1.502,
+      "step": 8351
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003269246192723457,
+      "loss": 1.6309,
+      "step": 8352
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032671154624543385,
+      "loss": 1.627,
+      "step": 8353
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003264985291181455,
+      "loss": 1.6074,
+      "step": 8354
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003262855679081669,
+      "loss": 1.6035,
+      "step": 8355
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003260726626331788,
+      "loss": 1.5234,
+      "step": 8356
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003258598133108577,
+      "loss": 1.6172,
+      "step": 8357
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032564701995887503,
+      "loss": 1.5898,
+      "step": 8358
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003254342825948985,
+      "loss": 1.6484,
+      "step": 8359
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003252216012365903,
+      "loss": 1.5742,
+      "step": 8360
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032500897590160817,
+      "loss": 1.6191,
+      "step": 8361
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032479640660760514,
+      "loss": 1.5625,
+      "step": 8362
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003245838933722302,
+      "loss": 1.6504,
+      "step": 8363
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032437143621312693,
+      "loss": 1.7148,
+      "step": 8364
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032415903514793456,
+      "loss": 1.5703,
+      "step": 8365
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032394669019428755,
+      "loss": 1.6191,
+      "step": 8366
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000323734401369816,
+      "loss": 1.6465,
+      "step": 8367
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003235221686921448,
+      "loss": 1.6387,
+      "step": 8368
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003233099921788949,
+      "loss": 1.5898,
+      "step": 8369
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032309787184768216,
+      "loss": 1.5938,
+      "step": 8370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00032288580771611765,
+      "loss": 1.5918,
+      "step": 8371
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000322673799801808,
+      "loss": 1.5039,
+      "step": 8372
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003224618481223548,
+      "loss": 1.4512,
+      "step": 8373
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0003222499526953561,
+      "loss": 1.7031,
+      "step": 8374
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00032203811353840385,
+      "loss": 1.5977,
+      "step": 8375
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00032182633066908607,
+      "loss": 1.748,
+      "step": 8376
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003216146041049858,
+      "loss": 1.7109,
+      "step": 8377
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000321402933863682,
+      "loss": 1.5723,
+      "step": 8378
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00032119131996274843,
+      "loss": 1.6914,
+      "step": 8379
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003209797624197542,
+      "loss": 1.5625,
+      "step": 8380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00032076826125226345,
+      "loss": 1.5625,
+      "step": 8381
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00032055681647783673,
+      "loss": 1.5957,
+      "step": 8382
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000320345428114029,
+      "loss": 1.6504,
+      "step": 8383
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003201340961783904,
+      "loss": 1.5625,
+      "step": 8384
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031992282068846667,
+      "loss": 1.5898,
+      "step": 8385
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031971160166179946,
+      "loss": 1.5957,
+      "step": 8386
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031950043911592495,
+      "loss": 1.6387,
+      "step": 8387
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003192893330683747,
+      "loss": 1.7051,
+      "step": 8388
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031907828353667557,
+      "loss": 1.6016,
+      "step": 8389
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003188672905383504,
+      "loss": 1.6445,
+      "step": 8390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031865635409091664,
+      "loss": 1.543,
+      "step": 8391
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003184454742118872,
+      "loss": 1.502,
+      "step": 8392
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031823465091877003,
+      "loss": 1.5977,
+      "step": 8393
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003180238842290694,
+      "loss": 1.543,
+      "step": 8394
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031781317416028354,
+      "loss": 1.5488,
+      "step": 8395
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003176025207299069,
+      "loss": 1.6523,
+      "step": 8396
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003173919239554286,
+      "loss": 1.5605,
+      "step": 8397
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031718138385433394,
+      "loss": 1.6602,
+      "step": 8398
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003169709004441027,
+      "loss": 1.6973,
+      "step": 8399
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003167604737422102,
+      "loss": 1.4844,
+      "step": 8400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003165501037661267,
+      "loss": 1.4668,
+      "step": 8401
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031633979053331883,
+      "loss": 1.5234,
+      "step": 8402
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031612953406124744,
+      "loss": 1.4785,
+      "step": 8403
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000315919334367369,
+      "loss": 1.4922,
+      "step": 8404
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031570919146913537,
+      "loss": 1.6367,
+      "step": 8405
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031549910538399353,
+      "loss": 1.6328,
+      "step": 8406
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031528907612938563,
+      "loss": 1.4375,
+      "step": 8407
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031507910372274985,
+      "loss": 1.4707,
+      "step": 8408
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031486918818151877,
+      "loss": 1.6055,
+      "step": 8409
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003146593295231206,
+      "loss": 1.6543,
+      "step": 8410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003144495277649789,
+      "loss": 1.623,
+      "step": 8411
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003142397829245123,
+      "loss": 1.5781,
+      "step": 8412
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031403009501913446,
+      "loss": 1.459,
+      "step": 8413
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003138204640662555,
+      "loss": 1.4707,
+      "step": 8414
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031361089008327946,
+      "loss": 1.709,
+      "step": 8415
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003134013730876063,
+      "loss": 1.6562,
+      "step": 8416
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031319191309663074,
+      "loss": 1.6152,
+      "step": 8417
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003129825101277439,
+      "loss": 1.582,
+      "step": 8418
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003127731641983309,
+      "loss": 1.6523,
+      "step": 8419
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031256387532577277,
+      "loss": 1.6289,
+      "step": 8420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031235464352744545,
+      "loss": 1.5215,
+      "step": 8421
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003121454688207207,
+      "loss": 1.5371,
+      "step": 8422
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003119363512229653,
+      "loss": 1.6465,
+      "step": 8423
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003117272907515407,
+      "loss": 1.5391,
+      "step": 8424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031151828742380415,
+      "loss": 1.6543,
+      "step": 8425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031130934125710863,
+      "loss": 1.541,
+      "step": 8426
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003111004522688016,
+      "loss": 1.5449,
+      "step": 8427
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003108916204762259,
+      "loss": 1.4492,
+      "step": 8428
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003106828458967194,
+      "loss": 1.6855,
+      "step": 8429
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003104741285476164,
+      "loss": 1.6367,
+      "step": 8430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00031026546844624506,
+      "loss": 1.5762,
+      "step": 8431
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003100568656099295,
+      "loss": 1.6309,
+      "step": 8432
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003098483200559885,
+      "loss": 1.582,
+      "step": 8433
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030963983180173725,
+      "loss": 1.4648,
+      "step": 8434
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003094314008644851,
+      "loss": 1.6016,
+      "step": 8435
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030922302726153696,
+      "loss": 1.6211,
+      "step": 8436
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030901471101019265,
+      "loss": 1.5879,
+      "step": 8437
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003088064521277482,
+      "loss": 1.623,
+      "step": 8438
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030859825063149406,
+      "loss": 1.6699,
+      "step": 8439
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003083901065387159,
+      "loss": 1.5508,
+      "step": 8440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003081820198666948,
+      "loss": 1.5371,
+      "step": 8441
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003079739906327076,
+      "loss": 1.6562,
+      "step": 8442
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030776601885402544,
+      "loss": 1.6641,
+      "step": 8443
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030755810454791534,
+      "loss": 1.4902,
+      "step": 8444
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003073502477316392,
+      "loss": 1.6699,
+      "step": 8445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030714244842245444,
+      "loss": 1.5801,
+      "step": 8446
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030693470663761303,
+      "loss": 1.6152,
+      "step": 8447
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030672702239436346,
+      "loss": 1.6055,
+      "step": 8448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030651939570994837,
+      "loss": 1.625,
+      "step": 8449
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003063118266016057,
+      "loss": 1.5234,
+      "step": 8450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003061043150865692,
+      "loss": 1.5781,
+      "step": 8451
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030589686118206685,
+      "loss": 1.5352,
+      "step": 8452
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003056894649053231,
+      "loss": 1.502,
+      "step": 8453
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030548212627355686,
+      "loss": 1.5469,
+      "step": 8454
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003052748453039822,
+      "loss": 1.541,
+      "step": 8455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003050676220138084,
+      "loss": 1.5195,
+      "step": 8456
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030486045642024064,
+      "loss": 1.5859,
+      "step": 8457
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030465334854047854,
+      "loss": 1.5371,
+      "step": 8458
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030444629839171725,
+      "loss": 1.627,
+      "step": 8459
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030423930599114647,
+      "loss": 1.7227,
+      "step": 8460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030403237135595264,
+      "loss": 1.6406,
+      "step": 8461
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003038254945033161,
+      "loss": 1.5977,
+      "step": 8462
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003036186754504127,
+      "loss": 1.5566,
+      "step": 8463
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030341191421441306,
+      "loss": 1.4863,
+      "step": 8464
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003032052108124843,
+      "loss": 1.5371,
+      "step": 8465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003029985652617877,
+      "loss": 1.5332,
+      "step": 8466
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030279197757947983,
+      "loss": 1.5625,
+      "step": 8467
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030258544778271224,
+      "loss": 1.5957,
+      "step": 8468
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003023789758886327,
+      "loss": 1.6309,
+      "step": 8469
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030217256191438315,
+      "loss": 1.5371,
+      "step": 8470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003019662058771011,
+      "loss": 1.6641,
+      "step": 8471
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030175990779391894,
+      "loss": 1.668,
+      "step": 8472
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030155366768196514,
+      "loss": 1.5059,
+      "step": 8473
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003013474855583622,
+      "loss": 1.5059,
+      "step": 8474
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030114136144022876,
+      "loss": 1.5898,
+      "step": 8475
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030093529534467743,
+      "loss": 1.502,
+      "step": 8476
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030072928728881786,
+      "loss": 1.6523,
+      "step": 8477
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00030052333728975333,
+      "loss": 1.5078,
+      "step": 8478
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003003174453645827,
+      "loss": 1.4902,
+      "step": 8479
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0003001116115304,
+      "loss": 1.541,
+      "step": 8480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029990583580429496,
+      "loss": 1.6152,
+      "step": 8481
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029970011820335186,
+      "loss": 1.5898,
+      "step": 8482
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002994944587446504,
+      "loss": 1.6738,
+      "step": 8483
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002992888574452652,
+      "loss": 1.5742,
+      "step": 8484
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029908331432226663,
+      "loss": 1.6348,
+      "step": 8485
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029887782939271913,
+      "loss": 1.6055,
+      "step": 8486
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000298672402673684,
+      "loss": 1.4766,
+      "step": 8487
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002984670341822162,
+      "loss": 1.625,
+      "step": 8488
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029826172393536654,
+      "loss": 1.6426,
+      "step": 8489
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002980564719501808,
+      "loss": 1.5898,
+      "step": 8490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029785127824370007,
+      "loss": 1.6191,
+      "step": 8491
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002976461428329601,
+      "loss": 1.709,
+      "step": 8492
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002974410657349929,
+      "loss": 1.6133,
+      "step": 8493
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002972360469668245,
+      "loss": 1.6387,
+      "step": 8494
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029703108654547674,
+      "loss": 1.623,
+      "step": 8495
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029682618448796595,
+      "loss": 1.5156,
+      "step": 8496
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029662134081130475,
+      "loss": 1.541,
+      "step": 8497
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002964165555324999,
+      "loss": 1.6016,
+      "step": 8498
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002962118286685538,
+      "loss": 1.5742,
+      "step": 8499
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029600716023646335,
+      "loss": 1.4492,
+      "step": 8500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002958025502532218,
+      "loss": 1.4707,
+      "step": 8501
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002955979987358165,
+      "loss": 1.4902,
+      "step": 8502
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002953935057012305,
+      "loss": 1.5703,
+      "step": 8503
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029518907116644113,
+      "loss": 1.5254,
+      "step": 8504
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002949846951484224,
+      "loss": 1.5508,
+      "step": 8505
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002947803776641421,
+      "loss": 1.752,
+      "step": 8506
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002945761187305639,
+      "loss": 1.4785,
+      "step": 8507
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002943719183646457,
+      "loss": 1.4941,
+      "step": 8508
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029416777658334214,
+      "loss": 1.5547,
+      "step": 8509
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002939636934036015,
+      "loss": 1.4238,
+      "step": 8510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029375966884236793,
+      "loss": 1.6309,
+      "step": 8511
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029355570291657995,
+      "loss": 1.4805,
+      "step": 8512
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029335179564317273,
+      "loss": 1.6074,
+      "step": 8513
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002931479470390751,
+      "loss": 1.6523,
+      "step": 8514
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002929441571212116,
+      "loss": 1.4121,
+      "step": 8515
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002927404259065014,
+      "loss": 1.6992,
+      "step": 8516
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029253675341186017,
+      "loss": 1.6895,
+      "step": 8517
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029233313965419715,
+      "loss": 1.582,
+      "step": 8518
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002921295846504176,
+      "loss": 1.4805,
+      "step": 8519
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002919260884174211,
+      "loss": 1.4434,
+      "step": 8520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002917226509721035,
+      "loss": 1.8359,
+      "step": 8521
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000291519272331355,
+      "loss": 1.5352,
+      "step": 8522
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000291315952512061,
+      "loss": 1.5762,
+      "step": 8523
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029111269153110197,
+      "loss": 1.582,
+      "step": 8524
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029090948940535376,
+      "loss": 1.4766,
+      "step": 8525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002907063461516869,
+      "loss": 1.6387,
+      "step": 8526
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002905032617869677,
+      "loss": 1.4648,
+      "step": 8527
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002903002363280571,
+      "loss": 1.6426,
+      "step": 8528
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002900972697918112,
+      "loss": 1.5703,
+      "step": 8529
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028989436219508123,
+      "loss": 1.6523,
+      "step": 8530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002896915135547135,
+      "loss": 1.5312,
+      "step": 8531
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002894887238875492,
+      "loss": 1.4707,
+      "step": 8532
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002892859932104256,
+      "loss": 1.6562,
+      "step": 8533
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002890833215401739,
+      "loss": 1.6211,
+      "step": 8534
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028888070889362093,
+      "loss": 1.5059,
+      "step": 8535
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028867815528758835,
+      "loss": 1.6387,
+      "step": 8536
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002884756607388936,
+      "loss": 1.4434,
+      "step": 8537
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028827322526434865,
+      "loss": 1.5566,
+      "step": 8538
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002880708488807601,
+      "loss": 1.4414,
+      "step": 8539
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028786853160493085,
+      "loss": 1.541,
+      "step": 8540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028766627345365816,
+      "loss": 1.541,
+      "step": 8541
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028746407444373424,
+      "loss": 1.5586,
+      "step": 8542
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002872619345919464,
+      "loss": 1.5312,
+      "step": 8543
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028705985391507796,
+      "loss": 1.5176,
+      "step": 8544
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028685783242990605,
+      "loss": 1.5332,
+      "step": 8545
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002866558701532038,
+      "loss": 1.584,
+      "step": 8546
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002864539671017385,
+      "loss": 1.6953,
+      "step": 8547
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028625212329227393,
+      "loss": 1.584,
+      "step": 8548
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002860503387415676,
+      "loss": 1.5918,
+      "step": 8549
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002858486134663728,
+      "loss": 1.4258,
+      "step": 8550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002856469474834373,
+      "loss": 1.6191,
+      "step": 8551
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028544534080950513,
+      "loss": 1.5215,
+      "step": 8552
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028524379346131425,
+      "loss": 1.6699,
+      "step": 8553
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028504230545559804,
+      "loss": 1.584,
+      "step": 8554
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002848408768090848,
+      "loss": 1.6035,
+      "step": 8555
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028463950753849853,
+      "loss": 1.5547,
+      "step": 8556
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028443819766055776,
+      "loss": 1.5586,
+      "step": 8557
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002842369471919761,
+      "loss": 1.5898,
+      "step": 8558
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002840357561494621,
+      "loss": 1.5996,
+      "step": 8559
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002838346245497201,
+      "loss": 1.5898,
+      "step": 8560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002836335524094488,
+      "loss": 1.5566,
+      "step": 8561
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028343253974534214,
+      "loss": 1.5469,
+      "step": 8562
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000283231586574089,
+      "loss": 1.4746,
+      "step": 8563
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002830306929123737,
+      "loss": 1.625,
+      "step": 8564
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028282985877687493,
+      "loss": 1.4844,
+      "step": 8565
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002826290841842676,
+      "loss": 1.5703,
+      "step": 8566
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002824283691512206,
+      "loss": 1.6367,
+      "step": 8567
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028222771369439836,
+      "loss": 1.5898,
+      "step": 8568
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002820271178304602,
+      "loss": 1.6113,
+      "step": 8569
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028182658157606047,
+      "loss": 1.4609,
+      "step": 8570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028162610494784847,
+      "loss": 1.5117,
+      "step": 8571
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028142568796246913,
+      "loss": 1.5762,
+      "step": 8572
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002812253306365621,
+      "loss": 1.5469,
+      "step": 8573
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002810250329867616,
+      "loss": 1.5156,
+      "step": 8574
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002808247950296972,
+      "loss": 1.5137,
+      "step": 8575
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002806246167819941,
+      "loss": 1.5781,
+      "step": 8576
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028042449826027195,
+      "loss": 1.5625,
+      "step": 8577
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002802244394811454,
+      "loss": 1.5449,
+      "step": 8578
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000280024440461224,
+      "loss": 1.6289,
+      "step": 8579
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00027982450121711325,
+      "loss": 1.5801,
+      "step": 8580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002796246217654127,
+      "loss": 1.6152,
+      "step": 8581
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002794248021227174,
+      "loss": 1.6191,
+      "step": 8582
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002792250423056168,
+      "loss": 1.5957,
+      "step": 8583
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00027902534233069685,
+      "loss": 1.6211,
+      "step": 8584
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002788257022145372,
+      "loss": 1.5117,
+      "step": 8585
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00027862612197371273,
+      "loss": 1.6328,
+      "step": 8586
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002784266016247935,
+      "loss": 1.5684,
+      "step": 8587
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000278227141184345,
+      "loss": 1.6836,
+      "step": 8588
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002780277406689272,
+      "loss": 1.5703,
+      "step": 8589
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002778284000950954,
+      "loss": 1.6641,
+      "step": 8590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002776291194793993,
+      "loss": 1.5215,
+      "step": 8591
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002774298988383849,
+      "loss": 1.3984,
+      "step": 8592
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000277230738188592,
+      "loss": 1.5684,
+      "step": 8593
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00027703163754655603,
+      "loss": 1.6035,
+      "step": 8594
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002768325969288068,
+      "loss": 1.6113,
+      "step": 8595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00027663361635187035,
+      "loss": 1.5371,
+      "step": 8596
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00027643469583226663,
+      "loss": 1.5703,
+      "step": 8597
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002762358353865111,
+      "loss": 1.4902,
+      "step": 8598
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00027603703503111356,
+      "loss": 1.627,
+      "step": 8599
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00027583829478258017,
+      "loss": 1.5605,
+      "step": 8600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00027563961465741096,
+      "loss": 1.6094,
+      "step": 8601
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00027544099467210115,
+      "loss": 1.6133,
+      "step": 8602
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002752424348431414,
+      "loss": 1.7188,
+      "step": 8603
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002750439351870164,
+      "loss": 1.5781,
+      "step": 8604
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002748454957202076,
+      "loss": 1.5137,
+      "step": 8605
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00027464711645918983,
+      "loss": 1.5957,
+      "step": 8606
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002744487974204334,
+      "loss": 1.5625,
+      "step": 8607
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00027425053862040395,
+      "loss": 1.5801,
+      "step": 8608
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002740523400755617,
+      "loss": 1.5781,
+      "step": 8609
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002738542018023621,
+      "loss": 1.5996,
+      "step": 8610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002736561238172552,
+      "loss": 1.582,
+      "step": 8611
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002734581061366871,
+      "loss": 1.6055,
+      "step": 8612
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00027326014877709757,
+      "loss": 1.6133,
+      "step": 8613
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002730622517549224,
+      "loss": 1.4668,
+      "step": 8614
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00027286441508659134,
+      "loss": 1.4375,
+      "step": 8615
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002726666387885305,
+      "loss": 1.5586,
+      "step": 8616
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002724689228771597,
+      "loss": 1.5293,
+      "step": 8617
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00027227126736889453,
+      "loss": 1.5898,
+      "step": 8618
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002720736722801448,
+      "loss": 1.5195,
+      "step": 8619
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00027187613762731644,
+      "loss": 1.6523,
+      "step": 8620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00027167866342680944,
+      "loss": 1.6465,
+      "step": 8621
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00027148124969501874,
+      "loss": 1.7539,
+      "step": 8622
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002712838964483353,
+      "loss": 1.5801,
+      "step": 8623
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00027108660370314387,
+      "loss": 1.5938,
+      "step": 8624
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002708893714758246,
+      "loss": 1.666,
+      "step": 8625
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002706921997827525,
+      "loss": 1.6582,
+      "step": 8626
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00027049508864029827,
+      "loss": 1.5605,
+      "step": 8627
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00027029803806482655,
+      "loss": 1.625,
+      "step": 8628
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00027010104807269765,
+      "loss": 1.5723,
+      "step": 8629
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002699041186802662,
+      "loss": 1.4531,
+      "step": 8630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026970724990388296,
+      "loss": 1.6172,
+      "step": 8631
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002695104417598924,
+      "loss": 1.5762,
+      "step": 8632
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026931369426463447,
+      "loss": 1.5234,
+      "step": 8633
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000269117007434444,
+      "loss": 1.6543,
+      "step": 8634
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026892038128565146,
+      "loss": 1.5391,
+      "step": 8635
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026872381583458116,
+      "loss": 1.5234,
+      "step": 8636
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000268527311097553,
+      "loss": 1.627,
+      "step": 8637
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002683308670908816,
+      "loss": 1.6895,
+      "step": 8638
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002681344838308771,
+      "loss": 1.543,
+      "step": 8639
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026793816133384397,
+      "loss": 1.6152,
+      "step": 8640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026774189961608177,
+      "loss": 1.5469,
+      "step": 8641
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026754569869388514,
+      "loss": 1.584,
+      "step": 8642
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002673495585835436,
+      "loss": 1.6309,
+      "step": 8643
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026715347930134125,
+      "loss": 1.5703,
+      "step": 8644
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002669574608635583,
+      "loss": 1.6113,
+      "step": 8645
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002667615032864688,
+      "loss": 1.5312,
+      "step": 8646
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000266565606586342,
+      "loss": 1.5547,
+      "step": 8647
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026636977077944214,
+      "loss": 1.5684,
+      "step": 8648
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002661739958820286,
+      "loss": 1.5371,
+      "step": 8649
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026597828191035524,
+      "loss": 1.6113,
+      "step": 8650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002657826288806715,
+      "loss": 1.6289,
+      "step": 8651
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002655870368092215,
+      "loss": 1.4531,
+      "step": 8652
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000265391505712244,
+      "loss": 1.6113,
+      "step": 8653
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002651960356059726,
+      "loss": 1.6309,
+      "step": 8654
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002650006265066369,
+      "loss": 1.6035,
+      "step": 8655
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026480527843046045,
+      "loss": 1.4766,
+      "step": 8656
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026460999139366173,
+      "loss": 1.4785,
+      "step": 8657
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026441476541245437,
+      "loss": 1.6582,
+      "step": 8658
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002642196005030473,
+      "loss": 1.5957,
+      "step": 8659
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026402449668164397,
+      "loss": 1.6406,
+      "step": 8660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002638294539644427,
+      "loss": 1.668,
+      "step": 8661
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002636344723676366,
+      "loss": 1.5977,
+      "step": 8662
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026343955190741465,
+      "loss": 1.6855,
+      "step": 8663
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026324469259995963,
+      "loss": 1.5703,
+      "step": 8664
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002630498944614498,
+      "loss": 1.5625,
+      "step": 8665
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002628551575080579,
+      "loss": 1.6191,
+      "step": 8666
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026266048175595247,
+      "loss": 1.502,
+      "step": 8667
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026246586722129627,
+      "loss": 1.5547,
+      "step": 8668
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026227131392024707,
+      "loss": 1.6621,
+      "step": 8669
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002620768218689571,
+      "loss": 1.4668,
+      "step": 8670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026188239108357493,
+      "loss": 1.5195,
+      "step": 8671
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026168802158024274,
+      "loss": 1.6094,
+      "step": 8672
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002614937133750979,
+      "loss": 1.707,
+      "step": 8673
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002612994664842726,
+      "loss": 1.5391,
+      "step": 8674
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026110528092389474,
+      "loss": 1.6133,
+      "step": 8675
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026091115671008616,
+      "loss": 1.4629,
+      "step": 8676
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026071709385896405,
+      "loss": 1.6348,
+      "step": 8677
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002605230923866402,
+      "loss": 1.6289,
+      "step": 8678
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00026032915230922194,
+      "loss": 1.6094,
+      "step": 8679
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002601352736428111,
+      "loss": 1.7031,
+      "step": 8680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002599414564035041,
+      "loss": 1.6465,
+      "step": 8681
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00025974770060739274,
+      "loss": 1.6465,
+      "step": 8682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002595540062705632,
+      "loss": 1.4551,
+      "step": 8683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00025936037340909757,
+      "loss": 1.6582,
+      "step": 8684
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00025916680203907184,
+      "loss": 1.498,
+      "step": 8685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00025897329217655716,
+      "loss": 1.5469,
+      "step": 8686
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002587798438376198,
+      "loss": 1.5723,
+      "step": 8687
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002585864570383206,
+      "loss": 1.4688,
+      "step": 8688
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002583931317947156,
+      "loss": 1.7734,
+      "step": 8689
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002581998681228552,
+      "loss": 1.582,
+      "step": 8690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002580066660387856,
+      "loss": 1.6328,
+      "step": 8691
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00025781352555854733,
+      "loss": 1.3945,
+      "step": 8692
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00025762044669817543,
+      "loss": 1.5918,
+      "step": 8693
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002574274294737003,
+      "loss": 1.4746,
+      "step": 8694
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002572344739011476,
+      "loss": 1.6621,
+      "step": 8695
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002570415799965371,
+      "loss": 1.5898,
+      "step": 8696
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00025684874777588375,
+      "loss": 1.5371,
+      "step": 8697
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002566559772551972,
+      "loss": 1.5371,
+      "step": 8698
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002564632684504827,
+      "loss": 1.6738,
+      "step": 8699
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002562706213777397,
+      "loss": 1.5352,
+      "step": 8700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00025607803605296256,
+      "loss": 1.6602,
+      "step": 8701
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002558855124921402,
+      "loss": 1.4824,
+      "step": 8702
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002556930507112577,
+      "loss": 1.7012,
+      "step": 8703
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002555006507262937,
+      "loss": 1.5039,
+      "step": 8704
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00025530831255322217,
+      "loss": 1.5527,
+      "step": 8705
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00025511603620801185,
+      "loss": 1.582,
+      "step": 8706
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00025492382170662673,
+      "loss": 1.5957,
+      "step": 8707
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002547316690650252,
+      "loss": 1.5996,
+      "step": 8708
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00025453957829916054,
+      "loss": 1.5254,
+      "step": 8709
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002543475494249814,
+      "loss": 1.6895,
+      "step": 8710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002541555824584308,
+      "loss": 1.6289,
+      "step": 8711
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00025396367741544677,
+      "loss": 1.5039,
+      "step": 8712
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002537718343119617,
+      "loss": 1.5332,
+      "step": 8713
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002535800531639041,
+      "loss": 1.5723,
+      "step": 8714
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002533883339871962,
+      "loss": 1.5488,
+      "step": 8715
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002531966767977554,
+      "loss": 1.5703,
+      "step": 8716
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002530050816114938,
+      "loss": 1.668,
+      "step": 8717
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00025281354844431905,
+      "loss": 1.6289,
+      "step": 8718
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002526220773121328,
+      "loss": 1.4922,
+      "step": 8719
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00025243066823083204,
+      "loss": 1.5449,
+      "step": 8720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00025223932121630846,
+      "loss": 1.5547,
+      "step": 8721
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002520480362844485,
+      "loss": 1.5293,
+      "step": 8722
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00025185681345113334,
+      "loss": 1.6289,
+      "step": 8723
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002516656527322396,
+      "loss": 1.4355,
+      "step": 8724
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002514745541436384,
+      "loss": 1.5176,
+      "step": 8725
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002512835177011954,
+      "loss": 1.5625,
+      "step": 8726
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002510925434207715,
+      "loss": 1.709,
+      "step": 8727
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002509016313182222,
+      "loss": 1.6016,
+      "step": 8728
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00025071078140939776,
+      "loss": 1.6934,
+      "step": 8729
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002505199937101438,
+      "loss": 1.6133,
+      "step": 8730
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00025032926823630045,
+      "loss": 1.5801,
+      "step": 8731
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002501386050037023,
+      "loss": 1.6133,
+      "step": 8732
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002499480040281792,
+      "loss": 1.7051,
+      "step": 8733
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002497574653255561,
+      "loss": 1.6328,
+      "step": 8734
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024956698891165217,
+      "loss": 1.5625,
+      "step": 8735
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002493765748022817,
+      "loss": 1.6621,
+      "step": 8736
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024918622301325357,
+      "loss": 1.5391,
+      "step": 8737
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002489959335603721,
+      "loss": 1.627,
+      "step": 8738
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002488057064594359,
+      "loss": 1.6328,
+      "step": 8739
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024861554172623835,
+      "loss": 1.6504,
+      "step": 8740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002484254393765678,
+      "loss": 1.5586,
+      "step": 8741
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002482353994262079,
+      "loss": 1.582,
+      "step": 8742
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024804542189093626,
+      "loss": 1.6484,
+      "step": 8743
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002478555067865259,
+      "loss": 1.6055,
+      "step": 8744
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002476656541287441,
+      "loss": 1.6875,
+      "step": 8745
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000247475863933354,
+      "loss": 1.4453,
+      "step": 8746
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002472861362161125,
+      "loss": 1.5391,
+      "step": 8747
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002470964709927719,
+      "loss": 1.4863,
+      "step": 8748
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002469068682790785,
+      "loss": 1.7129,
+      "step": 8749
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024671732809077485,
+      "loss": 1.5723,
+      "step": 8750
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000246527850443597,
+      "loss": 1.584,
+      "step": 8751
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002463384353532766,
+      "loss": 1.7812,
+      "step": 8752
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002461490828355392,
+      "loss": 1.4805,
+      "step": 8753
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024595979290610635,
+      "loss": 1.5957,
+      "step": 8754
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002457705655806938,
+      "loss": 1.4668,
+      "step": 8755
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002455814008750118,
+      "loss": 1.5684,
+      "step": 8756
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024539229880476553,
+      "loss": 1.5176,
+      "step": 8757
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002452032593856557,
+      "loss": 1.5742,
+      "step": 8758
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002450142826333771,
+      "loss": 1.5566,
+      "step": 8759
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024482536856361924,
+      "loss": 1.5996,
+      "step": 8760
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002446365171920669,
+      "loss": 1.5938,
+      "step": 8761
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002444477285343991,
+      "loss": 1.6328,
+      "step": 8762
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024425900260629043,
+      "loss": 1.5762,
+      "step": 8763
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002440703394234096,
+      "loss": 1.5039,
+      "step": 8764
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002438817390014204,
+      "loss": 1.5566,
+      "step": 8765
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024369320135598116,
+      "loss": 1.4961,
+      "step": 8766
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024350472650274535,
+      "loss": 1.5059,
+      "step": 8767
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024331631445736102,
+      "loss": 1.5078,
+      "step": 8768
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024312796523547064,
+      "loss": 1.6133,
+      "step": 8769
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002429396788527125,
+      "loss": 1.5449,
+      "step": 8770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024275145532471888,
+      "loss": 1.543,
+      "step": 8771
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002425632946671168,
+      "loss": 1.5781,
+      "step": 8772
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000242375196895528,
+      "loss": 1.6074,
+      "step": 8773
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002421871620255699,
+      "loss": 1.6777,
+      "step": 8774
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024199919007285387,
+      "loss": 1.6289,
+      "step": 8775
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000241811281052986,
+      "loss": 1.4922,
+      "step": 8776
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024162343498156726,
+      "loss": 1.6816,
+      "step": 8777
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024143565187419391,
+      "loss": 1.7676,
+      "step": 8778
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024124793174645664,
+      "loss": 1.5723,
+      "step": 8779
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024106027461394064,
+      "loss": 1.6504,
+      "step": 8780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024087268049222588,
+      "loss": 1.6309,
+      "step": 8781
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002406851493968879,
+      "loss": 1.5781,
+      "step": 8782
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024049768134349613,
+      "loss": 1.668,
+      "step": 8783
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00024031027634761505,
+      "loss": 1.6074,
+      "step": 8784
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002401229344248036,
+      "loss": 1.5957,
+      "step": 8785
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023993565559061647,
+      "loss": 1.5449,
+      "step": 8786
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023974843986060212,
+      "loss": 1.5898,
+      "step": 8787
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000239561287250304,
+      "loss": 1.6465,
+      "step": 8788
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023937419777526027,
+      "loss": 1.541,
+      "step": 8789
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023918717145100443,
+      "loss": 1.5059,
+      "step": 8790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023900020829306412,
+      "loss": 1.7344,
+      "step": 8791
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023881330831696158,
+      "loss": 1.6836,
+      "step": 8792
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002386264715382148,
+      "loss": 1.5176,
+      "step": 8793
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023843969797233546,
+      "loss": 1.668,
+      "step": 8794
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023825298763483039,
+      "loss": 1.5273,
+      "step": 8795
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023806634054120092,
+      "loss": 1.5293,
+      "step": 8796
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023787975670694396,
+      "loss": 1.5898,
+      "step": 8797
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023769323614755033,
+      "loss": 1.5801,
+      "step": 8798
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023750677887850592,
+      "loss": 1.666,
+      "step": 8799
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023732038491529106,
+      "loss": 1.6309,
+      "step": 8800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023713405427338132,
+      "loss": 1.6289,
+      "step": 8801
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023694778696824637,
+      "loss": 1.5957,
+      "step": 8802
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023676158301535167,
+      "loss": 1.7207,
+      "step": 8803
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023657544243015628,
+      "loss": 1.6289,
+      "step": 8804
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023638936522811461,
+      "loss": 1.541,
+      "step": 8805
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023620335142467575,
+      "loss": 1.5195,
+      "step": 8806
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023601740103528336,
+      "loss": 1.6133,
+      "step": 8807
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023583151407537562,
+      "loss": 1.6836,
+      "step": 8808
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002356456905603863,
+      "loss": 1.5742,
+      "step": 8809
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023545993050574332,
+      "loss": 1.4922,
+      "step": 8810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023527423392686908,
+      "loss": 1.6367,
+      "step": 8811
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002350886008391807,
+      "loss": 1.5684,
+      "step": 8812
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002349030312580912,
+      "loss": 1.5898,
+      "step": 8813
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002347175251990068,
+      "loss": 1.457,
+      "step": 8814
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002345320826773294,
+      "loss": 1.6426,
+      "step": 8815
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002343467037084549,
+      "loss": 1.5879,
+      "step": 8816
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002341613883077749,
+      "loss": 1.5391,
+      "step": 8817
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023397613649067496,
+      "loss": 1.7246,
+      "step": 8818
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023379094827253544,
+      "loss": 1.6309,
+      "step": 8819
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002336058236687314,
+      "loss": 1.5703,
+      "step": 8820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002334207626946333,
+      "loss": 1.584,
+      "step": 8821
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023323576536560554,
+      "loss": 1.584,
+      "step": 8822
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023305083169700735,
+      "loss": 1.6406,
+      "step": 8823
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00023286596170419261,
+      "loss": 1.6191,
+      "step": 8824
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00023268115540251078,
+      "loss": 1.4434,
+      "step": 8825
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00023249641280730484,
+      "loss": 1.668,
+      "step": 8826
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00023231173393391326,
+      "loss": 1.5527,
+      "step": 8827
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00023212711879766845,
+      "loss": 1.5039,
+      "step": 8828
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00023194256741389875,
+      "loss": 1.5332,
+      "step": 8829
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00023175807979792617,
+      "loss": 1.6543,
+      "step": 8830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00023157365596506785,
+      "loss": 1.5605,
+      "step": 8831
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00023138929593063506,
+      "loss": 1.5703,
+      "step": 8832
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00023120499970993492,
+      "loss": 1.5703,
+      "step": 8833
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00023102076731826838,
+      "loss": 1.5703,
+      "step": 8834
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002308365987709311,
+      "loss": 1.5918,
+      "step": 8835
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00023065249408321352,
+      "loss": 1.623,
+      "step": 8836
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00023046845327040144,
+      "loss": 1.5195,
+      "step": 8837
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002302844763477745,
+      "loss": 1.5352,
+      "step": 8838
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00023010056333060724,
+      "loss": 1.457,
+      "step": 8839
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022991671423416916,
+      "loss": 1.4805,
+      "step": 8840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002297329290737239,
+      "loss": 1.5645,
+      "step": 8841
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022954920786453083,
+      "loss": 1.582,
+      "step": 8842
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022936555062184294,
+      "loss": 1.5078,
+      "step": 8843
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022918195736090853,
+      "loss": 1.5684,
+      "step": 8844
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022899842809697012,
+      "loss": 1.5938,
+      "step": 8845
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002288149628452655,
+      "loss": 1.5195,
+      "step": 8846
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002286315616210266,
+      "loss": 1.5918,
+      "step": 8847
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002284482244394801,
+      "loss": 1.5703,
+      "step": 8848
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000228264951315848,
+      "loss": 1.6289,
+      "step": 8849
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002280817422653464,
+      "loss": 1.7305,
+      "step": 8850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022789859730318595,
+      "loss": 1.6973,
+      "step": 8851
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002277155164445721,
+      "loss": 1.6289,
+      "step": 8852
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002275324997047057,
+      "loss": 1.4551,
+      "step": 8853
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002273495470987812,
+      "loss": 1.4707,
+      "step": 8854
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002271666586419884,
+      "loss": 1.5039,
+      "step": 8855
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022698383434951118,
+      "loss": 1.5723,
+      "step": 8856
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022680107423652906,
+      "loss": 1.5195,
+      "step": 8857
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022661837831821542,
+      "loss": 1.4668,
+      "step": 8858
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022643574660973843,
+      "loss": 1.5254,
+      "step": 8859
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000226253179126261,
+      "loss": 1.6133,
+      "step": 8860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022607067588294107,
+      "loss": 1.5918,
+      "step": 8861
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022588823689493077,
+      "loss": 1.6133,
+      "step": 8862
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000225705862177377,
+      "loss": 1.6738,
+      "step": 8863
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022552355174542129,
+      "loss": 1.4941,
+      "step": 8864
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022534130561420018,
+      "loss": 1.5625,
+      "step": 8865
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022515912379884452,
+      "loss": 1.6309,
+      "step": 8866
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022497700631447993,
+      "loss": 1.5918,
+      "step": 8867
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022479495317622624,
+      "loss": 1.627,
+      "step": 8868
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002246129643991991,
+      "loss": 1.5449,
+      "step": 8869
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022443103999850777,
+      "loss": 1.625,
+      "step": 8870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022424917998925642,
+      "loss": 1.5195,
+      "step": 8871
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022406738438654372,
+      "loss": 1.6172,
+      "step": 8872
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022388565320546373,
+      "loss": 1.5879,
+      "step": 8873
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022370398646110435,
+      "loss": 1.5742,
+      "step": 8874
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002235223841685482,
+      "loss": 1.5762,
+      "step": 8875
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022334084634287332,
+      "loss": 1.666,
+      "step": 8876
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022315937299915146,
+      "loss": 1.5996,
+      "step": 8877
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022297796415244952,
+      "loss": 1.582,
+      "step": 8878
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022279661981782894,
+      "loss": 1.5254,
+      "step": 8879
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022261534001034568,
+      "loss": 1.5098,
+      "step": 8880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002224341247450503,
+      "loss": 1.6055,
+      "step": 8881
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022225297403698864,
+      "loss": 1.627,
+      "step": 8882
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022207188790120046,
+      "loss": 1.582,
+      "step": 8883
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022189086635272025,
+      "loss": 1.5645,
+      "step": 8884
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022170990940657752,
+      "loss": 1.5391,
+      "step": 8885
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022152901707779606,
+      "loss": 1.5742,
+      "step": 8886
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022134818938139412,
+      "loss": 1.5469,
+      "step": 8887
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022116742633238552,
+      "loss": 1.4863,
+      "step": 8888
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002209867279457777,
+      "loss": 1.6328,
+      "step": 8889
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022080609423657304,
+      "loss": 1.6699,
+      "step": 8890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022062552521976854,
+      "loss": 1.5469,
+      "step": 8891
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022044502091035634,
+      "loss": 1.4727,
+      "step": 8892
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002202645813233225,
+      "loss": 1.6621,
+      "step": 8893
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022008420647364791,
+      "loss": 1.6523,
+      "step": 8894
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021990389637630803,
+      "loss": 1.7422,
+      "step": 8895
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021972365104627346,
+      "loss": 1.7246,
+      "step": 8896
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021954347049850885,
+      "loss": 1.5586,
+      "step": 8897
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021936335474797365,
+      "loss": 1.5898,
+      "step": 8898
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002191833038096217,
+      "loss": 1.6445,
+      "step": 8899
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002190033176984021,
+      "loss": 1.5176,
+      "step": 8900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021882339642925787,
+      "loss": 1.6621,
+      "step": 8901
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021864354001712704,
+      "loss": 1.6055,
+      "step": 8902
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021846374847694195,
+      "loss": 1.4609,
+      "step": 8903
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021828402182363016,
+      "loss": 1.623,
+      "step": 8904
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002181043600721131,
+      "loss": 1.5098,
+      "step": 8905
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021792476323730738,
+      "loss": 1.625,
+      "step": 8906
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002177452313341235,
+      "loss": 1.6387,
+      "step": 8907
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021756576437746767,
+      "loss": 1.6191,
+      "step": 8908
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021738636238223985,
+      "loss": 1.5605,
+      "step": 8909
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002172070253633349,
+      "loss": 1.6016,
+      "step": 8910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021702775333564184,
+      "loss": 1.7168,
+      "step": 8911
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002168485463140454,
+      "loss": 1.6211,
+      "step": 8912
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021666940431342362,
+      "loss": 1.6465,
+      "step": 8913
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021649032734865004,
+      "loss": 1.4902,
+      "step": 8914
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021631131543459214,
+      "loss": 1.6016,
+      "step": 8915
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021613236858611284,
+      "loss": 1.5781,
+      "step": 8916
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002159534868180688,
+      "loss": 1.5918,
+      "step": 8917
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002157746701453118,
+      "loss": 1.6074,
+      "step": 8918
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000215595918582688,
+      "loss": 1.5723,
+      "step": 8919
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021541723214503795,
+      "loss": 1.502,
+      "step": 8920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021523861084719764,
+      "loss": 1.4844,
+      "step": 8921
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021506005470399659,
+      "loss": 1.6504,
+      "step": 8922
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021488156373025957,
+      "loss": 1.4805,
+      "step": 8923
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021470313794080576,
+      "loss": 1.6074,
+      "step": 8924
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021452477735044883,
+      "loss": 1.5254,
+      "step": 8925
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002143464819739971,
+      "loss": 1.5059,
+      "step": 8926
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021416825182625344,
+      "loss": 1.5664,
+      "step": 8927
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002139900869220156,
+      "loss": 1.5195,
+      "step": 8928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021381198727607565,
+      "loss": 1.6387,
+      "step": 8929
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002136339529032202,
+      "loss": 1.5527,
+      "step": 8930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002134559838182303,
+      "loss": 1.5723,
+      "step": 8931
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021327808003588223,
+      "loss": 1.5801,
+      "step": 8932
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021310024157094622,
+      "loss": 1.502,
+      "step": 8933
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021292246843818718,
+      "loss": 1.6367,
+      "step": 8934
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021274476065236448,
+      "loss": 1.5566,
+      "step": 8935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00021256711822823283,
+      "loss": 1.5859,
+      "step": 8936
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00021238954118054065,
+      "loss": 1.4844,
+      "step": 8937
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00021221202952403128,
+      "loss": 1.6895,
+      "step": 8938
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00021203458327344226,
+      "loss": 1.4941,
+      "step": 8939
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002118572024435066,
+      "loss": 1.6582,
+      "step": 8940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002116798870489509,
+      "loss": 1.6094,
+      "step": 8941
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00021150263710449692,
+      "loss": 1.5488,
+      "step": 8942
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00021132545262486047,
+      "loss": 1.498,
+      "step": 8943
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00021114833362475283,
+      "loss": 1.582,
+      "step": 8944
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002109712801188789,
+      "loss": 1.5898,
+      "step": 8945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00021079429212193847,
+      "loss": 1.5176,
+      "step": 8946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00021061736964862577,
+      "loss": 1.4746,
+      "step": 8947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00021044051271363028,
+      "loss": 1.6094,
+      "step": 8948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002102637213316353,
+      "loss": 1.4941,
+      "step": 8949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00021008699551731868,
+      "loss": 1.6328,
+      "step": 8950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020991033528535298,
+      "loss": 1.6895,
+      "step": 8951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020973374065040584,
+      "loss": 1.5684,
+      "step": 8952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002095572116271387,
+      "loss": 1.6113,
+      "step": 8953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002093807482302078,
+      "loss": 1.5762,
+      "step": 8954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002092043504742639,
+      "loss": 1.6094,
+      "step": 8955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002090280183739527,
+      "loss": 1.6973,
+      "step": 8956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020885175194391392,
+      "loss": 1.6621,
+      "step": 8957
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002086755511987821,
+      "loss": 1.5723,
+      "step": 8958
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000208499416153186,
+      "loss": 1.4922,
+      "step": 8959
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020832334682174968,
+      "loss": 1.6016,
+      "step": 8960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000208147343219091,
+      "loss": 1.4648,
+      "step": 8961
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020797140535982274,
+      "loss": 1.6328,
+      "step": 8962
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020779553325855183,
+      "loss": 1.5195,
+      "step": 8963
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020761972692988028,
+      "loss": 1.6055,
+      "step": 8964
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020744398638840422,
+      "loss": 1.5977,
+      "step": 8965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002072683116487143,
+      "loss": 1.7227,
+      "step": 8966
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020709270272539638,
+      "loss": 1.5332,
+      "step": 8967
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020691715963302992,
+      "loss": 1.5332,
+      "step": 8968
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020674168238618963,
+      "loss": 1.4355,
+      "step": 8969
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020656627099944404,
+      "loss": 1.5859,
+      "step": 8970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020639092548735717,
+      "loss": 1.6406,
+      "step": 8971
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020621564586448672,
+      "loss": 1.5508,
+      "step": 8972
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020604043214538537,
+      "loss": 1.6094,
+      "step": 8973
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020586528434459996,
+      "loss": 1.5,
+      "step": 8974
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020569020247667246,
+      "loss": 1.5039,
+      "step": 8975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000205515186556139,
+      "loss": 1.6328,
+      "step": 8976
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020534023659752988,
+      "loss": 1.4238,
+      "step": 8977
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020516535261537027,
+      "loss": 1.4941,
+      "step": 8978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020499053462418038,
+      "loss": 1.5391,
+      "step": 8979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002048157826384741,
+      "loss": 1.584,
+      "step": 8980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020464109667276033,
+      "loss": 1.6152,
+      "step": 8981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020446647674154174,
+      "loss": 1.5879,
+      "step": 8982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020429192285931696,
+      "loss": 1.5117,
+      "step": 8983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002041174350405779,
+      "loss": 1.6172,
+      "step": 8984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020394301329981136,
+      "loss": 1.6523,
+      "step": 8985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020376865765149842,
+      "loss": 1.5449,
+      "step": 8986
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020359436811011544,
+      "loss": 1.5078,
+      "step": 8987
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002034201446901325,
+      "loss": 1.5801,
+      "step": 8988
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020324598740601452,
+      "loss": 1.5977,
+      "step": 8989
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020307189627222057,
+      "loss": 1.5781,
+      "step": 8990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020289787130320502,
+      "loss": 1.6113,
+      "step": 8991
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020272391251341604,
+      "loss": 1.5215,
+      "step": 8992
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020255001991729648,
+      "loss": 1.5566,
+      "step": 8993
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002023761935292835,
+      "loss": 1.5,
+      "step": 8994
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002022024333638094,
+      "loss": 1.6875,
+      "step": 8995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020202873943530054,
+      "loss": 1.4414,
+      "step": 8996
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020185511175817772,
+      "loss": 1.6289,
+      "step": 8997
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020168155034685633,
+      "loss": 1.623,
+      "step": 8998
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020150805521574588,
+      "loss": 1.582,
+      "step": 8999
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020133462637925148,
+      "loss": 1.5352,
+      "step": 9000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002011612638517718,
+      "loss": 1.6348,
+      "step": 9001
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002009879676477,
+      "loss": 1.7578,
+      "step": 9002
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000200814737781424,
+      "loss": 1.5098,
+      "step": 9003
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020064157426732642,
+      "loss": 1.5488,
+      "step": 9004
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020046847711978376,
+      "loss": 1.6465,
+      "step": 9005
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020029544635316742,
+      "loss": 1.6016,
+      "step": 9006
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00020012248198184368,
+      "loss": 1.5234,
+      "step": 9007
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001999495840201725,
+      "loss": 1.6465,
+      "step": 9008
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001997767524825087,
+      "loss": 1.6562,
+      "step": 9009
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001996039873832015,
+      "loss": 1.5879,
+      "step": 9010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019943128873659499,
+      "loss": 1.6035,
+      "step": 9011
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019925865655702734,
+      "loss": 1.5215,
+      "step": 9012
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001990860908588311,
+      "loss": 1.5508,
+      "step": 9013
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001989135916563335,
+      "loss": 1.5293,
+      "step": 9014
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001987411589638566,
+      "loss": 1.582,
+      "step": 9015
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019856879279571638,
+      "loss": 1.4766,
+      "step": 9016
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019839649316622342,
+      "loss": 1.5859,
+      "step": 9017
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019822426008968275,
+      "loss": 1.5352,
+      "step": 9018
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019805209358039443,
+      "loss": 1.5566,
+      "step": 9019
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001978799936526522,
+      "loss": 1.584,
+      "step": 9020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019770796032074467,
+      "loss": 1.709,
+      "step": 9021
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001975359935989547,
+      "loss": 1.5254,
+      "step": 9022
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019736409350156026,
+      "loss": 1.5156,
+      "step": 9023
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019719226004283298,
+      "loss": 1.5391,
+      "step": 9024
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001970204932370393,
+      "loss": 1.6309,
+      "step": 9025
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019684879309843996,
+      "loss": 1.627,
+      "step": 9026
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019667715964129086,
+      "loss": 1.5879,
+      "step": 9027
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001965055928798414,
+      "loss": 1.5684,
+      "step": 9028
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019633409282833602,
+      "loss": 1.582,
+      "step": 9029
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001961626595010132,
+      "loss": 1.5273,
+      "step": 9030
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001959912929121066,
+      "loss": 1.5547,
+      "step": 9031
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019581999307584374,
+      "loss": 1.5039,
+      "step": 9032
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001956487600064466,
+      "loss": 1.625,
+      "step": 9033
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019547759371813168,
+      "loss": 1.6035,
+      "step": 9034
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019530649422511037,
+      "loss": 1.584,
+      "step": 9035
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019513546154158813,
+      "loss": 1.6328,
+      "step": 9036
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019496449568176468,
+      "loss": 1.5137,
+      "step": 9037
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019479359665983432,
+      "loss": 1.5352,
+      "step": 9038
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019462276448998627,
+      "loss": 1.6348,
+      "step": 9039
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019445199918640378,
+      "loss": 1.625,
+      "step": 9040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019428130076326434,
+      "loss": 1.6348,
+      "step": 9041
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001941106692347404,
+      "loss": 1.5957,
+      "step": 9042
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019394010461499844,
+      "loss": 1.6152,
+      "step": 9043
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019376960691819966,
+      "loss": 1.5508,
+      "step": 9044
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001935991761584993,
+      "loss": 1.6602,
+      "step": 9045
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019342881235004784,
+      "loss": 1.5879,
+      "step": 9046
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019325851550698936,
+      "loss": 1.5508,
+      "step": 9047
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019308828564346294,
+      "loss": 1.4629,
+      "step": 9048
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019291812277360133,
+      "loss": 1.5898,
+      "step": 9049
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019274802691153292,
+      "loss": 1.5898,
+      "step": 9050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019257799807137965,
+      "loss": 1.5586,
+      "step": 9051
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001924080362672581,
+      "loss": 1.4375,
+      "step": 9052
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001922381415132791,
+      "loss": 1.4512,
+      "step": 9053
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001920683138235486,
+      "loss": 1.5977,
+      "step": 9054
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001918985532121662,
+      "loss": 1.707,
+      "step": 9055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019172885969322628,
+      "loss": 1.6621,
+      "step": 9056
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019155923328081738,
+      "loss": 1.498,
+      "step": 9057
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019138967398902318,
+      "loss": 1.5918,
+      "step": 9058
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019122018183192102,
+      "loss": 1.582,
+      "step": 9059
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019105075682358298,
+      "loss": 1.6016,
+      "step": 9060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019088139897807522,
+      "loss": 1.7129,
+      "step": 9061
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019071210830945927,
+      "loss": 1.5957,
+      "step": 9062
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019054288483179007,
+      "loss": 1.7246,
+      "step": 9063
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001903737285591174,
+      "loss": 1.6719,
+      "step": 9064
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019020463950548527,
+      "loss": 1.5,
+      "step": 9065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001900356176849326,
+      "loss": 1.416,
+      "step": 9066
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018986666311149224,
+      "loss": 1.6152,
+      "step": 9067
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018969777579919155,
+      "loss": 1.4238,
+      "step": 9068
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001895289557620522,
+      "loss": 1.541,
+      "step": 9069
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018936020301409085,
+      "loss": 1.5371,
+      "step": 9070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018919151756931795,
+      "loss": 1.666,
+      "step": 9071
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018902289944173868,
+      "loss": 1.5781,
+      "step": 9072
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000188854348645352,
+      "loss": 1.4961,
+      "step": 9073
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018868586519415242,
+      "loss": 1.6875,
+      "step": 9074
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001885174491021281,
+      "loss": 1.4688,
+      "step": 9075
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001883491003832617,
+      "loss": 1.5879,
+      "step": 9076
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018818081905153027,
+      "loss": 1.5918,
+      "step": 9077
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018801260512090511,
+      "loss": 1.582,
+      "step": 9078
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018784445860535272,
+      "loss": 1.6797,
+      "step": 9079
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001876763795188331,
+      "loss": 1.5117,
+      "step": 9080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001875083678753009,
+      "loss": 1.6953,
+      "step": 9081
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001873404236887054,
+      "loss": 1.5332,
+      "step": 9082
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018717254697299003,
+      "loss": 1.4199,
+      "step": 9083
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001870047377420928,
+      "loss": 1.4824,
+      "step": 9084
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018683699600994576,
+      "loss": 1.5527,
+      "step": 9085
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018666932179047603,
+      "loss": 1.6777,
+      "step": 9086
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001865017150976046,
+      "loss": 1.6602,
+      "step": 9087
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018633417594524692,
+      "loss": 1.5469,
+      "step": 9088
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018616670434731263,
+      "loss": 1.5391,
+      "step": 9089
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001859993003177066,
+      "loss": 1.5371,
+      "step": 9090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018583196387032718,
+      "loss": 1.584,
+      "step": 9091
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018566469501906746,
+      "loss": 1.4824,
+      "step": 9092
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018549749377781466,
+      "loss": 1.4785,
+      "step": 9093
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018533036016045112,
+      "loss": 1.5742,
+      "step": 9094
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018516329418085276,
+      "loss": 1.5742,
+      "step": 9095
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018499629585289035,
+      "loss": 1.5371,
+      "step": 9096
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018482936519042859,
+      "loss": 1.5547,
+      "step": 9097
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018466250220732717,
+      "loss": 1.709,
+      "step": 9098
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018449570691743988,
+      "loss": 1.6562,
+      "step": 9099
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018432897933461467,
+      "loss": 1.5488,
+      "step": 9100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001841623194726939,
+      "loss": 1.6465,
+      "step": 9101
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018399572734551496,
+      "loss": 1.5078,
+      "step": 9102
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018382920296690876,
+      "loss": 1.4883,
+      "step": 9103
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018366274635070112,
+      "loss": 1.6504,
+      "step": 9104
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001834963575107117,
+      "loss": 1.7285,
+      "step": 9105
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018333003646075552,
+      "loss": 1.4941,
+      "step": 9106
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001831637832146409,
+      "loss": 1.5859,
+      "step": 9107
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001829975977861712,
+      "loss": 1.5508,
+      "step": 9108
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001828314801891434,
+      "loss": 1.5879,
+      "step": 9109
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001826654304373503,
+      "loss": 1.6172,
+      "step": 9110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018249944854457746,
+      "loss": 1.5605,
+      "step": 9111
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018233353452460567,
+      "loss": 1.6641,
+      "step": 9112
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018216768839120968,
+      "loss": 1.543,
+      "step": 9113
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018200191015815936,
+      "loss": 1.4902,
+      "step": 9114
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018183619983921806,
+      "loss": 1.5723,
+      "step": 9115
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018167055744814387,
+      "loss": 1.5273,
+      "step": 9116
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018150498299868901,
+      "loss": 1.6816,
+      "step": 9117
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001813394765046007,
+      "loss": 1.6562,
+      "step": 9118
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018117403797961984,
+      "loss": 1.5625,
+      "step": 9119
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018100866743748202,
+      "loss": 1.5566,
+      "step": 9120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000180843364891917,
+      "loss": 1.5742,
+      "step": 9121
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018067813035664893,
+      "loss": 1.6094,
+      "step": 9122
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001805129638453965,
+      "loss": 1.5781,
+      "step": 9123
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001803478653718723,
+      "loss": 1.5195,
+      "step": 9124
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018018283494978415,
+      "loss": 1.6074,
+      "step": 9125
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001800178725928333,
+      "loss": 1.6504,
+      "step": 9126
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017985297831471593,
+      "loss": 1.7441,
+      "step": 9127
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017968815212912205,
+      "loss": 1.3672,
+      "step": 9128
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001795233940497364,
+      "loss": 1.6094,
+      "step": 9129
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017935870409023814,
+      "loss": 1.502,
+      "step": 9130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017919408226430068,
+      "loss": 1.6094,
+      "step": 9131
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001790295285855912,
+      "loss": 1.541,
+      "step": 9132
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017886504306777241,
+      "loss": 1.5938,
+      "step": 9133
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001787006257245003,
+      "loss": 1.6406,
+      "step": 9134
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017853627656942573,
+      "loss": 1.6035,
+      "step": 9135
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017837199561619344,
+      "loss": 1.5859,
+      "step": 9136
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001782077828784433,
+      "loss": 1.5703,
+      "step": 9137
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017804363836980874,
+      "loss": 1.5957,
+      "step": 9138
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017787956210391797,
+      "loss": 1.6172,
+      "step": 9139
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017771555409439298,
+      "loss": 1.6484,
+      "step": 9140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000177551614354851,
+      "loss": 1.6445,
+      "step": 9141
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017738774289890303,
+      "loss": 1.7305,
+      "step": 9142
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017722393974015427,
+      "loss": 1.5742,
+      "step": 9143
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001770602048922042,
+      "loss": 1.582,
+      "step": 9144
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001768965383686475,
+      "loss": 1.582,
+      "step": 9145
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017673294018307218,
+      "loss": 1.5898,
+      "step": 9146
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017656941034906106,
+      "loss": 1.5215,
+      "step": 9147
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017640594888019068,
+      "loss": 1.5293,
+      "step": 9148
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017624255579003323,
+      "loss": 1.5879,
+      "step": 9149
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017607923109215384,
+      "loss": 1.5664,
+      "step": 9150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017591597480011269,
+      "loss": 1.5625,
+      "step": 9151
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017575278692746367,
+      "loss": 1.5566,
+      "step": 9152
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017558966748775607,
+      "loss": 1.5234,
+      "step": 9153
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017542661649453263,
+      "loss": 1.5918,
+      "step": 9154
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001752636339613305,
+      "loss": 1.5859,
+      "step": 9155
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001751007199016813,
+      "loss": 1.4961,
+      "step": 9156
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017493787432911067,
+      "loss": 1.5254,
+      "step": 9157
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017477509725713936,
+      "loss": 1.6523,
+      "step": 9158
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017461238869928165,
+      "loss": 1.6406,
+      "step": 9159
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001744497486690464,
+      "loss": 1.5449,
+      "step": 9160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001742871771799367,
+      "loss": 1.5859,
+      "step": 9161
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017412467424545008,
+      "loss": 1.5352,
+      "step": 9162
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017396223987907822,
+      "loss": 1.4883,
+      "step": 9163
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001737998740943071,
+      "loss": 1.5586,
+      "step": 9164
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017363757690461745,
+      "loss": 1.5234,
+      "step": 9165
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017347534832348377,
+      "loss": 1.5312,
+      "step": 9166
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017331318836437515,
+      "loss": 1.625,
+      "step": 9167
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001731510970407544,
+      "loss": 1.6465,
+      "step": 9168
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001729890743660798,
+      "loss": 1.623,
+      "step": 9169
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017282712035380287,
+      "loss": 1.5762,
+      "step": 9170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017266523501736997,
+      "loss": 1.6152,
+      "step": 9171
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001725034183702212,
+      "loss": 1.6465,
+      "step": 9172
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017234167042579196,
+      "loss": 1.5566,
+      "step": 9173
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017217999119751104,
+      "loss": 1.5762,
+      "step": 9174
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017201838069880182,
+      "loss": 1.668,
+      "step": 9175
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001718568389430817,
+      "loss": 1.625,
+      "step": 9176
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017169536594376311,
+      "loss": 1.5352,
+      "step": 9177
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017153396171425218,
+      "loss": 1.5918,
+      "step": 9178
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017137262626794935,
+      "loss": 1.5137,
+      "step": 9179
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001712113596182493,
+      "loss": 1.6074,
+      "step": 9180
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001710501617785416,
+      "loss": 1.4551,
+      "step": 9181
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017088903276220947,
+      "loss": 1.5918,
+      "step": 9182
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001707279725826306,
+      "loss": 1.4805,
+      "step": 9183
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001705669812531766,
+      "loss": 1.5254,
+      "step": 9184
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001704060587872144,
+      "loss": 1.6074,
+      "step": 9185
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017024520519810416,
+      "loss": 1.5117,
+      "step": 9186
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017008442049920092,
+      "loss": 1.4004,
+      "step": 9187
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016992370470385332,
+      "loss": 1.6152,
+      "step": 9188
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016976305782540535,
+      "loss": 1.5684,
+      "step": 9189
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016960247987719447,
+      "loss": 1.6328,
+      "step": 9190
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016944197087255254,
+      "loss": 1.6016,
+      "step": 9191
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016928153082480557,
+      "loss": 1.6426,
+      "step": 9192
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001691211597472747,
+      "loss": 1.459,
+      "step": 9193
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016896085765327417,
+      "loss": 1.5977,
+      "step": 9194
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016880062455611322,
+      "loss": 1.5645,
+      "step": 9195
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016864046046909497,
+      "loss": 1.6602,
+      "step": 9196
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016848036540551735,
+      "loss": 1.5957,
+      "step": 9197
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016832033937867196,
+      "loss": 1.666,
+      "step": 9198
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016816038240184518,
+      "loss": 1.6035,
+      "step": 9199
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000168000494488317,
+      "loss": 1.498,
+      "step": 9200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001678406756513624,
+      "loss": 1.5234,
+      "step": 9201
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001676809259042502,
+      "loss": 1.4883,
+      "step": 9202
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016752124526024336,
+      "loss": 1.4902,
+      "step": 9203
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016736163373259983,
+      "loss": 1.4062,
+      "step": 9204
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016720209133457099,
+      "loss": 1.5742,
+      "step": 9205
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001670426180794029,
+      "loss": 1.5312,
+      "step": 9206
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016688321398033578,
+      "loss": 1.6523,
+      "step": 9207
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016672387905060382,
+      "loss": 1.5508,
+      "step": 9208
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016656461330343632,
+      "loss": 1.498,
+      "step": 9209
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000166405416752056,
+      "loss": 1.582,
+      "step": 9210
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001662462894096801,
+      "loss": 1.5645,
+      "step": 9211
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016608723128951996,
+      "loss": 1.6133,
+      "step": 9212
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016592824240478177,
+      "loss": 1.5488,
+      "step": 9213
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001657693227686654,
+      "loss": 1.625,
+      "step": 9214
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016561047239436466,
+      "loss": 1.5352,
+      "step": 9215
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016545169129506877,
+      "loss": 1.4473,
+      "step": 9216
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016529297948396016,
+      "loss": 1.5254,
+      "step": 9217
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016513433697421586,
+      "loss": 1.5215,
+      "step": 9218
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001649757637790069,
+      "loss": 1.5195,
+      "step": 9219
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016481725991149932,
+      "loss": 1.6738,
+      "step": 9220
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016465882538485243,
+      "loss": 1.4727,
+      "step": 9221
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016450046021222042,
+      "loss": 1.6387,
+      "step": 9222
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001643421644067512,
+      "loss": 1.709,
+      "step": 9223
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001641839379815878,
+      "loss": 1.5449,
+      "step": 9224
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016402578094986653,
+      "loss": 1.668,
+      "step": 9225
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016386769332471851,
+      "loss": 1.4629,
+      "step": 9226
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001637096751192686,
+      "loss": 1.6914,
+      "step": 9227
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016355172634663673,
+      "loss": 1.4141,
+      "step": 9228
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016339384701993633,
+      "loss": 1.5488,
+      "step": 9229
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001632360371522752,
+      "loss": 1.4883,
+      "step": 9230
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016307829675675544,
+      "loss": 1.5742,
+      "step": 9231
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001629206258464737,
+      "loss": 1.6445,
+      "step": 9232
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001627630244345204,
+      "loss": 1.5703,
+      "step": 9233
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001626054925339804,
+      "loss": 1.5879,
+      "step": 9234
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016244803015793243,
+      "loss": 1.5566,
+      "step": 9235
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016229063731945038,
+      "loss": 1.6191,
+      "step": 9236
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016213331403160125,
+      "loss": 1.6328,
+      "step": 9237
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016197606030744706,
+      "loss": 1.6094,
+      "step": 9238
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016181887616004365,
+      "loss": 1.6523,
+      "step": 9239
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016166176160244128,
+      "loss": 1.6113,
+      "step": 9240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016150471664768418,
+      "loss": 1.418,
+      "step": 9241
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016134774130881102,
+      "loss": 1.5703,
+      "step": 9242
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001611908355988545,
+      "loss": 1.6387,
+      "step": 9243
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001610339995308422,
+      "loss": 1.7637,
+      "step": 9244
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001608772331177949,
+      "loss": 1.6094,
+      "step": 9245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016072053637272832,
+      "loss": 1.625,
+      "step": 9246
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016056390930865194,
+      "loss": 1.5273,
+      "step": 9247
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016040735193857003,
+      "loss": 1.4746,
+      "step": 9248
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016025086427548063,
+      "loss": 1.4277,
+      "step": 9249
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000160094446332376,
+      "loss": 1.5664,
+      "step": 9250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015993809812224247,
+      "loss": 1.4707,
+      "step": 9251
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015978181965806137,
+      "loss": 1.5547,
+      "step": 9252
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015962561095280738,
+      "loss": 1.7852,
+      "step": 9253
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015946947201944973,
+      "loss": 1.4258,
+      "step": 9254
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015931340287095154,
+      "loss": 1.5312,
+      "step": 9255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015915740352027085,
+      "loss": 1.6172,
+      "step": 9256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015900147398035936,
+      "loss": 1.5566,
+      "step": 9257
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000158845614264163,
+      "loss": 1.4453,
+      "step": 9258
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015868982438462176,
+      "loss": 1.666,
+      "step": 9259
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015853410435467052,
+      "loss": 1.7832,
+      "step": 9260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015837845418723774,
+      "loss": 1.6367,
+      "step": 9261
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015822287389524614,
+      "loss": 1.5664,
+      "step": 9262
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015806736349161256,
+      "loss": 1.5332,
+      "step": 9263
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015791192298924872,
+      "loss": 1.5469,
+      "step": 9264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001577565524010598,
+      "loss": 1.5352,
+      "step": 9265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001576012517399452,
+      "loss": 1.543,
+      "step": 9266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015744602101879879,
+      "loss": 1.5371,
+      "step": 9267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015729086025050888,
+      "loss": 1.5332,
+      "step": 9268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015713576944795748,
+      "loss": 1.6289,
+      "step": 9269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001569807486240209,
+      "loss": 1.5215,
+      "step": 9270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015682579779156946,
+      "loss": 1.6602,
+      "step": 9271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00015667091696346846,
+      "loss": 1.6328,
+      "step": 9272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001565161061525766,
+      "loss": 1.5645,
+      "step": 9273
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015636136537174695,
+      "loss": 1.5176,
+      "step": 9274
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015620669463382674,
+      "loss": 1.5586,
+      "step": 9275
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015605209395165776,
+      "loss": 1.5566,
+      "step": 9276
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001558975633380756,
+      "loss": 1.6035,
+      "step": 9277
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001557431028059101,
+      "loss": 1.4863,
+      "step": 9278
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015558871236798523,
+      "loss": 1.6426,
+      "step": 9279
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001554343920371193,
+      "loss": 1.6074,
+      "step": 9280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001552801418261247,
+      "loss": 1.6348,
+      "step": 9281
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001551259617478078,
+      "loss": 1.6328,
+      "step": 9282
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015497185181496975,
+      "loss": 1.6055,
+      "step": 9283
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015481781204040534,
+      "loss": 1.6445,
+      "step": 9284
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001546638424369038,
+      "loss": 1.6113,
+      "step": 9285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001545099430172482,
+      "loss": 1.6758,
+      "step": 9286
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000154356113794216,
+      "loss": 1.5488,
+      "step": 9287
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015420235478057898,
+      "loss": 1.502,
+      "step": 9288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015404866598910306,
+      "loss": 1.6836,
+      "step": 9289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015389504743254802,
+      "loss": 1.4785,
+      "step": 9290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015374149912366787,
+      "loss": 1.5137,
+      "step": 9291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001535880210752113,
+      "loss": 1.5137,
+      "step": 9292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001534346132999206,
+      "loss": 1.5605,
+      "step": 9293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001532812758105324,
+      "loss": 1.4922,
+      "step": 9294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015312800861977727,
+      "loss": 1.4199,
+      "step": 9295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015297481174038075,
+      "loss": 1.625,
+      "step": 9296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015282168518506155,
+      "loss": 1.5938,
+      "step": 9297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000152668628966533,
+      "loss": 1.5117,
+      "step": 9298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015251564309750275,
+      "loss": 1.5176,
+      "step": 9299
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015236272759067238,
+      "loss": 1.5508,
+      "step": 9300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015220988245873767,
+      "loss": 1.6191,
+      "step": 9301
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015205710771438817,
+      "loss": 1.5918,
+      "step": 9302
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015190440337030852,
+      "loss": 1.6289,
+      "step": 9303
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015175176943917667,
+      "loss": 1.6113,
+      "step": 9304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015159920593366515,
+      "loss": 1.5527,
+      "step": 9305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015144671286644018,
+      "loss": 1.541,
+      "step": 9306
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015129429025016294,
+      "loss": 1.5254,
+      "step": 9307
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001511419380974881,
+      "loss": 1.4766,
+      "step": 9308
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015098965642106456,
+      "loss": 1.7266,
+      "step": 9309
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015083744523353536,
+      "loss": 1.5527,
+      "step": 9310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015068530454753816,
+      "loss": 1.6719,
+      "step": 9311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001505332343757042,
+      "loss": 1.7246,
+      "step": 9312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001503812347306591,
+      "loss": 1.5859,
+      "step": 9313
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015022930562502246,
+      "loss": 1.5957,
+      "step": 9314
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00015007744707140846,
+      "loss": 1.5352,
+      "step": 9315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014992565908242495,
+      "loss": 1.5957,
+      "step": 9316
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014977394167067415,
+      "loss": 1.6172,
+      "step": 9317
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014962229484875224,
+      "loss": 1.5312,
+      "step": 9318
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014947071862924977,
+      "loss": 1.5898,
+      "step": 9319
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014931921302475127,
+      "loss": 1.6289,
+      "step": 9320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014916777804783554,
+      "loss": 1.6172,
+      "step": 9321
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001490164137110751,
+      "loss": 1.4395,
+      "step": 9322
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014886512002703733,
+      "loss": 1.6934,
+      "step": 9323
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001487138970082834,
+      "loss": 1.707,
+      "step": 9324
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001485627446673683,
+      "loss": 1.6465,
+      "step": 9325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001484116630168414,
+      "loss": 1.623,
+      "step": 9326
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014826065206924644,
+      "loss": 1.5723,
+      "step": 9327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014810971183712107,
+      "loss": 1.5176,
+      "step": 9328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014795884233299695,
+      "loss": 1.5059,
+      "step": 9329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001478080435693998,
+      "loss": 1.582,
+      "step": 9330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014765731555885009,
+      "loss": 1.6367,
+      "step": 9331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014750665831386167,
+      "loss": 1.5762,
+      "step": 9332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000147356071846943,
+      "loss": 1.6094,
+      "step": 9333
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014720555617059617,
+      "loss": 1.7559,
+      "step": 9334
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001470551112973181,
+      "loss": 1.5957,
+      "step": 9335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014690473723959929,
+      "loss": 1.6309,
+      "step": 9336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001467544340099246,
+      "loss": 1.625,
+      "step": 9337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014660420162077248,
+      "loss": 1.6094,
+      "step": 9338
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001464540400846166,
+      "loss": 1.6445,
+      "step": 9339
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014630394941392367,
+      "loss": 1.666,
+      "step": 9340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014615392962115514,
+      "loss": 1.6465,
+      "step": 9341
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014600398071876598,
+      "loss": 1.5098,
+      "step": 9342
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001458541027192062,
+      "loss": 1.5391,
+      "step": 9343
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014570429563491905,
+      "loss": 1.4941,
+      "step": 9344
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001455545594783424,
+      "loss": 1.5801,
+      "step": 9345
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014540489426190774,
+      "loss": 1.5625,
+      "step": 9346
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014525529999804144,
+      "loss": 1.6895,
+      "step": 9347
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014510577669916337,
+      "loss": 1.5664,
+      "step": 9348
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014495632437768757,
+      "loss": 1.5723,
+      "step": 9349
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014480694304602217,
+      "loss": 1.6035,
+      "step": 9350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014465763271656983,
+      "loss": 1.5273,
+      "step": 9351
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014450839340172695,
+      "loss": 1.5098,
+      "step": 9352
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014435922511388388,
+      "loss": 1.5723,
+      "step": 9353
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001442101278654253,
+      "loss": 1.5137,
+      "step": 9354
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001440611016687302,
+      "loss": 1.5762,
+      "step": 9355
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014391214653617146,
+      "loss": 1.6426,
+      "step": 9356
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014376326248011583,
+      "loss": 1.6172,
+      "step": 9357
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014361444951292447,
+      "loss": 1.7168,
+      "step": 9358
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014346570764695255,
+      "loss": 1.6016,
+      "step": 9359
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014331703689454933,
+      "loss": 1.584,
+      "step": 9360
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014316843726805794,
+      "loss": 1.5508,
+      "step": 9361
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001430199087798162,
+      "loss": 1.5293,
+      "step": 9362
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014287145144215553,
+      "loss": 1.5371,
+      "step": 9363
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014272306526740152,
+      "loss": 1.6172,
+      "step": 9364
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014257475026787403,
+      "loss": 1.6562,
+      "step": 9365
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001424265064558864,
+      "loss": 1.5898,
+      "step": 9366
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001422783338437472,
+      "loss": 1.5156,
+      "step": 9367
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001421302324437581,
+      "loss": 1.5078,
+      "step": 9368
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014198220226821522,
+      "loss": 1.5664,
+      "step": 9369
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014183424332940854,
+      "loss": 1.5254,
+      "step": 9370
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014168635563962262,
+      "loss": 1.5488,
+      "step": 9371
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001415385392111358,
+      "loss": 1.6348,
+      "step": 9372
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014139079405622035,
+      "loss": 1.5586,
+      "step": 9373
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001412431201871426,
+      "loss": 1.5801,
+      "step": 9374
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014109551761616357,
+      "loss": 1.7656,
+      "step": 9375
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014094798635553785,
+      "loss": 1.4922,
+      "step": 9376
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014080052641751394,
+      "loss": 1.6445,
+      "step": 9377
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014065313781433454,
+      "loss": 1.459,
+      "step": 9378
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014050582055823713,
+      "loss": 1.6309,
+      "step": 9379
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001403585746614523,
+      "loss": 1.6016,
+      "step": 9380
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014021140013620526,
+      "loss": 1.5859,
+      "step": 9381
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014006429699471478,
+      "loss": 1.6562,
+      "step": 9382
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013991726524919458,
+      "loss": 1.6055,
+      "step": 9383
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001397703049118517,
+      "loss": 1.5508,
+      "step": 9384
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013962341599488726,
+      "loss": 1.5391,
+      "step": 9385
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013947659851049722,
+      "loss": 1.5508,
+      "step": 9386
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013932985247087072,
+      "loss": 1.6426,
+      "step": 9387
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013918317788819135,
+      "loss": 1.5664,
+      "step": 9388
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013903657477463648,
+      "loss": 1.541,
+      "step": 9389
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001388900431423783,
+      "loss": 1.5742,
+      "step": 9390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013874358300358235,
+      "loss": 1.6309,
+      "step": 9391
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001385971943704084,
+      "loss": 1.6152,
+      "step": 9392
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013845087725501026,
+      "loss": 1.6113,
+      "step": 9393
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001383046316695361,
+      "loss": 1.6953,
+      "step": 9394
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001381584576261279,
+      "loss": 1.6445,
+      "step": 9395
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013801235513692146,
+      "loss": 1.4746,
+      "step": 9396
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001378663242140472,
+      "loss": 1.4141,
+      "step": 9397
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013772036486962925,
+      "loss": 1.5879,
+      "step": 9398
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013757447711578564,
+      "loss": 1.5977,
+      "step": 9399
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001374286609646289,
+      "loss": 1.4609,
+      "step": 9400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013728291642826508,
+      "loss": 1.5,
+      "step": 9401
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001371372435187951,
+      "loss": 1.6328,
+      "step": 9402
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013699164224831295,
+      "loss": 1.541,
+      "step": 9403
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013684611262890746,
+      "loss": 1.4609,
+      "step": 9404
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001367006546726608,
+      "loss": 1.5879,
+      "step": 9405
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013655526839165011,
+      "loss": 1.6055,
+      "step": 9406
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013640995379794575,
+      "loss": 1.6699,
+      "step": 9407
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013626471090361235,
+      "loss": 1.5918,
+      "step": 9408
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013611953972070867,
+      "loss": 1.5684,
+      "step": 9409
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013597444026128793,
+      "loss": 1.5234,
+      "step": 9410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013582941253739657,
+      "loss": 1.5703,
+      "step": 9411
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013568445656107563,
+      "loss": 1.5117,
+      "step": 9412
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013553957234435976,
+      "loss": 1.6094,
+      "step": 9413
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001353947598992785,
+      "loss": 1.5508,
+      "step": 9414
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001352500192378545,
+      "loss": 1.5352,
+      "step": 9415
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013510535037210492,
+      "loss": 1.6074,
+      "step": 9416
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013496075331404056,
+      "loss": 1.4902,
+      "step": 9417
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013481622807566708,
+      "loss": 1.584,
+      "step": 9418
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013467177466898328,
+      "loss": 1.5938,
+      "step": 9419
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013452739310598262,
+      "loss": 1.7051,
+      "step": 9420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013438308339865202,
+      "loss": 1.6055,
+      "step": 9421
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013423884555897303,
+      "loss": 1.4746,
+      "step": 9422
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013409467959892108,
+      "loss": 1.584,
+      "step": 9423
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013395058553046524,
+      "loss": 1.5137,
+      "step": 9424
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013380656336556873,
+      "loss": 1.5449,
+      "step": 9425
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013366261311618943,
+      "loss": 1.4824,
+      "step": 9426
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013351873479427857,
+      "loss": 1.5938,
+      "step": 9427
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013337492841178166,
+      "loss": 1.5234,
+      "step": 9428
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013323119398063788,
+      "loss": 1.6328,
+      "step": 9429
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013308753151278118,
+      "loss": 1.6562,
+      "step": 9430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013294394102013896,
+      "loss": 1.627,
+      "step": 9431
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013280042251463277,
+      "loss": 1.6133,
+      "step": 9432
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013265697600817794,
+      "loss": 1.5352,
+      "step": 9433
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013251360151268453,
+      "loss": 1.6641,
+      "step": 9434
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013237029904005605,
+      "loss": 1.5508,
+      "step": 9435
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013222706860219002,
+      "loss": 1.5957,
+      "step": 9436
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001320839102109781,
+      "loss": 1.4746,
+      "step": 9437
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001319408238783062,
+      "loss": 1.5566,
+      "step": 9438
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013179780961605382,
+      "loss": 1.502,
+      "step": 9439
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013165486743609456,
+      "loss": 1.6035,
+      "step": 9440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013151199735029662,
+      "loss": 1.5957,
+      "step": 9441
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001313691993705215,
+      "loss": 1.5996,
+      "step": 9442
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013122647350862494,
+      "loss": 1.5645,
+      "step": 9443
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013108381977645667,
+      "loss": 1.5312,
+      "step": 9444
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013094123818586044,
+      "loss": 1.5645,
+      "step": 9445
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013079872874867437,
+      "loss": 1.4766,
+      "step": 9446
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013065629147673009,
+      "loss": 1.5488,
+      "step": 9447
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013051392638185343,
+      "loss": 1.6191,
+      "step": 9448
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000130371633475864,
+      "loss": 1.6914,
+      "step": 9449
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001302294127705761,
+      "loss": 1.5801,
+      "step": 9450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013008726427779727,
+      "loss": 1.7324,
+      "step": 9451
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001299451880093293,
+      "loss": 1.5605,
+      "step": 9452
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001298031839769681,
+      "loss": 1.5371,
+      "step": 9453
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001296612521925037,
+      "loss": 1.6133,
+      "step": 9454
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012951939266771985,
+      "loss": 1.6719,
+      "step": 9455
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012937760541439436,
+      "loss": 1.582,
+      "step": 9456
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012923589044429884,
+      "loss": 1.5117,
+      "step": 9457
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012909424776919964,
+      "loss": 1.6426,
+      "step": 9458
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012895267740085649,
+      "loss": 1.4336,
+      "step": 9459
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012881117935102304,
+      "loss": 1.6445,
+      "step": 9460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012866975363144705,
+      "loss": 1.5566,
+      "step": 9461
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001285284002538707,
+      "loss": 1.6504,
+      "step": 9462
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012838711923002965,
+      "loss": 1.6875,
+      "step": 9463
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001282459105716537,
+      "loss": 1.5527,
+      "step": 9464
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012810477429046652,
+      "loss": 1.5469,
+      "step": 9465
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012796371039818633,
+      "loss": 1.5898,
+      "step": 9466
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012782271890652463,
+      "loss": 1.5352,
+      "step": 9467
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001276817998271872,
+      "loss": 1.5898,
+      "step": 9468
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012754095317187396,
+      "loss": 1.5781,
+      "step": 9469
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012740017895227862,
+      "loss": 1.5605,
+      "step": 9470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001272594771800889,
+      "loss": 1.7207,
+      "step": 9471
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012711884786698635,
+      "loss": 1.5723,
+      "step": 9472
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001269782910246471,
+      "loss": 1.5117,
+      "step": 9473
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012683780666474054,
+      "loss": 1.6289,
+      "step": 9474
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001266973947989306,
+      "loss": 1.6445,
+      "step": 9475
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012655705543887464,
+      "loss": 1.7324,
+      "step": 9476
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012641678859622453,
+      "loss": 1.6758,
+      "step": 9477
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012627659428262572,
+      "loss": 1.6953,
+      "step": 9478
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001261364725097177,
+      "loss": 1.6152,
+      "step": 9479
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012599642328913453,
+      "loss": 1.6367,
+      "step": 9480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001258564466325035,
+      "loss": 1.5488,
+      "step": 9481
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012571654255144604,
+      "loss": 1.6074,
+      "step": 9482
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012557671105757785,
+      "loss": 1.5742,
+      "step": 9483
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012543695216250806,
+      "loss": 1.7012,
+      "step": 9484
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001252972658778405,
+      "loss": 1.6387,
+      "step": 9485
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001251576522151725,
+      "loss": 1.627,
+      "step": 9486
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001250181111860954,
+      "loss": 1.5996,
+      "step": 9487
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001248786428021943,
+      "loss": 1.6426,
+      "step": 9488
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012473924707504913,
+      "loss": 1.5645,
+      "step": 9489
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012459992401623278,
+      "loss": 1.5996,
+      "step": 9490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001244606736373126,
+      "loss": 1.5684,
+      "step": 9491
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012432149594984965,
+      "loss": 1.5801,
+      "step": 9492
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001241823909653995,
+      "loss": 1.5781,
+      "step": 9493
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001240433586955111,
+      "loss": 1.5898,
+      "step": 9494
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012390439915172758,
+      "loss": 1.6699,
+      "step": 9495
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012376551234558576,
+      "loss": 1.6035,
+      "step": 9496
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012362669828861717,
+      "loss": 1.6641,
+      "step": 9497
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001234879569923466,
+      "loss": 1.5742,
+      "step": 9498
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012334928846829306,
+      "loss": 1.6719,
+      "step": 9499
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012321069272796915,
+      "loss": 1.4883,
+      "step": 9500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012307216978288228,
+      "loss": 1.5293,
+      "step": 9501
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012293371964453305,
+      "loss": 1.5234,
+      "step": 9502
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012279534232441624,
+      "loss": 1.5977,
+      "step": 9503
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012265703783402026,
+      "loss": 1.5801,
+      "step": 9504
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012251880618482846,
+      "loss": 1.5762,
+      "step": 9505
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012238064738831722,
+      "loss": 1.6328,
+      "step": 9506
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012224256145595692,
+      "loss": 1.5664,
+      "step": 9507
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012210454839921227,
+      "loss": 1.5762,
+      "step": 9508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012196660822954186,
+      "loss": 1.6582,
+      "step": 9509
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012182874095839813,
+      "loss": 1.5684,
+      "step": 9510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012169094659722734,
+      "loss": 1.5195,
+      "step": 9511
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012155322515746981,
+      "loss": 1.4629,
+      "step": 9512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012141557665056014,
+      "loss": 1.6035,
+      "step": 9513
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001212780010879263,
+      "loss": 1.5879,
+      "step": 9514
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012114049848099062,
+      "loss": 1.6074,
+      "step": 9515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012100306884116907,
+      "loss": 1.5137,
+      "step": 9516
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012086571217987185,
+      "loss": 1.5566,
+      "step": 9517
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012072842850850285,
+      "loss": 1.5586,
+      "step": 9518
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012059121783845994,
+      "loss": 1.5332,
+      "step": 9519
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012045408018113525,
+      "loss": 1.6719,
+      "step": 9520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012031701554791453,
+      "loss": 1.5156,
+      "step": 9521
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012018002395017746,
+      "loss": 1.6172,
+      "step": 9522
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00012004310539929785,
+      "loss": 1.7617,
+      "step": 9523
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011990625990664295,
+      "loss": 1.6699,
+      "step": 9524
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011976948748357485,
+      "loss": 1.5801,
+      "step": 9525
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011963278814144895,
+      "loss": 1.5762,
+      "step": 9526
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001194961618916145,
+      "loss": 1.6504,
+      "step": 9527
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011935960874541462,
+      "loss": 1.584,
+      "step": 9528
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011922312871418728,
+      "loss": 1.4805,
+      "step": 9529
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011908672180926327,
+      "loss": 1.5723,
+      "step": 9530
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011895038804196778,
+      "loss": 1.6133,
+      "step": 9531
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011881412742361974,
+      "loss": 1.5449,
+      "step": 9532
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011867793996553255,
+      "loss": 1.4902,
+      "step": 9533
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011854182567901294,
+      "loss": 1.5391,
+      "step": 9534
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011840578457536178,
+      "loss": 1.6738,
+      "step": 9535
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011826981666587356,
+      "loss": 1.5254,
+      "step": 9536
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011813392196183747,
+      "loss": 1.5391,
+      "step": 9537
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011799810047453596,
+      "loss": 1.5156,
+      "step": 9538
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011786235221524555,
+      "loss": 1.5625,
+      "step": 9539
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011772667719523655,
+      "loss": 1.6445,
+      "step": 9540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011759107542577362,
+      "loss": 1.4746,
+      "step": 9541
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000117455546918115,
+      "loss": 1.6387,
+      "step": 9542
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011732009168351276,
+      "loss": 1.5312,
+      "step": 9543
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011718470973321305,
+      "loss": 1.5801,
+      "step": 9544
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011704940107845618,
+      "loss": 1.5625,
+      "step": 9545
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011691416573047597,
+      "loss": 1.6133,
+      "step": 9546
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011677900370050032,
+      "loss": 1.6133,
+      "step": 9547
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001166439149997508,
+      "loss": 1.5527,
+      "step": 9548
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011650889963944344,
+      "loss": 1.4824,
+      "step": 9549
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011637395763078784,
+      "loss": 1.5449,
+      "step": 9550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011623908898498714,
+      "loss": 1.6094,
+      "step": 9551
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011610429371323938,
+      "loss": 1.5625,
+      "step": 9552
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011596957182673551,
+      "loss": 1.6289,
+      "step": 9553
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001158349233366609,
+      "loss": 1.5234,
+      "step": 9554
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011570034825419474,
+      "loss": 1.5977,
+      "step": 9555
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011556584659050994,
+      "loss": 1.4805,
+      "step": 9556
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011543141835677362,
+      "loss": 1.4492,
+      "step": 9557
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011529706356414638,
+      "loss": 1.6445,
+      "step": 9558
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001151627822237834,
+      "loss": 1.4902,
+      "step": 9559
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011502857434683334,
+      "loss": 1.627,
+      "step": 9560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011489443994443849,
+      "loss": 1.5938,
+      "step": 9561
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011476037902773539,
+      "loss": 1.5684,
+      "step": 9562
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011462639160785438,
+      "loss": 1.625,
+      "step": 9563
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011449247769592008,
+      "loss": 1.666,
+      "step": 9564
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011435863730305041,
+      "loss": 1.6484,
+      "step": 9565
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011422487044035735,
+      "loss": 1.5996,
+      "step": 9566
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011409117711894679,
+      "loss": 1.6328,
+      "step": 9567
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011395755734991909,
+      "loss": 1.5996,
+      "step": 9568
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011382401114436757,
+      "loss": 1.6152,
+      "step": 9569
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011369053851337996,
+      "loss": 1.4453,
+      "step": 9570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001135571394680377,
+      "loss": 1.5859,
+      "step": 9571
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011342381401941648,
+      "loss": 1.5039,
+      "step": 9572
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011329056217858547,
+      "loss": 1.6406,
+      "step": 9573
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011315738395660791,
+      "loss": 1.5996,
+      "step": 9574
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011302427936454062,
+      "loss": 1.4922,
+      "step": 9575
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000112891248413435,
+      "loss": 1.6875,
+      "step": 9576
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011275829111433577,
+      "loss": 1.6055,
+      "step": 9577
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011262540747828165,
+      "loss": 1.625,
+      "step": 9578
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011249259751630502,
+      "loss": 1.498,
+      "step": 9579
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011235986123943286,
+      "loss": 1.5352,
+      "step": 9580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011222719865868536,
+      "loss": 1.6465,
+      "step": 9581
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011209460978507668,
+      "loss": 1.4727,
+      "step": 9582
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011196209462961504,
+      "loss": 1.6113,
+      "step": 9583
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011182965320330263,
+      "loss": 1.7773,
+      "step": 9584
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011169728551713532,
+      "loss": 1.5254,
+      "step": 9585
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011156499158210288,
+      "loss": 1.5586,
+      "step": 9586
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011143277140918872,
+      "loss": 1.6328,
+      "step": 9587
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011130062500937089,
+      "loss": 1.6113,
+      "step": 9588
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011116855239362055,
+      "loss": 1.5391,
+      "step": 9589
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011103655357290298,
+      "loss": 1.5977,
+      "step": 9590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011090462855817729,
+      "loss": 1.5684,
+      "step": 9591
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001107727773603967,
+      "loss": 1.6504,
+      "step": 9592
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011064099999050803,
+      "loss": 1.6523,
+      "step": 9593
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011050929645945218,
+      "loss": 1.4668,
+      "step": 9594
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011037766677816364,
+      "loss": 1.5918,
+      "step": 9595
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011024611095757097,
+      "loss": 1.6133,
+      "step": 9596
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001101146290085967,
+      "loss": 1.5684,
+      "step": 9597
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010998322094215674,
+      "loss": 1.6211,
+      "step": 9598
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010985188676916168,
+      "loss": 1.5586,
+      "step": 9599
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010972062650051518,
+      "loss": 1.6445,
+      "step": 9600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010958944014711537,
+      "loss": 1.5391,
+      "step": 9601
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010945832771985365,
+      "loss": 1.5352,
+      "step": 9602
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010932728922961566,
+      "loss": 1.5742,
+      "step": 9603
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010919632468728113,
+      "loss": 1.5215,
+      "step": 9604
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010906543410372316,
+      "loss": 1.4824,
+      "step": 9605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010893461748980882,
+      "loss": 1.6406,
+      "step": 9606
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010880387485639919,
+      "loss": 1.5215,
+      "step": 9607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010867320621434929,
+      "loss": 1.6387,
+      "step": 9608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010854261157450773,
+      "loss": 1.5508,
+      "step": 9609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001084120909477172,
+      "loss": 1.6719,
+      "step": 9610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001082816443448138,
+      "loss": 1.5137,
+      "step": 9611
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010815127177662832,
+      "loss": 1.4883,
+      "step": 9612
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010802097325398474,
+      "loss": 1.4961,
+      "step": 9613
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010789074878770088,
+      "loss": 1.5859,
+      "step": 9614
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010776059838858865,
+      "loss": 1.6016,
+      "step": 9615
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010763052206745406,
+      "loss": 1.6543,
+      "step": 9616
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010750051983509646,
+      "loss": 1.6289,
+      "step": 9617
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010737059170230912,
+      "loss": 1.6074,
+      "step": 9618
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001072407376798793,
+      "loss": 1.5254,
+      "step": 9619
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010711095777858848,
+      "loss": 1.5195,
+      "step": 9620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010698125200921127,
+      "loss": 1.543,
+      "step": 9621
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010685162038251661,
+      "loss": 1.6094,
+      "step": 9622
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010672206290926678,
+      "loss": 1.6191,
+      "step": 9623
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010659257960021884,
+      "loss": 1.5898,
+      "step": 9624
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010646317046612274,
+      "loss": 1.5234,
+      "step": 9625
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001063338355177228,
+      "loss": 1.6152,
+      "step": 9626
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010620457476575662,
+      "loss": 1.4512,
+      "step": 9627
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010607538822095674,
+      "loss": 1.5977,
+      "step": 9628
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010594627589404838,
+      "loss": 1.5918,
+      "step": 9629
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010581723779575125,
+      "loss": 1.6602,
+      "step": 9630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010568827393677838,
+      "loss": 1.5781,
+      "step": 9631
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001055593843278374,
+      "loss": 1.5195,
+      "step": 9632
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001054305689796291,
+      "loss": 1.75,
+      "step": 9633
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010530182790284848,
+      "loss": 1.5547,
+      "step": 9634
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010517316110818409,
+      "loss": 1.6582,
+      "step": 9635
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001050445686063185,
+      "loss": 1.6582,
+      "step": 9636
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010491605040792806,
+      "loss": 1.6738,
+      "step": 9637
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010478760652368301,
+      "loss": 1.5137,
+      "step": 9638
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001046592369642475,
+      "loss": 1.543,
+      "step": 9639
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010453094174027921,
+      "loss": 1.6992,
+      "step": 9640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010440272086242986,
+      "loss": 1.5508,
+      "step": 9641
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001042745743413448,
+      "loss": 1.6602,
+      "step": 9642
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010414650218766376,
+      "loss": 1.5371,
+      "step": 9643
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010401850441201954,
+      "loss": 1.6016,
+      "step": 9644
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010389058102503934,
+      "loss": 1.543,
+      "step": 9645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010376273203734376,
+      "loss": 1.5293,
+      "step": 9646
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010363495745954766,
+      "loss": 1.6113,
+      "step": 9647
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010350725730225941,
+      "loss": 1.5645,
+      "step": 9648
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010337963157608132,
+      "loss": 1.4473,
+      "step": 9649
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010325208029160926,
+      "loss": 1.6562,
+      "step": 9650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010312460345943353,
+      "loss": 1.6133,
+      "step": 9651
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010299720109013765,
+      "loss": 1.627,
+      "step": 9652
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001028698731942992,
+      "loss": 1.5215,
+      "step": 9653
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010274261978248933,
+      "loss": 1.6602,
+      "step": 9654
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010261544086527363,
+      "loss": 1.5566,
+      "step": 9655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010248833645321087,
+      "loss": 1.6543,
+      "step": 9656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010236130655685404,
+      "loss": 1.6387,
+      "step": 9657
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010223435118674929,
+      "loss": 1.5977,
+      "step": 9658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001021074703534377,
+      "loss": 1.4707,
+      "step": 9659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010198066406745322,
+      "loss": 1.6074,
+      "step": 9660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010185393233932383,
+      "loss": 1.5332,
+      "step": 9661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010172727517957136,
+      "loss": 1.5488,
+      "step": 9662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010160069259871185,
+      "loss": 1.6445,
+      "step": 9663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010147418460725454,
+      "loss": 1.5449,
+      "step": 9664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010134775121570273,
+      "loss": 1.6465,
+      "step": 9665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010122139243455341,
+      "loss": 1.6641,
+      "step": 9666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010109510827429779,
+      "loss": 1.5605,
+      "step": 9667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010096889874542048,
+      "loss": 1.5508,
+      "step": 9668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010084276385839985,
+      "loss": 1.6738,
+      "step": 9669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010071670362370821,
+      "loss": 1.6289,
+      "step": 9670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010059071805181186,
+      "loss": 1.5898,
+      "step": 9671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010046480715317075,
+      "loss": 1.5586,
+      "step": 9672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010033897093823851,
+      "loss": 1.5098,
+      "step": 9673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010021320941746259,
+      "loss": 1.5645,
+      "step": 9674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001000875226012844,
+      "loss": 1.6289,
+      "step": 9675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.996191050013903e-05,
+      "loss": 1.4785,
+      "step": 9676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.983637312445525e-05,
+      "loss": 1.4883,
+      "step": 9677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.971091048465609e-05,
+      "loss": 1.627,
+      "step": 9678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.958552259115795e-05,
+      "loss": 1.627,
+      "step": 9679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.946020945437107e-05,
+      "loss": 1.5254,
+      "step": 9680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.933497108469946e-05,
+      "loss": 1.5664,
+      "step": 9681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.920980749254105e-05,
+      "loss": 1.5664,
+      "step": 9682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.908471868828761e-05,
+      "loss": 1.6426,
+      "step": 9683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.895970468232474e-05,
+      "loss": 1.6738,
+      "step": 9684
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.883476548503134e-05,
+      "loss": 1.5371,
+      "step": 9685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.870990110678058e-05,
+      "loss": 1.5977,
+      "step": 9686
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.85851115579396e-05,
+      "loss": 1.5723,
+      "step": 9687
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.846039684886864e-05,
+      "loss": 1.6387,
+      "step": 9688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.833575698992236e-05,
+      "loss": 1.5332,
+      "step": 9689
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.821119199144868e-05,
+      "loss": 1.5488,
+      "step": 9690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.808670186378998e-05,
+      "loss": 1.5371,
+      "step": 9691
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.796228661728169e-05,
+      "loss": 1.5273,
+      "step": 9692
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.783794626225362e-05,
+      "loss": 1.627,
+      "step": 9693
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.771368080902865e-05,
+      "loss": 1.5449,
+      "step": 9694
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.75894902679244e-05,
+      "loss": 1.5195,
+      "step": 9695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.746537464925154e-05,
+      "loss": 1.4453,
+      "step": 9696
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.73413339633148e-05,
+      "loss": 1.6289,
+      "step": 9697
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.721736822041239e-05,
+      "loss": 1.5352,
+      "step": 9698
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.709347743083696e-05,
+      "loss": 1.6074,
+      "step": 9699
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.69696616048743e-05,
+      "loss": 1.5527,
+      "step": 9700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.684592075280408e-05,
+      "loss": 1.5703,
+      "step": 9701
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.672225488489983e-05,
+      "loss": 1.7207,
+      "step": 9702
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.659866401142925e-05,
+      "loss": 1.4941,
+      "step": 9703
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.647514814265323e-05,
+      "loss": 1.5547,
+      "step": 9704
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.635170728882648e-05,
+      "loss": 1.5801,
+      "step": 9705
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.622834146019776e-05,
+      "loss": 1.707,
+      "step": 9706
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.610505066700958e-05,
+      "loss": 1.5703,
+      "step": 9707
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.598183491949819e-05,
+      "loss": 1.498,
+      "step": 9708
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.585869422789329e-05,
+      "loss": 1.5156,
+      "step": 9709
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.573562860241868e-05,
+      "loss": 1.666,
+      "step": 9710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.561263805329212e-05,
+      "loss": 1.5117,
+      "step": 9711
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.548972259072463e-05,
+      "loss": 1.5547,
+      "step": 9712
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.536688222492129e-05,
+      "loss": 1.4785,
+      "step": 9713
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.524411696608093e-05,
+      "loss": 1.6582,
+      "step": 9714
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.512142682439606e-05,
+      "loss": 1.541,
+      "step": 9715
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.499881181005287e-05,
+      "loss": 1.7051,
+      "step": 9716
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.487627193323167e-05,
+      "loss": 1.6211,
+      "step": 9717
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.475380720410631e-05,
+      "loss": 1.498,
+      "step": 9718
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.463141763284422e-05,
+      "loss": 1.541,
+      "step": 9719
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.450910322960694e-05,
+      "loss": 1.6172,
+      "step": 9720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.438686400454932e-05,
+      "loss": 1.5879,
+      "step": 9721
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.426469996782061e-05,
+      "loss": 1.6777,
+      "step": 9722
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.414261112956334e-05,
+      "loss": 1.5762,
+      "step": 9723
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.402059749991376e-05,
+      "loss": 1.5781,
+      "step": 9724
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.389865908900196e-05,
+      "loss": 1.5938,
+      "step": 9725
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.37767959069522e-05,
+      "loss": 1.5254,
+      "step": 9726
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.365500796388194e-05,
+      "loss": 1.5488,
+      "step": 9727
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.353329526990262e-05,
+      "loss": 1.6641,
+      "step": 9728
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.341165783511917e-05,
+      "loss": 1.6582,
+      "step": 9729
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.329009566963099e-05,
+      "loss": 1.5,
+      "step": 9730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.31686087835304e-05,
+      "loss": 1.5059,
+      "step": 9731
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.304719718690402e-05,
+      "loss": 1.6289,
+      "step": 9732
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.292586088983168e-05,
+      "loss": 1.6211,
+      "step": 9733
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.280459990238765e-05,
+      "loss": 1.498,
+      "step": 9734
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.268341423463956e-05,
+      "loss": 1.6504,
+      "step": 9735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.256230389664877e-05,
+      "loss": 1.6406,
+      "step": 9736
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.244126889847016e-05,
+      "loss": 1.5176,
+      "step": 9737
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.232030925015311e-05,
+      "loss": 1.5449,
+      "step": 9738
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.219942496174005e-05,
+      "loss": 1.5957,
+      "step": 9739
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.207861604326739e-05,
+      "loss": 1.6172,
+      "step": 9740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.195788250476511e-05,
+      "loss": 1.6152,
+      "step": 9741
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.18372243562573e-05,
+      "loss": 1.6191,
+      "step": 9742
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.17166416077615e-05,
+      "loss": 1.5957,
+      "step": 9743
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.159613426928903e-05,
+      "loss": 1.5137,
+      "step": 9744
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.14757023508449e-05,
+      "loss": 1.5879,
+      "step": 9745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.135534586242822e-05,
+      "loss": 1.5312,
+      "step": 9746
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.123506481403144e-05,
+      "loss": 1.625,
+      "step": 9747
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.111485921564078e-05,
+      "loss": 1.5664,
+      "step": 9748
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.099472907723605e-05,
+      "loss": 1.6992,
+      "step": 9749
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.08746744087916e-05,
+      "loss": 1.5977,
+      "step": 9750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.075469522027469e-05,
+      "loss": 1.5078,
+      "step": 9751
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.063479152164644e-05,
+      "loss": 1.5273,
+      "step": 9752
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.05149633228619e-05,
+      "loss": 1.5098,
+      "step": 9753
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.03952106338698e-05,
+      "loss": 1.6562,
+      "step": 9754
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.027553346461248e-05,
+      "loss": 1.5801,
+      "step": 9755
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.015593182502623e-05,
+      "loss": 1.459,
+      "step": 9756
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.003640572504101e-05,
+      "loss": 1.6523,
+      "step": 9757
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.991695517458031e-05,
+      "loss": 1.4629,
+      "step": 9758
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.979758018356166e-05,
+      "loss": 1.6797,
+      "step": 9759
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.967828076189589e-05,
+      "loss": 1.6367,
+      "step": 9760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.955905691948773e-05,
+      "loss": 1.5,
+      "step": 9761
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.943990866623619e-05,
+      "loss": 1.5156,
+      "step": 9762
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.932083601203322e-05,
+      "loss": 1.4668,
+      "step": 9763
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.920183896676481e-05,
+      "loss": 1.5742,
+      "step": 9764
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.90829175403105e-05,
+      "loss": 1.6387,
+      "step": 9765
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.896407174254417e-05,
+      "loss": 1.6367,
+      "step": 9766
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.884530158333259e-05,
+      "loss": 1.5977,
+      "step": 9767
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.872660707253678e-05,
+      "loss": 1.6914,
+      "step": 9768
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.860798822001115e-05,
+      "loss": 1.5391,
+      "step": 9769
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.84894450356043e-05,
+      "loss": 1.5801,
+      "step": 9770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.837097752915801e-05,
+      "loss": 1.6035,
+      "step": 9771
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.82525857105081e-05,
+      "loss": 1.5508,
+      "step": 9772
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.81342695894839e-05,
+      "loss": 1.5371,
+      "step": 9773
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.801602917590878e-05,
+      "loss": 1.4102,
+      "step": 9774
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.789786447959968e-05,
+      "loss": 1.4492,
+      "step": 9775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.777977551036697e-05,
+      "loss": 1.5469,
+      "step": 9776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.766176227801481e-05,
+      "loss": 1.5293,
+      "step": 9777
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.754382479234179e-05,
+      "loss": 1.668,
+      "step": 9778
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.742596306313922e-05,
+      "loss": 1.5781,
+      "step": 9779
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.730817710019257e-05,
+      "loss": 1.5762,
+      "step": 9780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.719046691328103e-05,
+      "loss": 1.6445,
+      "step": 9781
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.707283251217757e-05,
+      "loss": 1.6328,
+      "step": 9782
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.69552739066487e-05,
+      "loss": 1.6133,
+      "step": 9783
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.683779110645473e-05,
+      "loss": 1.4961,
+      "step": 9784
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.672038412134942e-05,
+      "loss": 1.5,
+      "step": 9785
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.660305296108074e-05,
+      "loss": 1.5605,
+      "step": 9786
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.648579763539e-05,
+      "loss": 1.4961,
+      "step": 9787
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.636861815401231e-05,
+      "loss": 1.6309,
+      "step": 9788
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.625151452667623e-05,
+      "loss": 1.625,
+      "step": 9789
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.613448676310465e-05,
+      "loss": 1.6133,
+      "step": 9790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.601753487301356e-05,
+      "loss": 1.6641,
+      "step": 9791
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.590065886611287e-05,
+      "loss": 1.541,
+      "step": 9792
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.578385875210627e-05,
+      "loss": 1.7168,
+      "step": 9793
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.566713454069097e-05,
+      "loss": 1.5918,
+      "step": 9794
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.555048624155793e-05,
+      "loss": 1.6211,
+      "step": 9795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.543391386439192e-05,
+      "loss": 1.582,
+      "step": 9796
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.531741741887145e-05,
+      "loss": 1.5195,
+      "step": 9797
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.520099691466854e-05,
+      "loss": 1.4902,
+      "step": 9798
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.50846523614488e-05,
+      "loss": 1.4668,
+      "step": 9799
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.496838376887195e-05,
+      "loss": 1.6406,
+      "step": 9800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.485219114659082e-05,
+      "loss": 1.6152,
+      "step": 9801
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.473607450425281e-05,
+      "loss": 1.5977,
+      "step": 9802
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.462003385149807e-05,
+      "loss": 1.7012,
+      "step": 9803
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.450406919796106e-05,
+      "loss": 1.5273,
+      "step": 9804
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.438818055326935e-05,
+      "loss": 1.5449,
+      "step": 9805
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.427236792704507e-05,
+      "loss": 1.584,
+      "step": 9806
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.415663132890327e-05,
+      "loss": 1.4199,
+      "step": 9807
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.404097076845285e-05,
+      "loss": 1.6328,
+      "step": 9808
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.392538625529689e-05,
+      "loss": 1.5254,
+      "step": 9809
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.380987779903138e-05,
+      "loss": 1.5293,
+      "step": 9810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.369444540924665e-05,
+      "loss": 1.4473,
+      "step": 9811
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.357908909552614e-05,
+      "loss": 1.5957,
+      "step": 9812
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.346380886744764e-05,
+      "loss": 1.6328,
+      "step": 9813
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.334860473458206e-05,
+      "loss": 1.6309,
+      "step": 9814
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.323347670649429e-05,
+      "loss": 1.6504,
+      "step": 9815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.311842479274246e-05,
+      "loss": 1.5117,
+      "step": 9816
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.300344900287937e-05,
+      "loss": 1.6504,
+      "step": 9817
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.28885493464504e-05,
+      "loss": 1.6406,
+      "step": 9818
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.277372583299514e-05,
+      "loss": 1.625,
+      "step": 9819
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.265897847204673e-05,
+      "loss": 1.5586,
+      "step": 9820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.254430727313234e-05,
+      "loss": 1.5996,
+      "step": 9821
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.242971224577234e-05,
+      "loss": 1.6152,
+      "step": 9822
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.23151933994808e-05,
+      "loss": 1.5547,
+      "step": 9823
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.220075074376565e-05,
+      "loss": 1.5117,
+      "step": 9824
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.208638428812876e-05,
+      "loss": 1.6719,
+      "step": 9825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.19720940420653e-05,
+      "loss": 1.4785,
+      "step": 9826
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.1857880015064e-05,
+      "loss": 1.5625,
+      "step": 9827
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.17437422166073e-05,
+      "loss": 1.5254,
+      "step": 9828
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.162968065617193e-05,
+      "loss": 1.5293,
+      "step": 9829
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.151569534322767e-05,
+      "loss": 1.6758,
+      "step": 9830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.140178628723793e-05,
+      "loss": 1.6152,
+      "step": 9831
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.128795349766016e-05,
+      "loss": 1.5605,
+      "step": 9832
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.117419698394534e-05,
+      "loss": 1.5449,
+      "step": 9833
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.106051675553772e-05,
+      "loss": 1.5352,
+      "step": 9834
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.094691282187605e-05,
+      "loss": 1.5234,
+      "step": 9835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.083338519239213e-05,
+      "loss": 1.5586,
+      "step": 9836
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.071993387651144e-05,
+      "loss": 1.6445,
+      "step": 9837
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.060655888365331e-05,
+      "loss": 1.5938,
+      "step": 9838
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.049326022323066e-05,
+      "loss": 1.623,
+      "step": 9839
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.038003790464999e-05,
+      "loss": 1.498,
+      "step": 9840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.026689193731184e-05,
+      "loss": 1.7012,
+      "step": 9841
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.015382233060997e-05,
+      "loss": 1.6582,
+      "step": 9842
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.004082909393195e-05,
+      "loss": 1.5879,
+      "step": 9843
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.992791223665885e-05,
+      "loss": 1.5977,
+      "step": 9844
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.981507176816583e-05,
+      "loss": 1.5352,
+      "step": 9845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.970230769782138e-05,
+      "loss": 1.5605,
+      "step": 9846
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.958962003498771e-05,
+      "loss": 1.6348,
+      "step": 9847
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.947700878902053e-05,
+      "loss": 1.4941,
+      "step": 9848
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.93644739692696e-05,
+      "loss": 1.5332,
+      "step": 9849
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.9252015585078e-05,
+      "loss": 1.584,
+      "step": 9850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.913963364578259e-05,
+      "loss": 1.5312,
+      "step": 9851
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.902732816071368e-05,
+      "loss": 1.6523,
+      "step": 9852
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.89150991391957e-05,
+      "loss": 1.4941,
+      "step": 9853
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.88029465905462e-05,
+      "loss": 1.5664,
+      "step": 9854
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.869087052407675e-05,
+      "loss": 1.5488,
+      "step": 9855
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.857887094909233e-05,
+      "loss": 1.582,
+      "step": 9856
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.846694787489183e-05,
+      "loss": 1.4473,
+      "step": 9857
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.83551013107675e-05,
+      "loss": 1.6035,
+      "step": 9858
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.824333126600547e-05,
+      "loss": 1.5996,
+      "step": 9859
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.813163774988529e-05,
+      "loss": 1.627,
+      "step": 9860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.802002077168047e-05,
+      "loss": 1.5391,
+      "step": 9861
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.790848034065801e-05,
+      "loss": 1.4551,
+      "step": 9862
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.779701646607828e-05,
+      "loss": 1.6836,
+      "step": 9863
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.768562915719557e-05,
+      "loss": 1.5566,
+      "step": 9864
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.757431842325812e-05,
+      "loss": 1.6797,
+      "step": 9865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.746308427350724e-05,
+      "loss": 1.5859,
+      "step": 9866
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.735192671717805e-05,
+      "loss": 1.541,
+      "step": 9867
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.724084576349944e-05,
+      "loss": 1.4766,
+      "step": 9868
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.7129841421694e-05,
+      "loss": 1.4863,
+      "step": 9869
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.701891370097769e-05,
+      "loss": 1.5234,
+      "step": 9870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.690806261056038e-05,
+      "loss": 1.5137,
+      "step": 9871
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.679728815964548e-05,
+      "loss": 1.6055,
+      "step": 9872
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.668659035742987e-05,
+      "loss": 1.5391,
+      "step": 9873
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.657596921310405e-05,
+      "loss": 1.6836,
+      "step": 9874
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.646542473585272e-05,
+      "loss": 1.6094,
+      "step": 9875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.635495693485373e-05,
+      "loss": 1.6074,
+      "step": 9876
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.62445658192784e-05,
+      "loss": 1.5254,
+      "step": 9877
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.61342513982921e-05,
+      "loss": 1.5566,
+      "step": 9878
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.602401368105372e-05,
+      "loss": 1.4434,
+      "step": 9879
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.591385267671547e-05,
+      "loss": 1.5332,
+      "step": 9880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.580376839442371e-05,
+      "loss": 1.6523,
+      "step": 9881
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.569376084331803e-05,
+      "loss": 1.5645,
+      "step": 9882
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.558383003253188e-05,
+      "loss": 1.6348,
+      "step": 9883
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.547397597119199e-05,
+      "loss": 1.5996,
+      "step": 9884
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.536419866841926e-05,
+      "loss": 1.4766,
+      "step": 9885
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.525449813332797e-05,
+      "loss": 1.6426,
+      "step": 9886
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.514487437502571e-05,
+      "loss": 1.4922,
+      "step": 9887
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.503532740261388e-05,
+      "loss": 1.5547,
+      "step": 9888
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.492585722518808e-05,
+      "loss": 1.5293,
+      "step": 9889
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.481646385183671e-05,
+      "loss": 1.6406,
+      "step": 9890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.470714729164207e-05,
+      "loss": 1.457,
+      "step": 9891
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.459790755368035e-05,
+      "loss": 1.5957,
+      "step": 9892
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.448874464702115e-05,
+      "loss": 1.457,
+      "step": 9893
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.437965858072748e-05,
+      "loss": 1.5117,
+      "step": 9894
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.42706493638563e-05,
+      "loss": 1.5371,
+      "step": 9895
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.416171700545815e-05,
+      "loss": 1.6172,
+      "step": 9896
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.405286151457713e-05,
+      "loss": 1.6582,
+      "step": 9897
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.39440829002509e-05,
+      "loss": 1.5215,
+      "step": 9898
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.383538117151056e-05,
+      "loss": 1.4805,
+      "step": 9899
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.372675633738135e-05,
+      "loss": 1.7109,
+      "step": 9900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.361820840688172e-05,
+      "loss": 1.543,
+      "step": 9901
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.35097373890239e-05,
+      "loss": 1.5645,
+      "step": 9902
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.340134329281344e-05,
+      "loss": 1.5625,
+      "step": 9903
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.329302612724997e-05,
+      "loss": 1.5605,
+      "step": 9904
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.318478590132649e-05,
+      "loss": 1.5215,
+      "step": 9905
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.30766226240296e-05,
+      "loss": 1.5703,
+      "step": 9906
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.296853630433919e-05,
+      "loss": 1.6211,
+      "step": 9907
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.286052695122968e-05,
+      "loss": 1.5469,
+      "step": 9908
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.275259457366822e-05,
+      "loss": 1.6035,
+      "step": 9909
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.264473918061588e-05,
+      "loss": 1.7109,
+      "step": 9910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.253696078102734e-05,
+      "loss": 1.752,
+      "step": 9911
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.242925938385093e-05,
+      "loss": 1.6484,
+      "step": 9912
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.232163499802824e-05,
+      "loss": 1.4062,
+      "step": 9913
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.221408763249526e-05,
+      "loss": 1.541,
+      "step": 9914
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.210661729618073e-05,
+      "loss": 1.3906,
+      "step": 9915
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.19992239980075e-05,
+      "loss": 1.5898,
+      "step": 9916
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.189190774689192e-05,
+      "loss": 1.4902,
+      "step": 9917
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.17846685517437e-05,
+      "loss": 1.5625,
+      "step": 9918
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.167750642146642e-05,
+      "loss": 1.5859,
+      "step": 9919
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.157042136495728e-05,
+      "loss": 1.4766,
+      "step": 9920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.146341339110695e-05,
+      "loss": 1.5098,
+      "step": 9921
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.135648250879979e-05,
+      "loss": 1.625,
+      "step": 9922
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.124962872691342e-05,
+      "loss": 1.625,
+      "step": 9923
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.11428520543198e-05,
+      "loss": 1.5449,
+      "step": 9924
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.103615249988382e-05,
+      "loss": 1.4277,
+      "step": 9925
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.092953007246416e-05,
+      "loss": 1.5508,
+      "step": 9926
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.082298478091309e-05,
+      "loss": 1.5059,
+      "step": 9927
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.071651663407663e-05,
+      "loss": 1.5625,
+      "step": 9928
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.061012564079428e-05,
+      "loss": 1.7188,
+      "step": 9929
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.050381180989918e-05,
+      "loss": 1.5273,
+      "step": 9930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.039757515021761e-05,
+      "loss": 1.5977,
+      "step": 9931
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.029141567057029e-05,
+      "loss": 1.5645,
+      "step": 9932
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.018533337977095e-05,
+      "loss": 1.5879,
+      "step": 9933
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.007932828662711e-05,
+      "loss": 1.6523,
+      "step": 9934
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.997340039993948e-05,
+      "loss": 1.6895,
+      "step": 9935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.986754972850317e-05,
+      "loss": 1.5293,
+      "step": 9936
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.976177628110625e-05,
+      "loss": 1.5254,
+      "step": 9937
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.965608006653035e-05,
+      "loss": 1.457,
+      "step": 9938
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.9550461093551e-05,
+      "loss": 1.5293,
+      "step": 9939
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.944491937093733e-05,
+      "loss": 1.4805,
+      "step": 9940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.933945490745175e-05,
+      "loss": 1.6309,
+      "step": 9941
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.923406771185048e-05,
+      "loss": 1.5508,
+      "step": 9942
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.912875779288308e-05,
+      "loss": 1.5469,
+      "step": 9943
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.902352515929322e-05,
+      "loss": 1.4902,
+      "step": 9944
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.891836981981769e-05,
+      "loss": 1.5039,
+      "step": 9945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.881329178318696e-05,
+      "loss": 1.5254,
+      "step": 9946
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.870829105812482e-05,
+      "loss": 1.6289,
+      "step": 9947
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.860336765334952e-05,
+      "loss": 1.6055,
+      "step": 9948
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.849852157757197e-05,
+      "loss": 1.5684,
+      "step": 9949
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.839375283949689e-05,
+      "loss": 1.5254,
+      "step": 9950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.828906144782298e-05,
+      "loss": 1.4824,
+      "step": 9951
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.818444741124196e-05,
+      "loss": 1.5879,
+      "step": 9952
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.807991073843933e-05,
+      "loss": 1.6621,
+      "step": 9953
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.797545143809447e-05,
+      "loss": 1.584,
+      "step": 9954
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.787106951888011e-05,
+      "loss": 1.5957,
+      "step": 9955
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.776676498946232e-05,
+      "loss": 1.541,
+      "step": 9956
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.766253785850107e-05,
+      "loss": 1.5742,
+      "step": 9957
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.755838813464987e-05,
+      "loss": 1.5801,
+      "step": 9958
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.745431582655537e-05,
+      "loss": 1.5449,
+      "step": 9959
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.735032094285864e-05,
+      "loss": 1.5605,
+      "step": 9960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.724640349219358e-05,
+      "loss": 1.6484,
+      "step": 9961
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.714256348318793e-05,
+      "loss": 1.4473,
+      "step": 9962
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.70388009244629e-05,
+      "loss": 1.6309,
+      "step": 9963
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.693511582463352e-05,
+      "loss": 1.6309,
+      "step": 9964
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.68315081923082e-05,
+      "loss": 1.4746,
+      "step": 9965
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.672797803608887e-05,
+      "loss": 1.4648,
+      "step": 9966
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.662452536457097e-05,
+      "loss": 1.6758,
+      "step": 9967
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.652115018634397e-05,
+      "loss": 1.5605,
+      "step": 9968
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.641785250999033e-05,
+      "loss": 1.6367,
+      "step": 9969
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.631463234408641e-05,
+      "loss": 1.625,
+      "step": 9970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.621148969720192e-05,
+      "loss": 1.6016,
+      "step": 9971
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.610842457790045e-05,
+      "loss": 1.584,
+      "step": 9972
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.600543699473882e-05,
+      "loss": 1.6035,
+      "step": 9973
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.590252695626742e-05,
+      "loss": 1.5586,
+      "step": 9974
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.579969447103062e-05,
+      "loss": 1.6113,
+      "step": 9975
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.569693954756594e-05,
+      "loss": 1.5215,
+      "step": 9976
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.559426219440468e-05,
+      "loss": 1.4922,
+      "step": 9977
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.549166242007132e-05,
+      "loss": 1.5645,
+      "step": 9978
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.538914023308451e-05,
+      "loss": 1.6094,
+      "step": 9979
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.528669564195589e-05,
+      "loss": 1.6875,
+      "step": 9980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.518432865519108e-05,
+      "loss": 1.6777,
+      "step": 9981
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.508203928128886e-05,
+      "loss": 1.5586,
+      "step": 9982
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.49798275287421e-05,
+      "loss": 1.543,
+      "step": 9983
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.487769340603666e-05,
+      "loss": 1.5762,
+      "step": 9984
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.477563692165233e-05,
+      "loss": 1.6016,
+      "step": 9985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4673658084062e-05,
+      "loss": 1.6309,
+      "step": 9986
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4571756901733e-05,
+      "loss": 1.4883,
+      "step": 9987
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.446993338312534e-05,
+      "loss": 1.5508,
+      "step": 9988
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.436818753669282e-05,
+      "loss": 1.6426,
+      "step": 9989
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.426651937088301e-05,
+      "loss": 1.543,
+      "step": 9990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.416492889413682e-05,
+      "loss": 1.6055,
+      "step": 9991
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.406341611488875e-05,
+      "loss": 1.5352,
+      "step": 9992
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.396198104156692e-05,
+      "loss": 1.5859,
+      "step": 9993
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.386062368259304e-05,
+      "loss": 1.6211,
+      "step": 9994
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.375934404638217e-05,
+      "loss": 1.6172,
+      "step": 9995
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.365814214134303e-05,
+      "loss": 1.5684,
+      "step": 9996
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.355701797587799e-05,
+      "loss": 1.5234,
+      "step": 9997
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.345597155838257e-05,
+      "loss": 1.5859,
+      "step": 9998
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.33550028972465e-05,
+      "loss": 1.5918,
+      "step": 9999
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.325411200085251e-05,
+      "loss": 1.6719,
+      "step": 10000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.315329887757714e-05,
+      "loss": 1.6934,
+      "step": 10001
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.305256353579003e-05,
+      "loss": 1.5625,
+      "step": 10002
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.295190598385514e-05,
+      "loss": 1.6562,
+      "step": 10003
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.285132623012946e-05,
+      "loss": 1.6211,
+      "step": 10004
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.275082428296341e-05,
+      "loss": 1.5234,
+      "step": 10005
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.265040015070111e-05,
+      "loss": 1.5215,
+      "step": 10006
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.255005384168056e-05,
+      "loss": 1.6328,
+      "step": 10007
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.244978536423285e-05,
+      "loss": 1.6094,
+      "step": 10008
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.234959472668266e-05,
+      "loss": 1.5293,
+      "step": 10009
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.224948193734826e-05,
+      "loss": 1.6016,
+      "step": 10010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.214944700454162e-05,
+      "loss": 1.5156,
+      "step": 10011
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.204948993656823e-05,
+      "loss": 1.6699,
+      "step": 10012
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.194961074172678e-05,
+      "loss": 1.6113,
+      "step": 10013
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.184980942830953e-05,
+      "loss": 1.541,
+      "step": 10014
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.175008600460297e-05,
+      "loss": 1.7168,
+      "step": 10015
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.165044047888646e-05,
+      "loss": 1.502,
+      "step": 10016
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.155087285943284e-05,
+      "loss": 1.5332,
+      "step": 10017
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.14513831545086e-05,
+      "loss": 1.5508,
+      "step": 10018
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.135197137237436e-05,
+      "loss": 1.6855,
+      "step": 10019
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.12526375212834e-05,
+      "loss": 1.623,
+      "step": 10020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.115338160948292e-05,
+      "loss": 1.5723,
+      "step": 10021
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.105420364521342e-05,
+      "loss": 1.5547,
+      "step": 10022
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.095510363670964e-05,
+      "loss": 1.4863,
+      "step": 10023
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.085608159219902e-05,
+      "loss": 1.6875,
+      "step": 10024
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.075713751990275e-05,
+      "loss": 1.6328,
+      "step": 10025
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.065827142803582e-05,
+      "loss": 1.5547,
+      "step": 10026
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.055948332480654e-05,
+      "loss": 1.5332,
+      "step": 10027
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.04607732184167e-05,
+      "loss": 1.6504,
+      "step": 10028
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.036214111706173e-05,
+      "loss": 1.5352,
+      "step": 10029
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.026358702893064e-05,
+      "loss": 1.6328,
+      "step": 10030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.016511096220556e-05,
+      "loss": 1.4922,
+      "step": 10031
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.006671292506272e-05,
+      "loss": 1.5859,
+      "step": 10032
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.996839292567158e-05,
+      "loss": 1.5273,
+      "step": 10033
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.987015097219506e-05,
+      "loss": 1.5195,
+      "step": 10034
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.9771987072789635e-05,
+      "loss": 1.6309,
+      "step": 10035
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.9673901235605454e-05,
+      "loss": 1.582,
+      "step": 10036
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.9575893468786e-05,
+      "loss": 1.5371,
+      "step": 10037
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.9477963780468105e-05,
+      "loss": 1.5645,
+      "step": 10038
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.9380112178782916e-05,
+      "loss": 1.6738,
+      "step": 10039
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.928233867185406e-05,
+      "loss": 1.6992,
+      "step": 10040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.918464326779949e-05,
+      "loss": 1.5938,
+      "step": 10041
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.908702597472993e-05,
+      "loss": 1.5898,
+      "step": 10042
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8989486800750555e-05,
+      "loss": 1.6426,
+      "step": 10043
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.889202575395913e-05,
+      "loss": 1.6406,
+      "step": 10044
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.879464284244762e-05,
+      "loss": 1.627,
+      "step": 10045
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8697338074300777e-05,
+      "loss": 1.6113,
+      "step": 10046
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.860011145759791e-05,
+      "loss": 1.6094,
+      "step": 10047
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.85029630004108e-05,
+      "loss": 1.5098,
+      "step": 10048
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.840589271080543e-05,
+      "loss": 1.5801,
+      "step": 10049
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8308900596840685e-05,
+      "loss": 1.5781,
+      "step": 10050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8211986666569684e-05,
+      "loss": 1.5723,
+      "step": 10051
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.811515092803865e-05,
+      "loss": 1.5332,
+      "step": 10052
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.801839338928716e-05,
+      "loss": 1.4648,
+      "step": 10053
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.792171405834845e-05,
+      "loss": 1.5859,
+      "step": 10054
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.782511294324955e-05,
+      "loss": 1.543,
+      "step": 10055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.77285900520107e-05,
+      "loss": 1.5254,
+      "step": 10056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.7632145392645494e-05,
+      "loss": 1.5547,
+      "step": 10057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.7535778973161315e-05,
+      "loss": 1.582,
+      "step": 10058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.743949080155908e-05,
+      "loss": 1.5605,
+      "step": 10059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.734328088583307e-05,
+      "loss": 1.5371,
+      "step": 10060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7247149233970785e-05,
+      "loss": 1.498,
+      "step": 10061
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.715109585395395e-05,
+      "loss": 1.5039,
+      "step": 10062
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7055120753757296e-05,
+      "loss": 1.6973,
+      "step": 10063
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.695922394134889e-05,
+      "loss": 1.6094,
+      "step": 10064
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.686340542469071e-05,
+      "loss": 1.498,
+      "step": 10065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.676766521173804e-05,
+      "loss": 1.6387,
+      "step": 10066
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.667200331043987e-05,
+      "loss": 1.5801,
+      "step": 10067
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.657641972873817e-05,
+      "loss": 1.502,
+      "step": 10068
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.648091447456905e-05,
+      "loss": 1.4863,
+      "step": 10069
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.638548755586159e-05,
+      "loss": 1.6152,
+      "step": 10070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.629013898053858e-05,
+      "loss": 1.5586,
+      "step": 10071
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.619486875651658e-05,
+      "loss": 1.5762,
+      "step": 10072
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.609967689170514e-05,
+      "loss": 1.5801,
+      "step": 10073
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.6004563394007724e-05,
+      "loss": 1.5762,
+      "step": 10074
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5909528271320896e-05,
+      "loss": 1.459,
+      "step": 10075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.581457153153502e-05,
+      "loss": 1.7305,
+      "step": 10076
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.571969318253378e-05,
+      "loss": 1.5254,
+      "step": 10077
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.562489323219466e-05,
+      "loss": 1.6289,
+      "step": 10078
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.553017168838826e-05,
+      "loss": 1.6152,
+      "step": 10079
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.543552855897882e-05,
+      "loss": 1.5898,
+      "step": 10080
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.534096385182386e-05,
+      "loss": 1.5312,
+      "step": 10081
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5246477574774966e-05,
+      "loss": 1.6934,
+      "step": 10082
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5152069735676655e-05,
+      "loss": 1.6328,
+      "step": 10083
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.505774034236721e-05,
+      "loss": 1.6133,
+      "step": 10084
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.496348940267803e-05,
+      "loss": 1.4883,
+      "step": 10085
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4869316924434646e-05,
+      "loss": 1.4629,
+      "step": 10086
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4775222915455446e-05,
+      "loss": 1.5312,
+      "step": 10087
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4681207383552754e-05,
+      "loss": 1.7148,
+      "step": 10088
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.458727033653188e-05,
+      "loss": 1.5664,
+      "step": 10089
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.449341178219236e-05,
+      "loss": 1.7129,
+      "step": 10090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4399631728326516e-05,
+      "loss": 1.5723,
+      "step": 10091
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4305930182720457e-05,
+      "loss": 1.6016,
+      "step": 10092
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4212307153153615e-05,
+      "loss": 1.6562,
+      "step": 10093
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.411876264739923e-05,
+      "loss": 1.584,
+      "step": 10094
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.402529667322375e-05,
+      "loss": 1.707,
+      "step": 10095
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.39319092383872e-05,
+      "loss": 1.5977,
+      "step": 10096
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3838600350642695e-05,
+      "loss": 1.6484,
+      "step": 10097
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.374537001773772e-05,
+      "loss": 1.5996,
+      "step": 10098
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.365221824741251e-05,
+      "loss": 1.6836,
+      "step": 10099
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3559145047400895e-05,
+      "loss": 1.6055,
+      "step": 10100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.346615042543002e-05,
+      "loss": 1.5898,
+      "step": 10101
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.337323438922126e-05,
+      "loss": 1.4453,
+      "step": 10102
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.328039694648867e-05,
+      "loss": 1.4688,
+      "step": 10103
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3187638104940094e-05,
+      "loss": 1.6289,
+      "step": 10104
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.309495787227669e-05,
+      "loss": 1.6152,
+      "step": 10105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.300235625619343e-05,
+      "loss": 1.5352,
+      "step": 10106
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.290983326437848e-05,
+      "loss": 1.5781,
+      "step": 10107
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2817388904513485e-05,
+      "loss": 1.5332,
+      "step": 10108
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2725023184273634e-05,
+      "loss": 1.5371,
+      "step": 10109
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.263273611132735e-05,
+      "loss": 1.6113,
+      "step": 10110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.254052769333717e-05,
+      "loss": 1.5605,
+      "step": 10111
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.244839793795853e-05,
+      "loss": 1.5449,
+      "step": 10112
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.23563468528403e-05,
+      "loss": 1.623,
+      "step": 10113
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.226437444562526e-05,
+      "loss": 1.7305,
+      "step": 10114
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2172480723949086e-05,
+      "loss": 1.5371,
+      "step": 10115
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.208066569544156e-05,
+      "loss": 1.5527,
+      "step": 10116
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.198892936772526e-05,
+      "loss": 1.6445,
+      "step": 10117
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.189727174841696e-05,
+      "loss": 1.4824,
+      "step": 10118
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1805692845126265e-05,
+      "loss": 1.6797,
+      "step": 10119
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.171419266545663e-05,
+      "loss": 1.6797,
+      "step": 10120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.162277121700454e-05,
+      "loss": 1.6621,
+      "step": 10121
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.153142850736059e-05,
+      "loss": 1.5684,
+      "step": 10122
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.144016454410849e-05,
+      "loss": 1.5039,
+      "step": 10123
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.134897933482519e-05,
+      "loss": 1.4863,
+      "step": 10124
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.125787288708117e-05,
+      "loss": 1.666,
+      "step": 10125
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.116684520844106e-05,
+      "loss": 1.6914,
+      "step": 10126
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1075896306462035e-05,
+      "loss": 1.6094,
+      "step": 10127
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.098502618869516e-05,
+      "loss": 1.6113,
+      "step": 10128
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0894234862684853e-05,
+      "loss": 1.5859,
+      "step": 10129
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.08035223359693e-05,
+      "loss": 1.5879,
+      "step": 10130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.07128886160797e-05,
+      "loss": 1.7773,
+      "step": 10131
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.062233371054081e-05,
+      "loss": 1.6094,
+      "step": 10132
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0531857626870957e-05,
+      "loss": 1.5898,
+      "step": 10133
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0441460372582125e-05,
+      "loss": 1.7285,
+      "step": 10134
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0351141955179314e-05,
+      "loss": 1.5234,
+      "step": 10135
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.02609023821613e-05,
+      "loss": 1.5098,
+      "step": 10136
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.017074166101998e-05,
+      "loss": 1.5566,
+      "step": 10137
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.008065979924126e-05,
+      "loss": 1.7285,
+      "step": 10138
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.999065680430403e-05,
+      "loss": 1.5879,
+      "step": 10139
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.990073268368078e-05,
+      "loss": 1.5957,
+      "step": 10140
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.981088744483719e-05,
+      "loss": 1.6133,
+      "step": 10141
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.972112109523297e-05,
+      "loss": 1.5293,
+      "step": 10142
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9631433642320924e-05,
+      "loss": 1.6133,
+      "step": 10143
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9541825093546985e-05,
+      "loss": 1.5391,
+      "step": 10144
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9452295456351435e-05,
+      "loss": 1.6309,
+      "step": 10145
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9362844738167104e-05,
+      "loss": 1.6016,
+      "step": 10146
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.927347294642059e-05,
+      "loss": 1.4297,
+      "step": 10147
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9184180088532095e-05,
+      "loss": 1.4902,
+      "step": 10148
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.909496617191522e-05,
+      "loss": 1.6387,
+      "step": 10149
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.90058312039765e-05,
+      "loss": 1.6406,
+      "step": 10150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.89167751921169e-05,
+      "loss": 1.6465,
+      "step": 10151
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.882779814372995e-05,
+      "loss": 1.6523,
+      "step": 10152
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8738900066203074e-05,
+      "loss": 1.6094,
+      "step": 10153
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8650080966916924e-05,
+      "loss": 1.582,
+      "step": 10154
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8561340853245813e-05,
+      "loss": 1.5586,
+      "step": 10155
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.847267973255709e-05,
+      "loss": 1.6152,
+      "step": 10156
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.838409761221207e-05,
+      "loss": 1.5801,
+      "step": 10157
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.829559449956533e-05,
+      "loss": 1.6406,
+      "step": 10158
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.820717040196465e-05,
+      "loss": 1.5156,
+      "step": 10159
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.811882532675138e-05,
+      "loss": 1.4922,
+      "step": 10160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8030559281260544e-05,
+      "loss": 1.5723,
+      "step": 10161
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.794237227282039e-05,
+      "loss": 1.6406,
+      "step": 10162
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.78542643087525e-05,
+      "loss": 1.6465,
+      "step": 10163
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7766235396371926e-05,
+      "loss": 1.5762,
+      "step": 10164
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7678285542987475e-05,
+      "loss": 1.5039,
+      "step": 10165
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7590414755901204e-05,
+      "loss": 1.5625,
+      "step": 10166
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.750262304240827e-05,
+      "loss": 1.7012,
+      "step": 10167
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.741491040979773e-05,
+      "loss": 1.5918,
+      "step": 10168
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7327276865351984e-05,
+      "loss": 1.6562,
+      "step": 10169
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7239722416346775e-05,
+      "loss": 1.5059,
+      "step": 10170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.715224707005117e-05,
+      "loss": 1.752,
+      "step": 10171
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.706485083372769e-05,
+      "loss": 1.623,
+      "step": 10172
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.697753371463265e-05,
+      "loss": 1.6094,
+      "step": 10173
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6890295720015465e-05,
+      "loss": 1.6543,
+      "step": 10174
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6803136857119013e-05,
+      "loss": 1.5332,
+      "step": 10175
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6716057133179614e-05,
+      "loss": 1.5176,
+      "step": 10176
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.662905655542704e-05,
+      "loss": 1.5273,
+      "step": 10177
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.654213513108463e-05,
+      "loss": 1.5488,
+      "step": 10178
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6455292867368826e-05,
+      "loss": 1.4922,
+      "step": 10179
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6368529771489644e-05,
+      "loss": 1.5879,
+      "step": 10180
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.628184585065087e-05,
+      "loss": 1.5254,
+      "step": 10181
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.619524111204931e-05,
+      "loss": 1.6875,
+      "step": 10182
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.61087155628751e-05,
+      "loss": 1.5645,
+      "step": 10183
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.602226921031205e-05,
+      "loss": 1.5586,
+      "step": 10184
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5935902061537635e-05,
+      "loss": 1.6953,
+      "step": 10185
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.584961412372224e-05,
+      "loss": 1.5566,
+      "step": 10186
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.57634054040299e-05,
+      "loss": 1.5293,
+      "step": 10187
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.567727590961812e-05,
+      "loss": 1.6113,
+      "step": 10188
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.559122564763773e-05,
+      "loss": 1.5586,
+      "step": 10189
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.550525462523314e-05,
+      "loss": 1.5684,
+      "step": 10190
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.541936284954196e-05,
+      "loss": 1.6348,
+      "step": 10191
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5333550327695484e-05,
+      "loss": 1.5078,
+      "step": 10192
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.524781706681802e-05,
+      "loss": 1.4902,
+      "step": 10193
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.516216307402787e-05,
+      "loss": 1.543,
+      "step": 10194
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.507658835643624e-05,
+      "loss": 1.7051,
+      "step": 10195
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.499109292114789e-05,
+      "loss": 1.5469,
+      "step": 10196
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.490567677526125e-05,
+      "loss": 1.5586,
+      "step": 10197
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.482033992586787e-05,
+      "loss": 1.4766,
+      "step": 10198
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4735082380052864e-05,
+      "loss": 1.584,
+      "step": 10199
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.464990414489456e-05,
+      "loss": 1.582,
+      "step": 10200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.45648052274652e-05,
+      "loss": 1.6055,
+      "step": 10201
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4479785634829794e-05,
+      "loss": 1.498,
+      "step": 10202
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4394845374047256e-05,
+      "loss": 1.5879,
+      "step": 10203
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4309984452169495e-05,
+      "loss": 1.5742,
+      "step": 10204
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.422520287624232e-05,
+      "loss": 1.6113,
+      "step": 10205
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.414050065330466e-05,
+      "loss": 1.5195,
+      "step": 10206
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.405587779038889e-05,
+      "loss": 1.4531,
+      "step": 10207
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.39713342945206e-05,
+      "loss": 1.5938,
+      "step": 10208
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3886870172719416e-05,
+      "loss": 1.582,
+      "step": 10209
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3802485431997494e-05,
+      "loss": 1.5879,
+      "step": 10210
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.371818007936124e-05,
+      "loss": 1.6289,
+      "step": 10211
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3633954121809704e-05,
+      "loss": 1.4902,
+      "step": 10212
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.354980756633609e-05,
+      "loss": 1.5586,
+      "step": 10213
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.346574041992646e-05,
+      "loss": 1.5469,
+      "step": 10214
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.338175268956046e-05,
+      "loss": 1.498,
+      "step": 10215
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.329784438221118e-05,
+      "loss": 1.5352,
+      "step": 10216
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.321401550484516e-05,
+      "loss": 1.5566,
+      "step": 10217
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.313026606442227e-05,
+      "loss": 1.6133,
+      "step": 10218
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.304659606789574e-05,
+      "loss": 1.6543,
+      "step": 10219
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2963005522211996e-05,
+      "loss": 1.7109,
+      "step": 10220
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.28794944343116e-05,
+      "loss": 1.6133,
+      "step": 10221
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.279606281112791e-05,
+      "loss": 1.627,
+      "step": 10222
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2712710659587594e-05,
+      "loss": 1.584,
+      "step": 10223
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.262943798661101e-05,
+      "loss": 1.543,
+      "step": 10224
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.254624479911207e-05,
+      "loss": 1.6523,
+      "step": 10225
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2463131103997575e-05,
+      "loss": 1.498,
+      "step": 10226
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.238009690816835e-05,
+      "loss": 1.5449,
+      "step": 10227
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2297142218517994e-05,
+      "loss": 1.4512,
+      "step": 10228
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2214267041933874e-05,
+      "loss": 1.5605,
+      "step": 10229
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2131471385296824e-05,
+      "loss": 1.4941,
+      "step": 10230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.20487552554808e-05,
+      "loss": 1.6621,
+      "step": 10231
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1966118659353405e-05,
+      "loss": 1.5586,
+      "step": 10232
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.188356160377538e-05,
+      "loss": 1.5156,
+      "step": 10233
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1801084095601016e-05,
+      "loss": 1.5137,
+      "step": 10234
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1718686141677955e-05,
+      "loss": 1.498,
+      "step": 10235
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.16363677488476e-05,
+      "loss": 1.5762,
+      "step": 10236
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.155412892394406e-05,
+      "loss": 1.4746,
+      "step": 10237
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1471969673795296e-05,
+      "loss": 1.5254,
+      "step": 10238
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.138989000522253e-05,
+      "loss": 1.666,
+      "step": 10239
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.130788992504053e-05,
+      "loss": 1.4297,
+      "step": 10240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1225969440057296e-05,
+      "loss": 1.5254,
+      "step": 10241
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.114412855707428e-05,
+      "loss": 1.5195,
+      "step": 10242
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.106236728288604e-05,
+      "loss": 1.5664,
+      "step": 10243
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0980685624281034e-05,
+      "loss": 1.5371,
+      "step": 10244
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.089908358804095e-05,
+      "loss": 1.4863,
+      "step": 10245
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.08175611809406e-05,
+      "loss": 1.6602,
+      "step": 10246
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.073611840974833e-05,
+      "loss": 1.5059,
+      "step": 10247
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0654755281226085e-05,
+      "loss": 1.3457,
+      "step": 10248
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.057347180212889e-05,
+      "loss": 1.5234,
+      "step": 10249
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.049226797920535e-05,
+      "loss": 1.5,
+      "step": 10250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0411143819197304e-05,
+      "loss": 1.5273,
+      "step": 10251
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.033009932884024e-05,
+      "loss": 1.6582,
+      "step": 10252
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0249134514862675e-05,
+      "loss": 1.6484,
+      "step": 10253
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.016824938398689e-05,
+      "loss": 1.7734,
+      "step": 10254
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.008744394292807e-05,
+      "loss": 1.5977,
+      "step": 10255
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0006718198395296e-05,
+      "loss": 1.5449,
+      "step": 10256
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9926072157090765e-05,
+      "loss": 1.6387,
+      "step": 10257
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9845505825710116e-05,
+      "loss": 1.5527,
+      "step": 10258
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.976501921094211e-05,
+      "loss": 1.541,
+      "step": 10259
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.968461231946952e-05,
+      "loss": 1.6445,
+      "step": 10260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.960428515796799e-05,
+      "loss": 1.543,
+      "step": 10261
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9524037733106424e-05,
+      "loss": 1.5156,
+      "step": 10262
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.944387005154748e-05,
+      "loss": 1.5996,
+      "step": 10263
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.936378211994729e-05,
+      "loss": 1.502,
+      "step": 10264
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.928377394495486e-05,
+      "loss": 1.6074,
+      "step": 10265
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.920384553321299e-05,
+      "loss": 1.5195,
+      "step": 10266
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9123996891357594e-05,
+      "loss": 1.541,
+      "step": 10267
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9044228026018146e-05,
+      "loss": 1.4551,
+      "step": 10268
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.896453894381746e-05,
+      "loss": 1.5137,
+      "step": 10269
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.88849296513718e-05,
+      "loss": 1.6523,
+      "step": 10270
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.880540015529055e-05,
+      "loss": 1.6016,
+      "step": 10271
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.872595046217664e-05,
+      "loss": 1.5195,
+      "step": 10272
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8646580578626464e-05,
+      "loss": 1.6523,
+      "step": 10273
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.856729051122965e-05,
+      "loss": 1.4766,
+      "step": 10274
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8488080266569136e-05,
+      "loss": 1.6426,
+      "step": 10275
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8408949851221456e-05,
+      "loss": 1.5977,
+      "step": 10276
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.832989927175634e-05,
+      "loss": 1.5215,
+      "step": 10277
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8250928534736995e-05,
+      "loss": 1.5664,
+      "step": 10278
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8172037646719834e-05,
+      "loss": 1.5098,
+      "step": 10279
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.809322661425496e-05,
+      "loss": 1.5781,
+      "step": 10280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.801449544388547e-05,
+      "loss": 1.5879,
+      "step": 10281
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7935844142148125e-05,
+      "loss": 1.5215,
+      "step": 10282
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.785727271557271e-05,
+      "loss": 1.5781,
+      "step": 10283
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.777878117068278e-05,
+      "loss": 1.668,
+      "step": 10284
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7700369513995135e-05,
+      "loss": 1.3711,
+      "step": 10285
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.762203775201978e-05,
+      "loss": 1.627,
+      "step": 10286
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7543785891260065e-05,
+      "loss": 1.668,
+      "step": 10287
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.746561393821313e-05,
+      "loss": 1.5977,
+      "step": 10288
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.738752189936911e-05,
+      "loss": 1.5195,
+      "step": 10289
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.730950978121139e-05,
+      "loss": 1.4961,
+      "step": 10290
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.723157759021689e-05,
+      "loss": 1.4707,
+      "step": 10291
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.715372533285621e-05,
+      "loss": 1.5156,
+      "step": 10292
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7075953015592854e-05,
+      "loss": 1.5391,
+      "step": 10293
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.699826064488388e-05,
+      "loss": 1.6191,
+      "step": 10294
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6920648227179446e-05,
+      "loss": 1.4512,
+      "step": 10295
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.684311576892363e-05,
+      "loss": 1.4004,
+      "step": 10296
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6765663276553396e-05,
+      "loss": 1.625,
+      "step": 10297
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6688290756499374e-05,
+      "loss": 1.4961,
+      "step": 10298
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6610998215185095e-05,
+      "loss": 1.5586,
+      "step": 10299
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6533785659027986e-05,
+      "loss": 1.5527,
+      "step": 10300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6456653094438574e-05,
+      "loss": 1.5762,
+      "step": 10301
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6379600527820856e-05,
+      "loss": 1.625,
+      "step": 10302
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.630262796557171e-05,
+      "loss": 1.6387,
+      "step": 10303
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.622573541408225e-05,
+      "loss": 1.5918,
+      "step": 10304
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.614892287973637e-05,
+      "loss": 1.5215,
+      "step": 10305
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.607219036891118e-05,
+      "loss": 1.666,
+      "step": 10306
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.599553788797738e-05,
+      "loss": 1.6875,
+      "step": 10307
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.591896544329931e-05,
+      "loss": 1.5879,
+      "step": 10308
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.584247304123422e-05,
+      "loss": 1.6543,
+      "step": 10309
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.576606068813293e-05,
+      "loss": 1.5371,
+      "step": 10310
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.568972839033946e-05,
+      "loss": 1.6523,
+      "step": 10311
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5613476154191304e-05,
+      "loss": 1.5645,
+      "step": 10312
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.553730398601929e-05,
+      "loss": 1.5566,
+      "step": 10313
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.546121189214757e-05,
+      "loss": 1.5762,
+      "step": 10314
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.538519987889388e-05,
+      "loss": 1.4844,
+      "step": 10315
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5309267952568835e-05,
+      "loss": 1.6973,
+      "step": 10316
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5233416119476834e-05,
+      "loss": 1.7441,
+      "step": 10317
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5157644385915174e-05,
+      "loss": 1.5781,
+      "step": 10318
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.508195275817527e-05,
+      "loss": 1.4727,
+      "step": 10319
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.50063412425411e-05,
+      "loss": 1.5332,
+      "step": 10320
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.493080984529029e-05,
+      "loss": 1.5078,
+      "step": 10321
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.485535857269373e-05,
+      "loss": 1.5684,
+      "step": 10322
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.477998743101607e-05,
+      "loss": 1.623,
+      "step": 10323
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.470469642651475e-05,
+      "loss": 1.5527,
+      "step": 10324
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4629485565440764e-05,
+      "loss": 1.541,
+      "step": 10325
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.455435485403846e-05,
+      "loss": 1.5254,
+      "step": 10326
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.447930429854585e-05,
+      "loss": 1.4961,
+      "step": 10327
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.440433390519371e-05,
+      "loss": 1.6582,
+      "step": 10328
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.432944368020641e-05,
+      "loss": 1.4844,
+      "step": 10329
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4254633629801745e-05,
+      "loss": 1.3613,
+      "step": 10330
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.417990376019098e-05,
+      "loss": 1.6016,
+      "step": 10331
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.410525407757836e-05,
+      "loss": 1.5957,
+      "step": 10332
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.403068458816172e-05,
+      "loss": 1.7285,
+      "step": 10333
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3956195298132096e-05,
+      "loss": 1.5547,
+      "step": 10334
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3881786213673995e-05,
+      "loss": 1.416,
+      "step": 10335
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3807457340965354e-05,
+      "loss": 1.6133,
+      "step": 10336
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.373320868617713e-05,
+      "loss": 1.5898,
+      "step": 10337
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.365904025547373e-05,
+      "loss": 1.668,
+      "step": 10338
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.358495205501322e-05,
+      "loss": 1.4297,
+      "step": 10339
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3510944090946685e-05,
+      "loss": 1.6211,
+      "step": 10340
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3437016369418535e-05,
+      "loss": 1.5391,
+      "step": 10341
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3363168896566545e-05,
+      "loss": 1.4707,
+      "step": 10342
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.328940167852201e-05,
+      "loss": 1.5508,
+      "step": 10343
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.32157147214095e-05,
+      "loss": 1.4395,
+      "step": 10344
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.314210803134676e-05,
+      "loss": 1.5625,
+      "step": 10345
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.306858161444493e-05,
+      "loss": 1.582,
+      "step": 10346
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2995135476808433e-05,
+      "loss": 1.5918,
+      "step": 10347
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.292176962453541e-05,
+      "loss": 1.5371,
+      "step": 10348
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.284848406371688e-05,
+      "loss": 1.5312,
+      "step": 10349
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.277527880043729e-05,
+      "loss": 1.6523,
+      "step": 10350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2702153840774686e-05,
+      "loss": 1.6777,
+      "step": 10351
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.262910919080009e-05,
+      "loss": 1.6289,
+      "step": 10352
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2556144856577984e-05,
+      "loss": 1.5703,
+      "step": 10353
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.248326084416631e-05,
+      "loss": 1.4902,
+      "step": 10354
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2410457159616217e-05,
+      "loss": 1.6035,
+      "step": 10355
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.233773380897231e-05,
+      "loss": 1.6562,
+      "step": 10356
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2265090798272225e-05,
+      "loss": 1.6562,
+      "step": 10357
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.219252813354723e-05,
+      "loss": 1.6113,
+      "step": 10358
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2120045820821954e-05,
+      "loss": 1.5762,
+      "step": 10359
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.204764386611414e-05,
+      "loss": 1.5547,
+      "step": 10360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.197532227543487e-05,
+      "loss": 1.5938,
+      "step": 10361
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.190308105478868e-05,
+      "loss": 1.5879,
+      "step": 10362
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.183092021017331e-05,
+      "loss": 1.4473,
+      "step": 10363
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.175883974758009e-05,
+      "loss": 1.7129,
+      "step": 10364
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.168683967299335e-05,
+      "loss": 1.4355,
+      "step": 10365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1614919992390746e-05,
+      "loss": 1.5801,
+      "step": 10366
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.154308071174372e-05,
+      "loss": 1.5977,
+      "step": 10367
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.14713218370164e-05,
+      "loss": 1.5195,
+      "step": 10368
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.139964337416678e-05,
+      "loss": 1.541,
+      "step": 10369
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1328045329145686e-05,
+      "loss": 1.5488,
+      "step": 10370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.125652770789789e-05,
+      "loss": 1.5137,
+      "step": 10371
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.118509051636087e-05,
+      "loss": 1.4688,
+      "step": 10372
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.111373376046578e-05,
+      "loss": 1.6172,
+      "step": 10373
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.104245744613676e-05,
+      "loss": 1.6426,
+      "step": 10374
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.097126157929198e-05,
+      "loss": 1.5723,
+      "step": 10375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.090014616584202e-05,
+      "loss": 1.5215,
+      "step": 10376
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.082911121169152e-05,
+      "loss": 1.4219,
+      "step": 10377
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.075815672273785e-05,
+      "loss": 1.5859,
+      "step": 10378
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.068728270487231e-05,
+      "loss": 1.6426,
+      "step": 10379
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.061648916397897e-05,
+      "loss": 1.541,
+      "step": 10380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0545776105935676e-05,
+      "loss": 1.5703,
+      "step": 10381
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.047514353661296e-05,
+      "loss": 1.5449,
+      "step": 10382
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0404591461875464e-05,
+      "loss": 1.541,
+      "step": 10383
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0334119887580614e-05,
+      "loss": 1.5645,
+      "step": 10384
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0263728819579396e-05,
+      "loss": 1.4805,
+      "step": 10385
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.019341826371569e-05,
+      "loss": 1.5,
+      "step": 10386
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.012318822582738e-05,
+      "loss": 1.5957,
+      "step": 10387
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.005303871174525e-05,
+      "loss": 1.6406,
+      "step": 10388
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9982969727293306e-05,
+      "loss": 1.6133,
+      "step": 10389
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.991298127828901e-05,
+      "loss": 1.5273,
+      "step": 10390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9843073370543262e-05,
+      "loss": 1.6152,
+      "step": 10391
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9773246009860088e-05,
+      "loss": 1.6855,
+      "step": 10392
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9703499202036855e-05,
+      "loss": 1.5508,
+      "step": 10393
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.963383295286426e-05,
+      "loss": 1.6211,
+      "step": 10394
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9564247268126344e-05,
+      "loss": 1.5879,
+      "step": 10395
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9494742153600594e-05,
+      "loss": 1.6523,
+      "step": 10396
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.94253176150574e-05,
+      "loss": 1.5781,
+      "step": 10397
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9355973658260926e-05,
+      "loss": 1.5508,
+      "step": 10398
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9286710288968454e-05,
+      "loss": 1.5996,
+      "step": 10399
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9217527512930386e-05,
+      "loss": 1.5762,
+      "step": 10400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9148425335890572e-05,
+      "loss": 1.5215,
+      "step": 10401
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9079403763586422e-05,
+      "loss": 1.5449,
+      "step": 10402
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.901046280174846e-05,
+      "loss": 1.582,
+      "step": 10403
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8941602456100225e-05,
+      "loss": 1.5059,
+      "step": 10404
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8872822732358916e-05,
+      "loss": 1.498,
+      "step": 10405
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8804123636235192e-05,
+      "loss": 1.5723,
+      "step": 10406
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8735505173432598e-05,
+      "loss": 1.7012,
+      "step": 10407
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.866696734964813e-05,
+      "loss": 1.5352,
+      "step": 10408
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8598510170572023e-05,
+      "loss": 1.5137,
+      "step": 10409
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.853013364188828e-05,
+      "loss": 1.5508,
+      "step": 10410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8461837769273578e-05,
+      "loss": 1.5391,
+      "step": 10411
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.839362255839817e-05,
+      "loss": 1.6074,
+      "step": 10412
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8325488014925627e-05,
+      "loss": 1.5996,
+      "step": 10413
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8257434144512984e-05,
+      "loss": 1.5605,
+      "step": 10414
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8189460952810163e-05,
+      "loss": 1.6113,
+      "step": 10415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8121568445460767e-05,
+      "loss": 1.582,
+      "step": 10416
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.805375662810139e-05,
+      "loss": 1.6719,
+      "step": 10417
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.79860255063622e-05,
+      "loss": 1.5449,
+      "step": 10418
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.79183750858667e-05,
+      "loss": 1.6582,
+      "step": 10419
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.785080537223139e-05,
+      "loss": 1.5938,
+      "step": 10420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7783316371066126e-05,
+      "loss": 1.6094,
+      "step": 10421
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7715908087974416e-05,
+      "loss": 1.7324,
+      "step": 10422
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.764858052855268e-05,
+      "loss": 1.4961,
+      "step": 10423
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7581333698390778e-05,
+      "loss": 1.5664,
+      "step": 10424
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7514167603071905e-05,
+      "loss": 1.5449,
+      "step": 10425
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7447082248172384e-05,
+      "loss": 1.6074,
+      "step": 10426
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7380077639262202e-05,
+      "loss": 1.5586,
+      "step": 10427
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7313153781904133e-05,
+      "loss": 1.5449,
+      "step": 10428
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.724631068165473e-05,
+      "loss": 1.4883,
+      "step": 10429
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7179548344063554e-05,
+      "loss": 1.541,
+      "step": 10430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7112866774673507e-05,
+      "loss": 1.6387,
+      "step": 10431
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7046265979020823e-05,
+      "loss": 1.5137,
+      "step": 10432
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.697974596263486e-05,
+      "loss": 1.459,
+      "step": 10433
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6913306731038756e-05,
+      "loss": 1.4902,
+      "step": 10434
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.684694828974843e-05,
+      "loss": 1.541,
+      "step": 10435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6780670644273254e-05,
+      "loss": 1.6621,
+      "step": 10436
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6714473800115825e-05,
+      "loss": 1.5469,
+      "step": 10437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6648357762772413e-05,
+      "loss": 1.6094,
+      "step": 10438
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6582322537732075e-05,
+      "loss": 1.5488,
+      "step": 10439
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.651636813047742e-05,
+      "loss": 1.6055,
+      "step": 10440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6450494546484184e-05,
+      "loss": 1.6211,
+      "step": 10441
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6384701791221764e-05,
+      "loss": 1.5957,
+      "step": 10442
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6318989870152464e-05,
+      "loss": 1.4531,
+      "step": 10443
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6253358788731917e-05,
+      "loss": 1.6172,
+      "step": 10444
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6187808552409096e-05,
+      "loss": 1.4473,
+      "step": 10445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.612233916662654e-05,
+      "loss": 1.5977,
+      "step": 10446
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6056950636819676e-05,
+      "loss": 1.8281,
+      "step": 10447
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5991642968417273e-05,
+      "loss": 1.5117,
+      "step": 10448
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.592641616684155e-05,
+      "loss": 1.5176,
+      "step": 10449
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.586127023750817e-05,
+      "loss": 1.541,
+      "step": 10450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5796205185825594e-05,
+      "loss": 1.5293,
+      "step": 10451
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5731221017195937e-05,
+      "loss": 1.6152,
+      "step": 10452
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5666317737014333e-05,
+      "loss": 1.4941,
+      "step": 10453
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.560149535066969e-05,
+      "loss": 1.4609,
+      "step": 10454
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5536753863543595e-05,
+      "loss": 1.6094,
+      "step": 10455
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.54720932810113e-05,
+      "loss": 1.6465,
+      "step": 10456
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.540751360844118e-05,
+      "loss": 1.4941,
+      "step": 10457
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5343014851195056e-05,
+      "loss": 1.5645,
+      "step": 10458
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5278597014627758e-05,
+      "loss": 1.5859,
+      "step": 10459
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5214260104087783e-05,
+      "loss": 1.6074,
+      "step": 10460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5150004124916413e-05,
+      "loss": 1.5547,
+      "step": 10461
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.508582908244883e-05,
+      "loss": 1.6953,
+      "step": 10462
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.502173498201288e-05,
+      "loss": 1.5977,
+      "step": 10463
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.495772182893008e-05,
+      "loss": 1.4316,
+      "step": 10464
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.489378962851485e-05,
+      "loss": 1.6172,
+      "step": 10465
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4829938386075613e-05,
+      "loss": 1.502,
+      "step": 10466
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4766168106913345e-05,
+      "loss": 1.5332,
+      "step": 10467
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.470247879632248e-05,
+      "loss": 1.5293,
+      "step": 10468
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.463887045959101e-05,
+      "loss": 1.5117,
+      "step": 10469
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4575343101999825e-05,
+      "loss": 1.6348,
+      "step": 10470
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4511896728823368e-05,
+      "loss": 1.5273,
+      "step": 10471
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4448531345329206e-05,
+      "loss": 1.6035,
+      "step": 10472
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4385246956778128e-05,
+      "loss": 1.6523,
+      "step": 10473
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.43220435684246e-05,
+      "loss": 1.543,
+      "step": 10474
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.425892118551598e-05,
+      "loss": 1.5215,
+      "step": 10475
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4195879813292744e-05,
+      "loss": 1.5293,
+      "step": 10476
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4132919456989145e-05,
+      "loss": 1.6699,
+      "step": 10477
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4070040121832338e-05,
+      "loss": 1.5586,
+      "step": 10478
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4007241813042923e-05,
+      "loss": 1.5938,
+      "step": 10479
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3944524535834733e-05,
+      "loss": 1.6289,
+      "step": 10480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.388188829541471e-05,
+      "loss": 1.5469,
+      "step": 10481
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3819333096983366e-05,
+      "loss": 1.6445,
+      "step": 10482
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3756858945734317e-05,
+      "loss": 1.5273,
+      "step": 10483
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3694465846854307e-05,
+      "loss": 1.5117,
+      "step": 10484
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3632153805523747e-05,
+      "loss": 1.6934,
+      "step": 10485
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3569922826916057e-05,
+      "loss": 1.4473,
+      "step": 10486
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3507772916197767e-05,
+      "loss": 1.6738,
+      "step": 10487
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.344570407852886e-05,
+      "loss": 1.6934,
+      "step": 10488
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3383716319062774e-05,
+      "loss": 1.4922,
+      "step": 10489
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3321809642945944e-05,
+      "loss": 1.6797,
+      "step": 10490
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.325998405531815e-05,
+      "loss": 1.5742,
+      "step": 10491
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3198239561312394e-05,
+      "loss": 1.5449,
+      "step": 10492
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3136576166055025e-05,
+      "loss": 1.6699,
+      "step": 10493
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3074993874665828e-05,
+      "loss": 1.6406,
+      "step": 10494
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.301349269225739e-05,
+      "loss": 1.5977,
+      "step": 10495
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.295207262393584e-05,
+      "loss": 1.5684,
+      "step": 10496
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.289073367480077e-05,
+      "loss": 1.6875,
+      "step": 10497
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.282947584994477e-05,
+      "loss": 1.4766,
+      "step": 10498
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2768299154453664e-05,
+      "loss": 1.623,
+      "step": 10499
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.27072035934065e-05,
+      "loss": 1.6582,
+      "step": 10500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2646189171876107e-05,
+      "loss": 1.6484,
+      "step": 10501
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.258525589492799e-05,
+      "loss": 1.5156,
+      "step": 10502
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2524403767621104e-05,
+      "loss": 1.7129,
+      "step": 10503
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.246363279500763e-05,
+      "loss": 1.6172,
+      "step": 10504
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.240294298213308e-05,
+      "loss": 1.5879,
+      "step": 10505
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.234233433403643e-05,
+      "loss": 1.6621,
+      "step": 10506
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2281806855749432e-05,
+      "loss": 1.6523,
+      "step": 10507
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.222136055229751e-05,
+      "loss": 1.6406,
+      "step": 10508
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2160995428699204e-05,
+      "loss": 1.5098,
+      "step": 10509
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2100711489966175e-05,
+      "loss": 1.5039,
+      "step": 10510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.204050874110364e-05,
+      "loss": 1.5,
+      "step": 10511
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1980387187109817e-05,
+      "loss": 1.5312,
+      "step": 10512
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.192034683297639e-05,
+      "loss": 1.4785,
+      "step": 10513
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1860387683688142e-05,
+      "loss": 1.6094,
+      "step": 10514
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1800509744223097e-05,
+      "loss": 1.4844,
+      "step": 10515
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1740713019552714e-05,
+      "loss": 1.5879,
+      "step": 10516
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1680997514641477e-05,
+      "loss": 1.5312,
+      "step": 10517
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1621363234447524e-05,
+      "loss": 1.4961,
+      "step": 10518
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.156181018392167e-05,
+      "loss": 1.4863,
+      "step": 10519
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.15023383680083e-05,
+      "loss": 1.6387,
+      "step": 10520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.144294779164535e-05,
+      "loss": 1.5742,
+      "step": 10521
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1383638459763433e-05,
+      "loss": 1.4785,
+      "step": 10522
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1324410377286828e-05,
+      "loss": 1.6426,
+      "step": 10523
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.126526354913283e-05,
+      "loss": 1.6465,
+      "step": 10524
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1206197980212174e-05,
+      "loss": 1.4746,
+      "step": 10525
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.114721367542882e-05,
+      "loss": 1.5801,
+      "step": 10526
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1088310639679753e-05,
+      "loss": 1.4766,
+      "step": 10527
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1029488877855497e-05,
+      "loss": 1.5234,
+      "step": 10528
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.09707483948397e-05,
+      "loss": 1.6875,
+      "step": 10529
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0912089195509244e-05,
+      "loss": 1.582,
+      "step": 10530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.085351128473445e-05,
+      "loss": 1.5566,
+      "step": 10531
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.079501466737843e-05,
+      "loss": 1.625,
+      "step": 10532
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.073659934829808e-05,
+      "loss": 1.6816,
+      "step": 10533
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0678265332343405e-05,
+      "loss": 1.4824,
+      "step": 10534
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.062001262435742e-05,
+      "loss": 1.6113,
+      "step": 10535
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.056184122917637e-05,
+      "loss": 1.7168,
+      "step": 10536
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.050375115163028e-05,
+      "loss": 1.6172,
+      "step": 10537
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0445742396541955e-05,
+      "loss": 1.6816,
+      "step": 10538
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0387814968727438e-05,
+      "loss": 1.6875,
+      "step": 10539
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0329968872996095e-05,
+      "loss": 1.5176,
+      "step": 10540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0272204114150873e-05,
+      "loss": 1.4688,
+      "step": 10541
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0214520696987372e-05,
+      "loss": 1.7383,
+      "step": 10542
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0156918626294985e-05,
+      "loss": 1.543,
+      "step": 10543
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0099397906855887e-05,
+      "loss": 1.6055,
+      "step": 10544
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0041958543445925e-05,
+      "loss": 1.4707,
+      "step": 10545
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984600540833954e-05,
+      "loss": 1.6758,
+      "step": 10546
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9927323903782045e-05,
+      "loss": 1.623,
+      "step": 10547
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9870128637045627e-05,
+      "loss": 1.5977,
+      "step": 10548
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9813014745373225e-05,
+      "loss": 1.6309,
+      "step": 10549
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.975598223350672e-05,
+      "loss": 1.6719,
+      "step": 10550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.969903110618132e-05,
+      "loss": 1.6113,
+      "step": 10551
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9642161368125357e-05,
+      "loss": 1.5625,
+      "step": 10552
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.958537302406038e-05,
+      "loss": 1.6562,
+      "step": 10553
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9528666078701296e-05,
+      "loss": 1.5488,
+      "step": 10554
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9472040536756107e-05,
+      "loss": 1.627,
+      "step": 10555
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.941549640292617e-05,
+      "loss": 1.5801,
+      "step": 10556
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.935903368190606e-05,
+      "loss": 1.6367,
+      "step": 10557
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.930265237838358e-05,
+      "loss": 1.4766,
+      "step": 10558
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9246352497039765e-05,
+      "loss": 1.5879,
+      "step": 10559
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9190134042548878e-05,
+      "loss": 1.5547,
+      "step": 10560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9133997019578518e-05,
+      "loss": 1.6641,
+      "step": 10561
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9077941432789402e-05,
+      "loss": 1.5449,
+      "step": 10562
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9021967286835583e-05,
+      "loss": 1.4922,
+      "step": 10563
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8966074586364125e-05,
+      "loss": 1.5703,
+      "step": 10564
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8910263336015754e-05,
+      "loss": 1.75,
+      "step": 10565
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.88545335404241e-05,
+      "loss": 1.6738,
+      "step": 10566
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.879888520421591e-05,
+      "loss": 1.5488,
+      "step": 10567
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8743318332011596e-05,
+      "loss": 1.4609,
+      "step": 10568
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8687832928424687e-05,
+      "loss": 1.502,
+      "step": 10569
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8632428998061613e-05,
+      "loss": 1.5527,
+      "step": 10570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8577106545522247e-05,
+      "loss": 1.5156,
+      "step": 10571
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.852186557539992e-05,
+      "loss": 1.6367,
+      "step": 10572
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.846670609228096e-05,
+      "loss": 1.6445,
+      "step": 10573
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8411628100744925e-05,
+      "loss": 1.5879,
+      "step": 10574
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8356631605364603e-05,
+      "loss": 1.5352,
+      "step": 10575
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8301716610706122e-05,
+      "loss": 1.6738,
+      "step": 10576
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8246883121328826e-05,
+      "loss": 1.541,
+      "step": 10577
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8192131141785194e-05,
+      "loss": 1.541,
+      "step": 10578
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.813746067662092e-05,
+      "loss": 1.5625,
+      "step": 10579
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.808287173037515e-05,
+      "loss": 1.5859,
+      "step": 10580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8028364307580147e-05,
+      "loss": 1.5684,
+      "step": 10581
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7973938412761183e-05,
+      "loss": 1.6797,
+      "step": 10582
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7919594050437082e-05,
+      "loss": 1.4746,
+      "step": 10583
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.786533122511991e-05,
+      "loss": 1.6191,
+      "step": 10584
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7811149941314498e-05,
+      "loss": 1.7031,
+      "step": 10585
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7757050203519587e-05,
+      "loss": 1.5645,
+      "step": 10586
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7703032016226472e-05,
+      "loss": 1.5156,
+      "step": 10587
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7649095383920234e-05,
+      "loss": 1.5879,
+      "step": 10588
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7595240311078952e-05,
+      "loss": 1.6191,
+      "step": 10589
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.754146680217372e-05,
+      "loss": 1.5918,
+      "step": 10590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7487774861669193e-05,
+      "loss": 1.4453,
+      "step": 10591
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.743416449402313e-05,
+      "loss": 1.5742,
+      "step": 10592
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.738063570368664e-05,
+      "loss": 1.6953,
+      "step": 10593
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.732718849510373e-05,
+      "loss": 1.6875,
+      "step": 10594
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.727382287271195e-05,
+      "loss": 1.5566,
+      "step": 10595
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.722053884094199e-05,
+      "loss": 1.543,
+      "step": 10596
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7167336404217637e-05,
+      "loss": 1.4766,
+      "step": 10597
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.711421556695625e-05,
+      "loss": 1.5391,
+      "step": 10598
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7061176333567852e-05,
+      "loss": 1.5645,
+      "step": 10599
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7008218708456148e-05,
+      "loss": 1.5801,
+      "step": 10600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.695534269601806e-05,
+      "loss": 1.5488,
+      "step": 10601
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6902548300643527e-05,
+      "loss": 1.6797,
+      "step": 10602
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6849835526715707e-05,
+      "loss": 1.5801,
+      "step": 10603
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6797204378611208e-05,
+      "loss": 1.5898,
+      "step": 10604
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6744654860699536e-05,
+      "loss": 1.6035,
+      "step": 10605
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.669218697734387e-05,
+      "loss": 1.6387,
+      "step": 10606
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6639800732899946e-05,
+      "loss": 1.5078,
+      "step": 10607
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6587496131717505e-05,
+      "loss": 1.541,
+      "step": 10608
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6535273178139076e-05,
+      "loss": 1.5312,
+      "step": 10609
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.648313187650019e-05,
+      "loss": 1.5859,
+      "step": 10610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6431072231130052e-05,
+      "loss": 1.4648,
+      "step": 10611
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.637909424635098e-05,
+      "loss": 1.623,
+      "step": 10612
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.63271979264783e-05,
+      "loss": 1.6797,
+      "step": 10613
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6275383275820677e-05,
+      "loss": 1.6465,
+      "step": 10614
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6223650298680116e-05,
+      "loss": 1.5059,
+      "step": 10615
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6171998999351733e-05,
+      "loss": 1.5371,
+      "step": 10616
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6120429382123768e-05,
+      "loss": 1.5586,
+      "step": 10617
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6068941451277907e-05,
+      "loss": 1.5332,
+      "step": 10618
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.601753521108862e-05,
+      "loss": 1.5645,
+      "step": 10619
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5966210665824377e-05,
+      "loss": 1.4453,
+      "step": 10620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5914967819745997e-05,
+      "loss": 1.3887,
+      "step": 10621
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5863806677108182e-05,
+      "loss": 1.5703,
+      "step": 10622
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5812727242158208e-05,
+      "loss": 1.498,
+      "step": 10623
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5761729519137347e-05,
+      "loss": 1.5684,
+      "step": 10624
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5710813512279432e-05,
+      "loss": 1.6035,
+      "step": 10625
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.565997922581186e-05,
+      "loss": 1.6602,
+      "step": 10626
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5609226663955033e-05,
+      "loss": 1.5195,
+      "step": 10627
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5558555830922805e-05,
+      "loss": 1.4805,
+      "step": 10628
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.550796673092192e-05,
+      "loss": 1.7539,
+      "step": 10629
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5457459368152683e-05,
+      "loss": 1.5918,
+      "step": 10630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5407033746808408e-05,
+      "loss": 1.5469,
+      "step": 10631
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5356689871075637e-05,
+      "loss": 1.543,
+      "step": 10632
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5306427745134244e-05,
+      "loss": 1.5938,
+      "step": 10633
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5256247373157227e-05,
+      "loss": 1.5039,
+      "step": 10634
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.520614875931059e-05,
+      "loss": 1.5488,
+      "step": 10635
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5156131907754e-05,
+      "loss": 1.6406,
+      "step": 10636
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5106196822640138e-05,
+      "loss": 1.6797,
+      "step": 10637
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.505634350811469e-05,
+      "loss": 1.6211,
+      "step": 10638
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5006571968316563e-05,
+      "loss": 1.5371,
+      "step": 10639
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4956882207378453e-05,
+      "loss": 1.6211,
+      "step": 10640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4907274229425505e-05,
+      "loss": 1.459,
+      "step": 10641
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4857748038576647e-05,
+      "loss": 1.6953,
+      "step": 10642
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4808303638943476e-05,
+      "loss": 1.6055,
+      "step": 10643
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.475894103463138e-05,
+      "loss": 1.5449,
+      "step": 10644
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.470966022973852e-05,
+      "loss": 1.4941,
+      "step": 10645
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4660461228356515e-05,
+      "loss": 1.623,
+      "step": 10646
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4611344034570095e-05,
+      "loss": 1.5488,
+      "step": 10647
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4562308652457112e-05,
+      "loss": 1.5098,
+      "step": 10648
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4513355086088864e-05,
+      "loss": 1.627,
+      "step": 10649
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4464483339529544e-05,
+      "loss": 1.6309,
+      "step": 10650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4415693416836795e-05,
+      "loss": 1.5176,
+      "step": 10651
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4366985322061488e-05,
+      "loss": 1.7461,
+      "step": 10652
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.43183590592475e-05,
+      "loss": 1.6895,
+      "step": 10653
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4269814632431932e-05,
+      "loss": 1.6152,
+      "step": 10654
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4221352045645341e-05,
+      "loss": 1.5859,
+      "step": 10655
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4172971302911286e-05,
+      "loss": 1.5938,
+      "step": 10656
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4124672408246442e-05,
+      "loss": 1.5254,
+      "step": 10657
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4076455365660823e-05,
+      "loss": 1.5059,
+      "step": 10658
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4028320179157894e-05,
+      "loss": 1.5508,
+      "step": 10659
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3980266852733903e-05,
+      "loss": 1.4941,
+      "step": 10660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3932295390378324e-05,
+      "loss": 1.5488,
+      "step": 10661
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3884405796074084e-05,
+      "loss": 1.5605,
+      "step": 10662
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3836598073797336e-05,
+      "loss": 1.6758,
+      "step": 10663
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3788872227517235e-05,
+      "loss": 1.6719,
+      "step": 10664
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.374122826119617e-05,
+      "loss": 1.4883,
+      "step": 10665
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3693666178789754e-05,
+      "loss": 1.7246,
+      "step": 10666
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3646185984246717e-05,
+      "loss": 1.6875,
+      "step": 10667
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.359878768150935e-05,
+      "loss": 1.4844,
+      "step": 10668
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3551471274512727e-05,
+      "loss": 1.5996,
+      "step": 10669
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3504236767185263e-05,
+      "loss": 1.5195,
+      "step": 10670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3457084163448597e-05,
+      "loss": 1.5547,
+      "step": 10671
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3410013467217708e-05,
+      "loss": 1.5762,
+      "step": 10672
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.336302468240047e-05,
+      "loss": 1.5605,
+      "step": 10673
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3316117812898098e-05,
+      "loss": 1.5586,
+      "step": 10674
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3269292862605253e-05,
+      "loss": 1.5312,
+      "step": 10675
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.322254983540927e-05,
+      "loss": 1.584,
+      "step": 10676
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3175888735191155e-05,
+      "loss": 1.5195,
+      "step": 10677
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.312930956582492e-05,
+      "loss": 1.5801,
+      "step": 10678
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3082812331177696e-05,
+      "loss": 1.582,
+      "step": 10679
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3036397035110059e-05,
+      "loss": 1.6797,
+      "step": 10680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2990063681475594e-05,
+      "loss": 1.668,
+      "step": 10681
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2943812274120893e-05,
+      "loss": 1.502,
+      "step": 10682
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2897642816886323e-05,
+      "loss": 1.5449,
+      "step": 10683
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2851555313604824e-05,
+      "loss": 1.5488,
+      "step": 10684
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2805549768102997e-05,
+      "loss": 1.5391,
+      "step": 10685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2759626184200123e-05,
+      "loss": 1.459,
+      "step": 10686
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2713784565709375e-05,
+      "loss": 1.6816,
+      "step": 10687
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2668024916436593e-05,
+      "loss": 1.6387,
+      "step": 10688
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2622347240180854e-05,
+      "loss": 1.6367,
+      "step": 10689
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2576751540734677e-05,
+      "loss": 1.5996,
+      "step": 10690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.253123782188359e-05,
+      "loss": 1.6094,
+      "step": 10691
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2485806087406348e-05,
+      "loss": 1.5605,
+      "step": 10692
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2440456341075047e-05,
+      "loss": 1.582,
+      "step": 10693
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2395188586654448e-05,
+      "loss": 1.5957,
+      "step": 10694
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2350002827903329e-05,
+      "loss": 1.5781,
+      "step": 10695
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2304899068573128e-05,
+      "loss": 1.748,
+      "step": 10696
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2259877312408407e-05,
+      "loss": 1.5918,
+      "step": 10697
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2214937563147177e-05,
+      "loss": 1.5312,
+      "step": 10698
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2170079824520563e-05,
+      "loss": 1.5605,
+      "step": 10699
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.212530410025292e-05,
+      "loss": 1.418,
+      "step": 10700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.208061039406172e-05,
+      "loss": 1.5098,
+      "step": 10701
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2035998709657548e-05,
+      "loss": 1.6816,
+      "step": 10702
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.199146905074433e-05,
+      "loss": 1.582,
+      "step": 10703
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1947021421019222e-05,
+      "loss": 1.5117,
+      "step": 10704
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.190265582417238e-05,
+      "loss": 1.498,
+      "step": 10705
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1858372263887307e-05,
+      "loss": 1.5312,
+      "step": 10706
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1814170743840503e-05,
+      "loss": 1.5039,
+      "step": 10707
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.177005126770192e-05,
+      "loss": 1.5371,
+      "step": 10708
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1726013839134519e-05,
+      "loss": 1.6191,
+      "step": 10709
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1682058461794599e-05,
+      "loss": 1.6504,
+      "step": 10710
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1638185139331459e-05,
+      "loss": 1.75,
+      "step": 10711
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1594393875387632e-05,
+      "loss": 1.5352,
+      "step": 10712
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1550684673598988e-05,
+      "loss": 1.6035,
+      "step": 10713
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.150705753759429e-05,
+      "loss": 1.543,
+      "step": 10714
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1463512470995751e-05,
+      "loss": 1.6309,
+      "step": 10715
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1420049477418926e-05,
+      "loss": 1.6367,
+      "step": 10716
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1376668560471926e-05,
+      "loss": 1.5176,
+      "step": 10717
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1333369723756759e-05,
+      "loss": 1.543,
+      "step": 10718
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1290152970868106e-05,
+      "loss": 1.5723,
+      "step": 10719
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1247018305394207e-05,
+      "loss": 1.6641,
+      "step": 10720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1203965730916088e-05,
+      "loss": 1.5645,
+      "step": 10721
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1160995251008332e-05,
+      "loss": 1.752,
+      "step": 10722
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1118106869238532e-05,
+      "loss": 1.5508,
+      "step": 10723
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1075300589167392e-05,
+      "loss": 1.6543,
+      "step": 10724
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1032576414349071e-05,
+      "loss": 1.6895,
+      "step": 10725
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0989934348330622e-05,
+      "loss": 1.5996,
+      "step": 10726
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.094737439465232e-05,
+      "loss": 1.7871,
+      "step": 10727
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0904896556847898e-05,
+      "loss": 1.5352,
+      "step": 10728
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0862500838443979e-05,
+      "loss": 1.5,
+      "step": 10729
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.08201872429603e-05,
+      "loss": 1.4688,
+      "step": 10730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0777955773910053e-05,
+      "loss": 1.7285,
+      "step": 10731
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0735806434799655e-05,
+      "loss": 1.623,
+      "step": 10732
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0693739229128308e-05,
+      "loss": 1.5488,
+      "step": 10733
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0651754160388771e-05,
+      "loss": 1.459,
+      "step": 10734
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0609851232066813e-05,
+      "loss": 1.5293,
+      "step": 10735
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0568030447641319e-05,
+      "loss": 1.6406,
+      "step": 10736
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0526291810584621e-05,
+      "loss": 1.6035,
+      "step": 10737
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0484635324361947e-05,
+      "loss": 1.5391,
+      "step": 10738
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0443060992431864e-05,
+      "loss": 1.625,
+      "step": 10739
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0401568818246054e-05,
+      "loss": 1.6797,
+      "step": 10740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.036015880524932e-05,
+      "loss": 1.5566,
+      "step": 10741
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0318830956879799e-05,
+      "loss": 1.5156,
+      "step": 10742
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0277585276568746e-05,
+      "loss": 1.709,
+      "step": 10743
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0236421767740645e-05,
+      "loss": 1.584,
+      "step": 10744
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0195340433812872e-05,
+      "loss": 1.5586,
+      "step": 10745
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0154341278196366e-05,
+      "loss": 1.4004,
+      "step": 10746
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0113424304295071e-05,
+      "loss": 1.5352,
+      "step": 10747
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0072589515506048e-05,
+      "loss": 1.5898,
+      "step": 10748
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0031836915219472e-05,
+      "loss": 1.6445,
+      "step": 10749
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.991166506819083e-06,
+      "loss": 1.5156,
+      "step": 10750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.9505782936814e-06,
+      "loss": 1.6582,
+      "step": 10751
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.910072279176175e-06,
+      "loss": 1.543,
+      "step": 10752
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.869648466666493e-06,
+      "loss": 1.5664,
+      "step": 10753
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.829306859508557e-06,
+      "loss": 1.5781,
+      "step": 10754
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.789047461051803e-06,
+      "loss": 1.5664,
+      "step": 10755
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.748870274638555e-06,
+      "loss": 1.5918,
+      "step": 10756
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.708775303604588e-06,
+      "loss": 1.4219,
+      "step": 10757
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.668762551278797e-06,
+      "loss": 1.5957,
+      "step": 10758
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.628832020983302e-06,
+      "loss": 1.5547,
+      "step": 10759
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.588983716033228e-06,
+      "loss": 1.5527,
+      "step": 10760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.549217639736929e-06,
+      "loss": 1.5234,
+      "step": 10761
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.509533795396097e-06,
+      "loss": 1.7285,
+      "step": 10762
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.46993218630543e-06,
+      "loss": 1.7871,
+      "step": 10763
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.430412815752854e-06,
+      "loss": 1.5938,
+      "step": 10764
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.390975687019298e-06,
+      "loss": 1.6855,
+      "step": 10765
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.351620803379368e-06,
+      "loss": 1.5996,
+      "step": 10766
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.312348168100226e-06,
+      "loss": 1.6445,
+      "step": 10767
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.273157784442488e-06,
+      "loss": 1.4355,
+      "step": 10768
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.234049655659992e-06,
+      "loss": 1.5918,
+      "step": 10769
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.19502378499959e-06,
+      "loss": 1.6699,
+      "step": 10770
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.156080175701576e-06,
+      "loss": 1.5273,
+      "step": 10771
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.117218830999141e-06,
+      "loss": 1.5586,
+      "step": 10772
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.078439754118594e-06,
+      "loss": 1.5801,
+      "step": 10773
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.039742948279806e-06,
+      "loss": 1.6387,
+      "step": 10774
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.001128416695537e-06,
+      "loss": 1.5078,
+      "step": 10775
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.962596162571557e-06,
+      "loss": 1.5527,
+      "step": 10776
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.924146189107086e-06,
+      "loss": 1.6504,
+      "step": 10777
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.885778499494457e-06,
+      "loss": 1.5859,
+      "step": 10778
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.847493096919235e-06,
+      "loss": 1.5312,
+      "step": 10779
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.809289984559877e-06,
+      "loss": 1.6836,
+      "step": 10780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.771169165588177e-06,
+      "loss": 1.5312,
+      "step": 10781
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.733130643169274e-06,
+      "loss": 1.5059,
+      "step": 10782
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.695174420461305e-06,
+      "loss": 1.4844,
+      "step": 10783
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.657300500615418e-06,
+      "loss": 1.5645,
+      "step": 10784
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.619508886776205e-06,
+      "loss": 1.541,
+      "step": 10785
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.581799582081274e-06,
+      "loss": 1.6016,
+      "step": 10786
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.544172589661448e-06,
+      "loss": 1.666,
+      "step": 10787
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.506627912640563e-06,
+      "loss": 1.627,
+      "step": 10788
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.469165554136016e-06,
+      "loss": 1.6172,
+      "step": 10789
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.431785517257873e-06,
+      "loss": 1.5957,
+      "step": 10790
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.394487805109763e-06,
+      "loss": 1.4863,
+      "step": 10791
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.357272420788208e-06,
+      "loss": 1.5449,
+      "step": 10792
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.320139367383074e-06,
+      "loss": 1.6875,
+      "step": 10793
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.283088647977222e-06,
+      "loss": 1.543,
+      "step": 10794
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.246120265646973e-06,
+      "loss": 1.5449,
+      "step": 10795
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.209234223461538e-06,
+      "loss": 1.6797,
+      "step": 10796
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.172430524483243e-06,
+      "loss": 1.6445,
+      "step": 10797
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.135709171767869e-06,
+      "loss": 1.6074,
+      "step": 10798
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.099070168364086e-06,
+      "loss": 1.668,
+      "step": 10799
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.062513517314019e-06,
+      "loss": 1.6094,
+      "step": 10800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.02603922165246e-06,
+      "loss": 1.5488,
+      "step": 10801
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.989647284407986e-06,
+      "loss": 1.5918,
+      "step": 10802
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.953337708601848e-06,
+      "loss": 1.5625,
+      "step": 10803
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.917110497248747e-06,
+      "loss": 1.5938,
+      "step": 10804
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.880965653356386e-06,
+      "loss": 1.418,
+      "step": 10805
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.8449031799257e-06,
+      "loss": 1.5586,
+      "step": 10806
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.808923079950735e-06,
+      "loss": 1.5117,
+      "step": 10807
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.773025356418772e-06,
+      "loss": 1.5,
+      "step": 10808
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.737210012310203e-06,
+      "loss": 1.4473,
+      "step": 10809
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.701477050598537e-06,
+      "loss": 1.5488,
+      "step": 10810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.665826474250625e-06,
+      "loss": 1.4414,
+      "step": 10811
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.630258286226211e-06,
+      "loss": 1.5586,
+      "step": 10812
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.594772489478263e-06,
+      "loss": 1.5957,
+      "step": 10813
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.559369086953205e-06,
+      "loss": 1.5605,
+      "step": 10814
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.524048081590351e-06,
+      "loss": 1.709,
+      "step": 10815
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.488809476322134e-06,
+      "loss": 1.5273,
+      "step": 10816
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.453653274074213e-06,
+      "loss": 1.6621,
+      "step": 10817
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.418579477765586e-06,
+      "loss": 1.6387,
+      "step": 10818
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.383588090308035e-06,
+      "loss": 1.5176,
+      "step": 10819
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.348679114606793e-06,
+      "loss": 1.582,
+      "step": 10820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.313852553560208e-06,
+      "loss": 1.5547,
+      "step": 10821
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.279108410059743e-06,
+      "loss": 1.6562,
+      "step": 10822
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.2444466869900915e-06,
+      "loss": 1.6367,
+      "step": 10823
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.209867387228841e-06,
+      "loss": 1.6035,
+      "step": 10824
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.175370513647139e-06,
+      "loss": 1.5254,
+      "step": 10825
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.140956069108917e-06,
+      "loss": 1.5391,
+      "step": 10826
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.106624056471445e-06,
+      "loss": 1.5117,
+      "step": 10827
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.0723744785852195e-06,
+      "loss": 1.584,
+      "step": 10828
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.038207338293745e-06,
+      "loss": 1.6836,
+      "step": 10829
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.004122638433863e-06,
+      "loss": 1.6113,
+      "step": 10830
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.970120381835199e-06,
+      "loss": 1.5332,
+      "step": 10831
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.9362005713209385e-06,
+      "loss": 1.6855,
+      "step": 10832
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.902363209707385e-06,
+      "loss": 1.5586,
+      "step": 10833
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.8686082998036245e-06,
+      "loss": 1.5117,
+      "step": 10834
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.834935844412305e-06,
+      "loss": 1.6113,
+      "step": 10835
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.801345846329077e-06,
+      "loss": 1.5391,
+      "step": 10836
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.7678383083426e-06,
+      "loss": 1.4785,
+      "step": 10837
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.734413233235093e-06,
+      "loss": 1.5684,
+      "step": 10838
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.701070623781447e-06,
+      "loss": 1.4902,
+      "step": 10839
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.667810482749892e-06,
+      "loss": 1.5273,
+      "step": 10840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.634632812902109e-06,
+      "loss": 1.541,
+      "step": 10841
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.601537616992448e-06,
+      "loss": 1.6211,
+      "step": 10842
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.568524897768602e-06,
+      "loss": 1.6309,
+      "step": 10843
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.5355946579714885e-06,
+      "loss": 1.541,
+      "step": 10844
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.502746900335366e-06,
+      "loss": 1.4082,
+      "step": 10845
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.469981627587052e-06,
+      "loss": 1.6895,
+      "step": 10846
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.437298842447148e-06,
+      "loss": 1.4844,
+      "step": 10847
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.4046985476289285e-06,
+      "loss": 1.7324,
+      "step": 10848
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.372180745839118e-06,
+      "loss": 1.6777,
+      "step": 10849
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.339745439777556e-06,
+      "loss": 1.5273,
+      "step": 10850
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.3073926321370875e-06,
+      "loss": 1.627,
+      "step": 10851
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.275122325603788e-06,
+      "loss": 1.5547,
+      "step": 10852
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.242934522857069e-06,
+      "loss": 1.6426,
+      "step": 10853
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.2108292265691255e-06,
+      "loss": 1.6797,
+      "step": 10854
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.1788064394056045e-06,
+      "loss": 1.5527,
+      "step": 10855
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.146866164025044e-06,
+      "loss": 1.5391,
+      "step": 10856
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.1150084030794364e-06,
+      "loss": 1.4844,
+      "step": 10857
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.083233159213663e-06,
+      "loss": 1.5469,
+      "step": 10858
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.05154043506595e-06,
+      "loss": 1.6758,
+      "step": 10859
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.019930233267412e-06,
+      "loss": 1.5918,
+      "step": 10860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.988402556442618e-06,
+      "loss": 1.5977,
+      "step": 10861
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.956957407209251e-06,
+      "loss": 1.666,
+      "step": 10862
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.925594788177779e-06,
+      "loss": 1.5762,
+      "step": 10863
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8943147019522305e-06,
+      "loss": 1.5781,
+      "step": 10864
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8631171511296374e-06,
+      "loss": 1.6641,
+      "step": 10865
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.832002138300152e-06,
+      "loss": 1.5957,
+      "step": 10866
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.800969666046929e-06,
+      "loss": 1.4609,
+      "step": 10867
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.770019736946686e-06,
+      "loss": 1.5586,
+      "step": 10868
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.739152353568922e-06,
+      "loss": 1.5234,
+      "step": 10869
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.708367518476476e-06,
+      "loss": 1.6914,
+      "step": 10870
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.677665234225083e-06,
+      "loss": 1.6445,
+      "step": 10871
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.647045503363812e-06,
+      "loss": 1.5215,
+      "step": 10872
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.616508328434966e-06,
+      "loss": 1.6133,
+      "step": 10873
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.586053711973959e-06,
+      "loss": 1.5547,
+      "step": 10874
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.555681656509104e-06,
+      "loss": 1.5059,
+      "step": 10875
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.525392164562048e-06,
+      "loss": 1.5566,
+      "step": 10876
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4951852386476706e-06,
+      "loss": 1.6582,
+      "step": 10877
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.465060881273854e-06,
+      "loss": 1.5078,
+      "step": 10878
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.435019094941595e-06,
+      "loss": 1.5469,
+      "step": 10879
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.405059882145236e-06,
+      "loss": 1.5254,
+      "step": 10880
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.375183245372007e-06,
+      "loss": 1.5176,
+      "step": 10881
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.345389187102479e-06,
+      "loss": 1.5156,
+      "step": 10882
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3156777098103405e-06,
+      "loss": 1.6445,
+      "step": 10883
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.286048815962286e-06,
+      "loss": 1.4707,
+      "step": 10884
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.256502508018346e-06,
+      "loss": 1.5176,
+      "step": 10885
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.227038788431449e-06,
+      "loss": 1.4746,
+      "step": 10886
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.197657659647969e-06,
+      "loss": 1.4824,
+      "step": 10887
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.16835912410718e-06,
+      "loss": 1.5312,
+      "step": 10888
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.13914318424169e-06,
+      "loss": 1.5879,
+      "step": 10889
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.110009842477003e-06,
+      "loss": 1.4922,
+      "step": 10890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.080959101231963e-06,
+      "loss": 1.5391,
+      "step": 10891
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.051990962918418e-06,
+      "loss": 1.5547,
+      "step": 10892
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0231054299416655e-06,
+      "loss": 1.5801,
+      "step": 10893
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.994302504699788e-06,
+      "loss": 1.5723,
+      "step": 10894
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9655821895840945e-06,
+      "loss": 1.6328,
+      "step": 10895
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.936944486979012e-06,
+      "loss": 1.6211,
+      "step": 10896
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.908389399262525e-06,
+      "loss": 1.6855,
+      "step": 10897
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.879916928805073e-06,
+      "loss": 1.5156,
+      "step": 10898
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.851527077970652e-06,
+      "loss": 1.5586,
+      "step": 10899
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.823219849116378e-06,
+      "loss": 1.6016,
+      "step": 10900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.794995244592482e-06,
+      "loss": 1.5254,
+      "step": 10901
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7668532667422e-06,
+      "loss": 1.502,
+      "step": 10902
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.738793917902106e-06,
+      "loss": 1.4453,
+      "step": 10903
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7108172004017846e-06,
+      "loss": 1.625,
+      "step": 10904
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.682923116564042e-06,
+      "loss": 1.4434,
+      "step": 10905
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.655111668704582e-06,
+      "loss": 1.584,
+      "step": 10906
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.627382859132778e-06,
+      "loss": 1.5449,
+      "step": 10907
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.599736690150569e-06,
+      "loss": 1.6387,
+      "step": 10908
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.572173164053228e-06,
+      "loss": 1.4688,
+      "step": 10909
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.544692283129482e-06,
+      "loss": 1.4746,
+      "step": 10910
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.517294049660614e-06,
+      "loss": 1.5801,
+      "step": 10911
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.489978465921474e-06,
+      "loss": 1.5625,
+      "step": 10912
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.462745534180135e-06,
+      "loss": 1.6523,
+      "step": 10913
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4355952566972336e-06,
+      "loss": 1.6895,
+      "step": 10914
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.408527635727189e-06,
+      "loss": 1.502,
+      "step": 10915
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.381542673517203e-06,
+      "loss": 1.5566,
+      "step": 10916
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.354640372307705e-06,
+      "loss": 1.6934,
+      "step": 10917
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.327820734332244e-06,
+      "loss": 1.6191,
+      "step": 10918
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.301083761817482e-06,
+      "loss": 1.5625,
+      "step": 10919
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.274429456983309e-06,
+      "loss": 1.7012,
+      "step": 10920
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.247857822042733e-06,
+      "loss": 1.6816,
+      "step": 10921
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.221368859201657e-06,
+      "loss": 1.4609,
+      "step": 10922
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.194962570659544e-06,
+      "loss": 1.4395,
+      "step": 10923
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.168638958608639e-06,
+      "loss": 1.6445,
+      "step": 10924
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1423980252345285e-06,
+      "loss": 1.6094,
+      "step": 10925
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.116239772715802e-06,
+      "loss": 1.6094,
+      "step": 10926
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.090164203224278e-06,
+      "loss": 1.5293,
+      "step": 10927
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.064171318924781e-06,
+      "loss": 1.5312,
+      "step": 10928
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.038261121975584e-06,
+      "loss": 1.5117,
+      "step": 10929
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.012433614527633e-06,
+      "loss": 1.7168,
+      "step": 10930
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.986688798725435e-06,
+      "loss": 1.5586,
+      "step": 10931
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.961026676706392e-06,
+      "loss": 1.6016,
+      "step": 10932
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.93544725060102e-06,
+      "loss": 1.6172,
+      "step": 10933
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.909950522533068e-06,
+      "loss": 1.5703,
+      "step": 10934
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.884536494619506e-06,
+      "loss": 1.6465,
+      "step": 10935
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.859205168970204e-06,
+      "loss": 1.5957,
+      "step": 10936
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.833956547688367e-06,
+      "loss": 1.6699,
+      "step": 10937
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.808790632870207e-06,
+      "loss": 1.6895,
+      "step": 10938
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.783707426605054e-06,
+      "loss": 1.5996,
+      "step": 10939
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.758706930975575e-06,
+      "loss": 1.6113,
+      "step": 10940
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7337891480574426e-06,
+      "loss": 1.6992,
+      "step": 10941
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7089540799193356e-06,
+      "loss": 1.6035,
+      "step": 10942
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.684201728623271e-06,
+      "loss": 1.6582,
+      "step": 10943
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6595320962241607e-06,
+      "loss": 1.5605,
+      "step": 10944
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6349451847703664e-06,
+      "loss": 1.543,
+      "step": 10945
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6104409963031436e-06,
+      "loss": 1.5742,
+      "step": 10946
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5860195328568655e-06,
+      "loss": 1.5371,
+      "step": 10947
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5616807964592434e-06,
+      "loss": 1.5312,
+      "step": 10948
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.537424789130994e-06,
+      "loss": 1.5723,
+      "step": 10949
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.513251512885951e-06,
+      "loss": 1.6426,
+      "step": 10950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4891609697309536e-06,
+      "loss": 1.5234,
+      "step": 10951
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.46515316166629e-06,
+      "loss": 1.5273,
+      "step": 10952
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4412280906851444e-06,
+      "loss": 1.4492,
+      "step": 10953
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4173857587738166e-06,
+      "loss": 1.543,
+      "step": 10954
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.393626167911945e-06,
+      "loss": 1.5059,
+      "step": 10955
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3699493200720633e-06,
+      "loss": 1.5605,
+      "step": 10956
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.346355217220043e-06,
+      "loss": 1.5371,
+      "step": 10957
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3228438613146507e-06,
+      "loss": 1.584,
+      "step": 10958
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2994152543079915e-06,
+      "loss": 1.5332,
+      "step": 10959
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.276069398145176e-06,
+      "loss": 1.4941,
+      "step": 10960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.252806294764543e-06,
+      "loss": 1.5703,
+      "step": 10961
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2296259460974366e-06,
+      "loss": 1.6133,
+      "step": 10962
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2065283540684274e-06,
+      "loss": 1.5469,
+      "step": 10963
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.183513520595205e-06,
+      "loss": 1.4336,
+      "step": 10964
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.160581447588684e-06,
+      "loss": 1.5508,
+      "step": 10965
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1377321369525645e-06,
+      "loss": 1.6387,
+      "step": 10966
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.114965590583996e-06,
+      "loss": 1.6074,
+      "step": 10967
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0922818103732433e-06,
+      "loss": 1.5449,
+      "step": 10968
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.069680798203578e-06,
+      "loss": 1.5664,
+      "step": 10969
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0471625559513883e-06,
+      "loss": 1.5332,
+      "step": 10970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0247270854862896e-06,
+      "loss": 1.498,
+      "step": 10971
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.002374388671014e-06,
+      "loss": 1.6094,
+      "step": 10972
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9801044673612996e-06,
+      "loss": 1.5801,
+      "step": 10973
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.957917323406223e-06,
+      "loss": 1.5703,
+      "step": 10974
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9358129586477546e-06,
+      "loss": 1.5098,
+      "step": 10975
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.913791374921093e-06,
+      "loss": 1.4863,
+      "step": 10976
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8918525740547763e-06,
+      "loss": 1.5488,
+      "step": 10977
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.869996557870014e-06,
+      "loss": 1.584,
+      "step": 10978
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.848223328181576e-06,
+      "loss": 1.5645,
+      "step": 10979
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8265328867971283e-06,
+      "loss": 1.7285,
+      "step": 10980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8049252355174526e-06,
+      "loss": 1.4551,
+      "step": 10981
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.783400376136558e-06,
+      "loss": 1.4297,
+      "step": 10982
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7619583104415703e-06,
+      "loss": 1.6055,
+      "step": 10983
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7405990402127323e-06,
+      "loss": 1.5488,
+      "step": 10984
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7193225672232925e-06,
+      "loss": 1.6094,
+      "step": 10985
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6981288932398375e-06,
+      "loss": 1.5898,
+      "step": 10986
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6770180200218485e-06,
+      "loss": 1.6699,
+      "step": 10987
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.655989949322146e-06,
+      "loss": 1.4961,
+      "step": 10988
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6350446828865558e-06,
+      "loss": 1.5508,
+      "step": 10989
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6141822224540202e-06,
+      "loss": 1.5469,
+      "step": 10990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5934025697567087e-06,
+      "loss": 1.5996,
+      "step": 10991
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.572705726519797e-06,
+      "loss": 1.5625,
+      "step": 10992
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.552091694461689e-06,
+      "loss": 1.582,
+      "step": 10993
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5315604752937924e-06,
+      "loss": 1.4355,
+      "step": 10994
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5111120707207445e-06,
+      "loss": 1.543,
+      "step": 10995
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.490746482440187e-06,
+      "loss": 1.5312,
+      "step": 10996
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4704637121432116e-06,
+      "loss": 1.5625,
+      "step": 10997
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4502637615134716e-06,
+      "loss": 1.5391,
+      "step": 10998
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4301466322281805e-06,
+      "loss": 1.6055,
+      "step": 10999
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4101123259576696e-06,
+      "loss": 1.5391,
+      "step": 11000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3901608443651634e-06,
+      "loss": 1.6191,
+      "step": 11001
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.370292189107226e-06,
+      "loss": 1.6348,
+      "step": 11002
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3505063618333155e-06,
+      "loss": 1.4961,
+      "step": 11003
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.33080336418634e-06,
+      "loss": 1.5469,
+      "step": 11004
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3111831978018805e-06,
+      "loss": 1.5977,
+      "step": 11005
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.291645864309078e-06,
+      "loss": 1.6191,
+      "step": 11006
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2721913653299676e-06,
+      "loss": 1.6016,
+      "step": 11007
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2528197024798136e-06,
+      "loss": 1.5996,
+      "step": 11008
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.233530877366885e-06,
+      "loss": 1.5566,
+      "step": 11009
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.214324891592567e-06,
+      "loss": 1.4746,
+      "step": 11010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.195201746751474e-06,
+      "loss": 1.6895,
+      "step": 11011
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.176161444431446e-06,
+      "loss": 1.5371,
+      "step": 11012
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1572039862131077e-06,
+      "loss": 1.4844,
+      "step": 11013
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1383293736705334e-06,
+      "loss": 1.4297,
+      "step": 11014
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.119537608370581e-06,
+      "loss": 1.5742,
+      "step": 11015
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1008286918737796e-06,
+      "loss": 1.6348,
+      "step": 11016
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0822026257331094e-06,
+      "loss": 1.7227,
+      "step": 11017
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.063659411495e-06,
+      "loss": 1.6211,
+      "step": 11018
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0451990506993312e-06,
+      "loss": 1.4727,
+      "step": 11019
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.026821544878432e-06,
+      "loss": 1.6035,
+      "step": 11020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0085268955581936e-06,
+      "loss": 1.5645,
+      "step": 11021
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.990315104257623e-06,
+      "loss": 1.7246,
+      "step": 11022
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9721861724886213e-06,
+      "loss": 1.5176,
+      "step": 11023
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.954140101756319e-06,
+      "loss": 1.6113,
+      "step": 11024
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.936176893559072e-06,
+      "loss": 1.6191,
+      "step": 11025
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.918296549388354e-06,
+      "loss": 1.582,
+      "step": 11026
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9004990707285342e-06,
+      "loss": 1.5684,
+      "step": 11027
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8827844590572074e-06,
+      "loss": 1.4961,
+      "step": 11028
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8651527158453087e-06,
+      "loss": 1.4609,
+      "step": 11029
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8476038425565556e-06,
+      "loss": 1.6953,
+      "step": 11030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.830137840648005e-06,
+      "loss": 1.4336,
+      "step": 11031
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8127547115697196e-06,
+      "loss": 1.5371,
+      "step": 11032
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7954544567649889e-06,
+      "loss": 1.5645,
+      "step": 11033
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7782370776701084e-06,
+      "loss": 1.6953,
+      "step": 11034
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7611025757144906e-06,
+      "loss": 1.6816,
+      "step": 11035
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7440509523209968e-06,
+      "loss": 1.5254,
+      "step": 11036
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7270822089050509e-06,
+      "loss": 1.7676,
+      "step": 11037
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7101963468755255e-06,
+      "loss": 1.5059,
+      "step": 11038
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6933933676344105e-06,
+      "loss": 1.6758,
+      "step": 11039
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6766732725768119e-06,
+      "loss": 1.5098,
+      "step": 11040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6600360630909527e-06,
+      "loss": 1.5059,
+      "step": 11041
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6434817405579506e-06,
+      "loss": 1.6133,
+      "step": 11042
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6270103063523722e-06,
+      "loss": 1.5625,
+      "step": 11043
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6106217618416796e-06,
+      "loss": 1.5273,
+      "step": 11044
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.594316108386562e-06,
+      "loss": 1.6914,
+      "step": 11045
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5780933473407144e-06,
+      "loss": 1.5488,
+      "step": 11046
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5619534800511704e-06,
+      "loss": 1.7051,
+      "step": 11047
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.545896507857858e-06,
+      "loss": 1.584,
+      "step": 11048
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5299224320939332e-06,
+      "loss": 1.543,
+      "step": 11049
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5140312540856682e-06,
+      "loss": 1.4629,
+      "step": 11050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4982229751523413e-06,
+      "loss": 1.6953,
+      "step": 11051
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.482497596606458e-06,
+      "loss": 1.4277,
+      "step": 11052
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4668551197536406e-06,
+      "loss": 1.5996,
+      "step": 11053
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.451295545892517e-06,
+      "loss": 1.5996,
+      "step": 11054
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4358188763150536e-06,
+      "loss": 1.6973,
+      "step": 11055
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4204251123061118e-06,
+      "loss": 1.6328,
+      "step": 11056
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.40511425514378e-06,
+      "loss": 1.584,
+      "step": 11057
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.389886306099153e-06,
+      "loss": 1.5156,
+      "step": 11058
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3747412664367742e-06,
+      "loss": 1.5918,
+      "step": 11059
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3596791374137497e-06,
+      "loss": 1.4922,
+      "step": 11060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3446999202808564e-06,
+      "loss": 1.5918,
+      "step": 11061
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3298036162815442e-06,
+      "loss": 1.5723,
+      "step": 11062
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3149902266527126e-06,
+      "loss": 1.6211,
+      "step": 11063
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3002597526241555e-06,
+      "loss": 1.6094,
+      "step": 11064
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2856121954188947e-06,
+      "loss": 1.627,
+      "step": 11065
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2710475562530688e-06,
+      "loss": 1.5059,
+      "step": 11066
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2565658363359323e-06,
+      "loss": 1.6016,
+      "step": 11067
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2421670368697457e-06,
+      "loss": 1.5957,
+      "step": 11068
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2278511590499975e-06,
+      "loss": 1.5391,
+      "step": 11069
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2136182040652921e-06,
+      "loss": 1.6426,
+      "step": 11070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1994681730972402e-06,
+      "loss": 1.5566,
+      "step": 11071
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1854010673207905e-06,
+      "loss": 1.5801,
+      "step": 11072
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1714168879036758e-06,
+      "loss": 1.5508,
+      "step": 11073
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1575156360069672e-06,
+      "loss": 1.6172,
+      "step": 11074
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1436973127849638e-06,
+      "loss": 1.6016,
+      "step": 11075
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1299619193847476e-06,
+      "loss": 1.4824,
+      "step": 11076
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1163094569468513e-06,
+      "loss": 1.5293,
+      "step": 11077
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.102739926604701e-06,
+      "loss": 1.6016,
+      "step": 11078
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0892533294848405e-06,
+      "loss": 1.5605,
+      "step": 11079
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0758496667070405e-06,
+      "loss": 1.5605,
+      "step": 11080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0625289393841886e-06,
+      "loss": 1.6055,
+      "step": 11081
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0492911486221778e-06,
+      "loss": 1.5723,
+      "step": 11082
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0361362955200183e-06,
+      "loss": 1.5234,
+      "step": 11083
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.023064381170058e-06,
+      "loss": 1.6543,
+      "step": 11084
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.010075406657429e-06,
+      "loss": 1.4941,
+      "step": 11085
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.971693730606024e-07,
+      "loss": 1.6113,
+      "step": 11086
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.84346281451165e-07,
+      "loss": 1.6035,
+      "step": 11087
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.716061328935987e-07,
+      "loss": 1.502,
+      "step": 11088
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.589489284457243e-07,
+      "loss": 1.5625,
+      "step": 11089
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.463746691584785e-07,
+      "loss": 1.6738,
+      "step": 11090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.338833560756932e-07,
+      "loss": 1.7246,
+      "step": 11091
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.214749902346497e-07,
+      "loss": 1.6426,
+      "step": 11092
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.091495726653021e-07,
+      "loss": 1.5117,
+      "step": 11093
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.969071043911648e-07,
+      "loss": 1.6172,
+      "step": 11094
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.84747586428647e-07,
+      "loss": 1.6504,
+      "step": 11095
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.726710197871634e-07,
+      "loss": 1.5742,
+      "step": 11096
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.606774054693567e-07,
+      "loss": 1.5938,
+      "step": 11097
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.487667444712077e-07,
+      "loss": 1.5723,
+      "step": 11098
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.369390377814812e-07,
+      "loss": 1.5332,
+      "step": 11099
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.251942863821693e-07,
+      "loss": 1.4824,
+      "step": 11100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.1353249124827e-07,
+      "loss": 1.5215,
+      "step": 11101
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.019536533482308e-07,
+      "loss": 1.6504,
+      "step": 11102
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.904577736430607e-07,
+      "loss": 1.4648,
+      "step": 11103
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.790448530875516e-07,
+      "loss": 1.5273,
+      "step": 11104
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.677148926290567e-07,
+      "loss": 1.5566,
+      "step": 11105
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.56467893208268e-07,
+      "loss": 1.5176,
+      "step": 11106
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.453038557589941e-07,
+      "loss": 1.5371,
+      "step": 11107
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.342227812081603e-07,
+      "loss": 1.543,
+      "step": 11108
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.232246704756972e-07,
+      "loss": 1.4668,
+      "step": 11109
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.123095244747635e-07,
+      "loss": 1.5664,
+      "step": 11110
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.01477344111523e-07,
+      "loss": 1.584,
+      "step": 11111
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.907281302853674e-07,
+      "loss": 1.5566,
+      "step": 11112
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.800618838888051e-07,
+      "loss": 1.5332,
+      "step": 11113
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.694786058072388e-07,
+      "loss": 1.498,
+      "step": 11114
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.589782969195213e-07,
+      "loss": 1.7559,
+      "step": 11115
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.485609580973994e-07,
+      "loss": 1.6133,
+      "step": 11116
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.382265902056261e-07,
+      "loss": 1.7676,
+      "step": 11117
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.279751941024037e-07,
+      "loss": 1.5195,
+      "step": 11118
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.17806770638718e-07,
+      "loss": 1.5625,
+      "step": 11119
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.077213206588939e-07,
+      "loss": 1.6094,
+      "step": 11120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.977188450001503e-07,
+      "loss": 1.5332,
+      "step": 11121
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.877993444930451e-07,
+      "loss": 1.5723,
+      "step": 11122
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.77962819961142e-07,
+      "loss": 1.6445,
+      "step": 11123
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.682092722210097e-07,
+      "loss": 1.5391,
+      "step": 11124
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.585387020825561e-07,
+      "loss": 1.4434,
+      "step": 11125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.489511103486944e-07,
+      "loss": 1.5273,
+      "step": 11126
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.394464978152324e-07,
+      "loss": 1.6621,
+      "step": 11127
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.30024865271539e-07,
+      "loss": 1.5254,
+      "step": 11128
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.206862134997659e-07,
+      "loss": 1.5742,
+      "step": 11129
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.11430543275071e-07,
+      "loss": 1.5254,
+      "step": 11130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.022578553661727e-07,
+      "loss": 1.5938,
+      "step": 11131
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.931681505344621e-07,
+      "loss": 1.5801,
+      "step": 11132
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.841614295346686e-07,
+      "loss": 1.5996,
+      "step": 11133
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.752376931146385e-07,
+      "loss": 1.5957,
+      "step": 11134
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.663969420151126e-07,
+      "loss": 1.5859,
+      "step": 11135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5763917697017046e-07,
+      "loss": 1.5762,
+      "step": 11136
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.48964398706897e-07,
+      "loss": 1.541,
+      "step": 11137
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4037260794549396e-07,
+      "loss": 1.6152,
+      "step": 11138
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3186380539950165e-07,
+      "loss": 1.4668,
+      "step": 11139
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2343799177502197e-07,
+      "loss": 1.5801,
+      "step": 11140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1509516777193945e-07,
+      "loss": 1.4492,
+      "step": 11141
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0683533408258923e-07,
+      "loss": 1.6504,
+      "step": 11142
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9865849139297805e-07,
+      "loss": 1.6016,
+      "step": 11143
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9056464038200733e-07,
+      "loss": 1.6934,
+      "step": 11144
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.82553781721473e-07,
+      "loss": 1.4863,
+      "step": 11145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7462591607662077e-07,
+      "loss": 1.4727,
+      "step": 11146
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.667810441055908e-07,
+      "loss": 1.584,
+      "step": 11147
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5901916645975084e-07,
+      "loss": 1.6289,
+      "step": 11148
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.513402837834745e-07,
+      "loss": 1.6426,
+      "step": 11149
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4374439671436275e-07,
+      "loss": 1.625,
+      "step": 11150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.362315058830223e-07,
+      "loss": 1.5449,
+      "step": 11151
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.288016119131765e-07,
+      "loss": 1.418,
+      "step": 11152
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2145471542177615e-07,
+      "loss": 1.7285,
+      "step": 11153
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1419081701877796e-07,
+      "loss": 1.582,
+      "step": 11154
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.070099173071439e-07,
+      "loss": 1.5645,
+      "step": 11155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9991201688328585e-07,
+      "loss": 1.6445,
+      "step": 11156
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.928971163362881e-07,
+      "loss": 1.5078,
+      "step": 11157
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8596521624868477e-07,
+      "loss": 1.5,
+      "step": 11158
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.791163171960154e-07,
+      "loss": 1.5645,
+      "step": 11159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7235041974682515e-07,
+      "loss": 1.5566,
+      "step": 11160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6566752446277596e-07,
+      "loss": 1.5996,
+      "step": 11161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.590676318989793e-07,
+      "loss": 1.4766,
+      "step": 11162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.525507426032192e-07,
+      "loss": 1.5078,
+      "step": 11163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4611685711650733e-07,
+      "loss": 1.5625,
+      "step": 11164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3976597597308304e-07,
+      "loss": 1.4062,
+      "step": 11165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.334980997001912e-07,
+      "loss": 1.5762,
+      "step": 11166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.273132288183044e-07,
+      "loss": 1.6387,
+      "step": 11167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.212113638409008e-07,
+      "loss": 1.5488,
+      "step": 11168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.151925052744641e-07,
+      "loss": 1.5137,
+      "step": 11169
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0925665361892776e-07,
+      "loss": 1.5859,
+      "step": 11170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0340380936689773e-07,
+      "loss": 1.5918,
+      "step": 11171
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9763397300442965e-07,
+      "loss": 1.502,
+      "step": 11172
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9194714501047374e-07,
+      "loss": 1.7285,
+      "step": 11173
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8634332585731884e-07,
+      "loss": 1.5566,
+      "step": 11174
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8082251601003742e-07,
+      "loss": 1.627,
+      "step": 11175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7538471592715155e-07,
+      "loss": 1.5391,
+      "step": 11176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7002992606007795e-07,
+      "loss": 1.5664,
+      "step": 11177
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6475814685334988e-07,
+      "loss": 1.4746,
+      "step": 11178
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5956937874472832e-07,
+      "loss": 1.5137,
+      "step": 11179
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.544636221649798e-07,
+      "loss": 1.4941,
+      "step": 11180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4944087753798742e-07,
+      "loss": 1.7344,
+      "step": 11181
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4450114528075098e-07,
+      "loss": 1.5234,
+      "step": 11182
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.396444258034979e-07,
+      "loss": 1.7402,
+      "step": 11183
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3487071950935014e-07,
+      "loss": 1.5801,
+      "step": 11184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3018002679465736e-07,
+      "loss": 1.5098,
+      "step": 11185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2557234804888574e-07,
+      "loss": 1.4922,
+      "step": 11186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2104768365450712e-07,
+      "loss": 1.582,
+      "step": 11187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1660603398733205e-07,
+      "loss": 1.6387,
+      "step": 11188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1224739941606554e-07,
+      "loss": 1.5176,
+      "step": 11189
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.079717803025293e-07,
+      "loss": 1.6562,
+      "step": 11190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0377917700177264e-07,
+      "loss": 1.7188,
+      "step": 11191
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.966958986185049e-08,
+      "loss": 1.582,
+      "step": 11192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.564301922393437e-08,
+      "loss": 1.5332,
+      "step": 11193
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.169946542242347e-08,
+      "loss": 1.5996,
+      "step": 11194
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.783892878461153e-08,
+      "loss": 1.5273,
+      "step": 11195
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.406140963101993e-08,
+      "loss": 1.5527,
+      "step": 11196
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.036690827539772e-08,
+      "loss": 1.707,
+      "step": 11197
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.675542502438848e-08,
+      "loss": 1.5703,
+      "step": 11198
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.322696017797447e-08,
+      "loss": 1.4863,
+      "step": 11199
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.97815140288105e-08,
+      "loss": 1.5938,
+      "step": 11200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.641908686322307e-08,
+      "loss": 1.625,
+      "step": 11201
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.313967896032225e-08,
+      "loss": 1.5723,
+      "step": 11202
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.994329059233472e-08,
+      "loss": 1.5312,
+      "step": 11203
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.6829922024603794e-08,
+      "loss": 1.6113,
+      "step": 11204
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.379957351570042e-08,
+      "loss": 1.6602,
+      "step": 11205
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.085224531720112e-08,
+      "loss": 1.5117,
+      "step": 11206
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7987937673688034e-08,
+      "loss": 1.5254,
+      "step": 11207
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5206650823192977e-08,
+      "loss": 1.4785,
+      "step": 11208
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.250838499642029e-08,
+      "loss": 1.582,
+      "step": 11209
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.989314041752401e-08,
+      "loss": 1.3945,
+      "step": 11210
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7360917303552735e-08,
+      "loss": 1.6777,
+      "step": 11211
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.49117158647827e-08,
+      "loss": 1.6055,
+      "step": 11212
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.254553630460677e-08,
+      "loss": 1.6289,
+      "step": 11213
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.02623788194234e-08,
+      "loss": 1.582,
+      "step": 11214
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.806224359874765e-08,
+      "loss": 1.543,
+      "step": 11215
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.594513082532224e-08,
+      "loss": 1.5918,
+      "step": 11216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3911040674895468e-08,
+      "loss": 1.5449,
+      "step": 11217
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.195997331633226e-08,
+      "loss": 1.6094,
+      "step": 11218
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0091928911614156e-08,
+      "loss": 1.6074,
+      "step": 11219
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.830690761595033e-08,
+      "loss": 1.7402,
+      "step": 11220
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6604909577333517e-08,
+      "loss": 1.6074,
+      "step": 11221
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4985934937206124e-08,
+      "loss": 1.6191,
+      "step": 11222
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3449983829905143e-08,
+      "loss": 1.6582,
+      "step": 11223
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.199705638310622e-08,
+      "loss": 1.5156,
+      "step": 11224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.062715271726855e-08,
+      "loss": 1.5996,
+      "step": 11225
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.340272946189998e-09,
+      "loss": 1.6016,
+      "step": 11226
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.136417176785039e-09,
+      "loss": 1.5781,
+      "step": 11227
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.015585508862721e-09,
+      "loss": 1.5293,
+      "step": 11228
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.97777803568178e-09,
+      "loss": 1.543,
+      "step": 11229
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.022994843173478e-09,
+      "loss": 1.5859,
+      "step": 11230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.151236010718762e-09,
+      "loss": 1.5117,
+      "step": 11231
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3625016107041717e-09,
+      "loss": 1.5977,
+      "step": 11232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6567917086328663e-09,
+      "loss": 1.6582,
+      "step": 11233
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0341063630135993e-09,
+      "loss": 1.5605,
+      "step": 11234
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4944456255827632e-09,
+      "loss": 1.5879,
+      "step": 11235
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0378095411933685e-09,
+      "loss": 1.5547,
+      "step": 11236
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.641981477040204e-10,
+      "loss": 1.5352,
+      "step": 11237
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.736114762009635e-10,
+      "loss": 1.5918,
+      "step": 11238
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6604955066501503e-10,
+      "loss": 1.5078,
+      "step": 11239
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.15123885266766e-11,
+      "loss": 1.5586,
+      "step": 11240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.4893,
+      "step": 11241
+    },
+    {
+      "epoch": 1.0,
+      "step": 11241,
+      "total_flos": 9.459725673640755e+18,
+      "train_loss": 1.645774518920692,
+      "train_runtime": 28024.9955,
+      "train_samples_per_second": 25.669,
+      "train_steps_per_second": 0.401
+    }
+  ],
+  "max_steps": 11241,
+  "num_train_epochs": 1,
+  "total_flos": 9.459725673640755e+18,
+  "trial_name": null,
+  "trial_params": null
+}