diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4849 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 804,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0303030303030305e-07,
+      "loss": 0.7678,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.060606060606061e-07,
+      "loss": 0.7486,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.090909090909091e-07,
+      "loss": 0.6844,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2121212121212122e-06,
+      "loss": 0.6423,
+      "step": 4
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5151515151515152e-06,
+      "loss": 0.7242,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 0.6894,
+      "step": 6
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1212121212121216e-06,
+      "loss": 0.6789,
+      "step": 7
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.4242424242424244e-06,
+      "loss": 0.6631,
+      "step": 8
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7272727272727272e-06,
+      "loss": 0.6086,
+      "step": 9
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.0303030303030305e-06,
+      "loss": 0.6119,
+      "step": 10
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.6622,
+      "step": 11
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 0.6068,
+      "step": 12
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.93939393939394e-06,
+      "loss": 0.6927,
+      "step": 13
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.242424242424243e-06,
+      "loss": 0.6237,
+      "step": 14
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 0.5622,
+      "step": 15
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.848484848484849e-06,
+      "loss": 0.6201,
+      "step": 16
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.151515151515152e-06,
+      "loss": 0.5536,
+      "step": 17
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 0.601,
+      "step": 18
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.7575757575757586e-06,
+      "loss": 0.6038,
+      "step": 19
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.060606060606061e-06,
+      "loss": 0.5969,
+      "step": 20
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.363636363636364e-06,
+      "loss": 0.5707,
+      "step": 21
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.5728,
+      "step": 22
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.969696969696971e-06,
+      "loss": 0.585,
+      "step": 23
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.6294,
+      "step": 24
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.5757575757575764e-06,
+      "loss": 0.5554,
+      "step": 25
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.87878787878788e-06,
+      "loss": 0.5958,
+      "step": 26
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.181818181818183e-06,
+      "loss": 0.5265,
+      "step": 27
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.484848484848486e-06,
+      "loss": 0.4352,
+      "step": 28
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.787878787878788e-06,
+      "loss": 0.5238,
+      "step": 29
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 0.5666,
+      "step": 30
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.393939393939396e-06,
+      "loss": 0.6112,
+      "step": 31
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.696969696969698e-06,
+      "loss": 0.6233,
+      "step": 32
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1e-05,
+      "loss": 0.5839,
+      "step": 33
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.999958492134858e-06,
+      "loss": 0.6107,
+      "step": 34
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.999833969228587e-06,
+      "loss": 0.5078,
+      "step": 35
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.999626433348664e-06,
+      "loss": 0.5617,
+      "step": 36
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.999335887940834e-06,
+      "loss": 0.5442,
+      "step": 37
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.998962337829067e-06,
+      "loss": 0.5857,
+      "step": 38
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.998505789215469e-06,
+      "loss": 0.513,
+      "step": 39
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.997966249680185e-06,
+      "loss": 0.5383,
+      "step": 40
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.997343728181266e-06,
+      "loss": 0.5579,
+      "step": 41
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.996638235054527e-06,
+      "loss": 0.5556,
+      "step": 42
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.995849782013378e-06,
+      "loss": 0.5051,
+      "step": 43
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.994978382148616e-06,
+      "loss": 0.6008,
+      "step": 44
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.994024049928222e-06,
+      "loss": 0.5312,
+      "step": 45
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.99298680119711e-06,
+      "loss": 0.5143,
+      "step": 46
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.991866653176877e-06,
+      "loss": 0.4882,
+      "step": 47
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.990663624465504e-06,
+      "loss": 0.5685,
+      "step": 48
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.989377735037047e-06,
+      "loss": 0.486,
+      "step": 49
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.988009006241321e-06,
+      "loss": 0.5476,
+      "step": 50
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.986557460803527e-06,
+      "loss": 0.5981,
+      "step": 51
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.985023122823889e-06,
+      "loss": 0.5334,
+      "step": 52
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.983406017777244e-06,
+      "loss": 0.554,
+      "step": 53
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.98170617251262e-06,
+      "loss": 0.4702,
+      "step": 54
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.9799236152528e-06,
+      "loss": 0.5905,
+      "step": 55
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.97805837559384e-06,
+      "loss": 0.5348,
+      "step": 56
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.976110484504587e-06,
+      "loss": 0.4669,
+      "step": 57
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.974079974326162e-06,
+      "loss": 0.5359,
+      "step": 58
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.971966878771423e-06,
+      "loss": 0.6028,
+      "step": 59
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.969771232924404e-06,
+      "loss": 0.4909,
+      "step": 60
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.967493073239729e-06,
+      "loss": 0.5464,
+      "step": 61
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.965132437542023e-06,
+      "loss": 0.5131,
+      "step": 62
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.962689365025259e-06,
+      "loss": 0.5379,
+      "step": 63
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.960163896252131e-06,
+      "loss": 0.4797,
+      "step": 64
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.957556073153364e-06,
+      "loss": 0.5163,
+      "step": 65
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.954865939027028e-06,
+      "loss": 0.5366,
+      "step": 66
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.95209353853781e-06,
+      "loss": 0.5051,
+      "step": 67
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.949238917716283e-06,
+      "loss": 0.471,
+      "step": 68
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.94630212395813e-06,
+      "loss": 0.5505,
+      "step": 69
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.94328320602337e-06,
+      "loss": 0.5167,
+      "step": 70
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.940182214035537e-06,
+      "loss": 0.4822,
+      "step": 71
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.936999199480854e-06,
+      "loss": 0.4917,
+      "step": 72
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.933734215207376e-06,
+      "loss": 0.4735,
+      "step": 73
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.930387315424114e-06,
+      "loss": 0.523,
+      "step": 74
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.926958555700134e-06,
+      "loss": 0.445,
+      "step": 75
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.923447992963635e-06,
+      "loss": 0.5564,
+      "step": 76
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.919855685501003e-06,
+      "loss": 0.4953,
+      "step": 77
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.916181692955841e-06,
+      "loss": 0.4794,
+      "step": 78
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.91242607632799e-06,
+      "loss": 0.4279,
+      "step": 79
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.908588897972495e-06,
+      "loss": 0.4863,
+      "step": 80
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.90467022159859e-06,
+      "loss": 0.5865,
+      "step": 81
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.900670112268633e-06,
+      "loss": 0.482,
+      "step": 82
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.896588636397022e-06,
+      "loss": 0.4903,
+      "step": 83
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.8924258617491e-06,
+      "loss": 0.5028,
+      "step": 84
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.888181857440019e-06,
+      "loss": 0.5701,
+      "step": 85
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.883856693933602e-06,
+      "loss": 0.5387,
+      "step": 86
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.879450443041172e-06,
+      "loss": 0.5311,
+      "step": 87
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.874963177920358e-06,
+      "loss": 0.526,
+      "step": 88
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.870394973073874e-06,
+      "loss": 0.468,
+      "step": 89
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.865745904348296e-06,
+      "loss": 0.4907,
+      "step": 90
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.86101604893279e-06,
+      "loss": 0.4735,
+      "step": 91
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.856205485357837e-06,
+      "loss": 0.5558,
+      "step": 92
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.851314293493923e-06,
+      "loss": 0.477,
+      "step": 93
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.846342554550225e-06,
+      "loss": 0.5064,
+      "step": 94
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.841290351073248e-06,
+      "loss": 0.5061,
+      "step": 95
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.836157766945467e-06,
+      "loss": 0.4644,
+      "step": 96
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.830944887383925e-06,
+      "loss": 0.4992,
+      "step": 97
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.82565179893882e-06,
+      "loss": 0.4907,
+      "step": 98
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.820278589492076e-06,
+      "loss": 0.5114,
+      "step": 99
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.814825348255875e-06,
+      "loss": 0.4487,
+      "step": 100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.809292165771172e-06,
+      "loss": 0.5144,
+      "step": 101
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.80367913390621e-06,
+      "loss": 0.4406,
+      "step": 102
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.797986345854976e-06,
+      "loss": 0.4793,
+      "step": 103
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.792213896135659e-06,
+      "loss": 0.4944,
+      "step": 104
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.786361880589084e-06,
+      "loss": 0.4827,
+      "step": 105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.780430396377124e-06,
+      "loss": 0.5,
+      "step": 106
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.774419541981075e-06,
+      "loss": 0.5129,
+      "step": 107
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.768329417200029e-06,
+      "loss": 0.4435,
+      "step": 108
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.76216012314922e-06,
+      "loss": 0.528,
+      "step": 109
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.755911762258338e-06,
+      "loss": 0.5105,
+      "step": 110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.749584438269833e-06,
+      "loss": 0.4923,
+      "step": 111
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.743178256237185e-06,
+      "loss": 0.5192,
+      "step": 112
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.736693322523172e-06,
+      "loss": 0.4617,
+      "step": 113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.730129744798096e-06,
+      "loss": 0.488,
+      "step": 114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.723487632037996e-06,
+      "loss": 0.4696,
+      "step": 115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.71676709452284e-06,
+      "loss": 0.4897,
+      "step": 116
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.709968243834698e-06,
+      "loss": 0.451,
+      "step": 117
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.703091192855876e-06,
+      "loss": 0.4831,
+      "step": 118
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.696136055767055e-06,
+      "loss": 0.49,
+      "step": 119
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.689102948045398e-06,
+      "loss": 0.5338,
+      "step": 120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.681991986462616e-06,
+      "loss": 0.4177,
+      "step": 121
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.67480328908304e-06,
+      "loss": 0.5144,
+      "step": 122
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.667536975261667e-06,
+      "loss": 0.4238,
+      "step": 123
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.660193165642167e-06,
+      "loss": 0.4411,
+      "step": 124
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.652771982154878e-06,
+      "loss": 0.5001,
+      "step": 125
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.6452735480148e-06,
+      "loss": 0.4207,
+      "step": 126
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.637697987719527e-06,
+      "loss": 0.4681,
+      "step": 127
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.630045427047194e-06,
+      "loss": 0.4679,
+      "step": 128
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.622315993054384e-06,
+      "loss": 0.4971,
+      "step": 129
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.614509814074015e-06,
+      "loss": 0.5188,
+      "step": 130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.60662701971322e-06,
+      "loss": 0.4483,
+      "step": 131
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.598667740851187e-06,
+      "loss": 0.5159,
+      "step": 132
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.590632109636979e-06,
+      "loss": 0.5182,
+      "step": 133
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.58252025948736e-06,
+      "loss": 0.4488,
+      "step": 134
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.574332325084564e-06,
+      "loss": 0.478,
+      "step": 135
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.566068442374057e-06,
+      "loss": 0.4757,
+      "step": 136
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.557728748562294e-06,
+      "loss": 0.5225,
+      "step": 137
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.549313382114427e-06,
+      "loss": 0.4631,
+      "step": 138
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.540822482752017e-06,
+      "loss": 0.4413,
+      "step": 139
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.532256191450705e-06,
+      "loss": 0.4665,
+      "step": 140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.523614650437876e-06,
+      "loss": 0.4692,
+      "step": 141
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.514898003190298e-06,
+      "loss": 0.4044,
+      "step": 142
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.506106394431737e-06,
+      "loss": 0.4428,
+      "step": 143
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.497239970130561e-06,
+      "loss": 0.4821,
+      "step": 144
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.488298877497307e-06,
+      "loss": 0.4566,
+      "step": 145
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.479283264982237e-06,
+      "loss": 0.4868,
+      "step": 146
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.470193282272886e-06,
+      "loss": 0.4678,
+      "step": 147
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.461029080291567e-06,
+      "loss": 0.5092,
+      "step": 148
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.45179081119286e-06,
+      "loss": 0.488,
+      "step": 149
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.442478628361098e-06,
+      "loss": 0.5545,
+      "step": 150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.43309268640781e-06,
+      "loss": 0.4638,
+      "step": 151
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.423633141169167e-06,
+      "loss": 0.4493,
+      "step": 152
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.414100149703373e-06,
+      "loss": 0.4748,
+      "step": 153
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.404493870288083e-06,
+      "loss": 0.5025,
+      "step": 154
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.394814462417756e-06,
+      "loss": 0.4529,
+      "step": 155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.385062086801013e-06,
+      "loss": 0.4367,
+      "step": 156
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.375236905357972e-06,
+      "loss": 0.4705,
+      "step": 157
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.365339081217556e-06,
+      "loss": 0.3987,
+      "step": 158
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.355368778714784e-06,
+      "loss": 0.5246,
+      "step": 159
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.345326163388044e-06,
+      "loss": 0.5063,
+      "step": 160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.335211401976348e-06,
+      "loss": 0.4336,
+      "step": 161
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.325024662416553e-06,
+      "loss": 0.4375,
+      "step": 162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.314766113840587e-06,
+      "loss": 0.4676,
+      "step": 163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.304435926572628e-06,
+      "loss": 0.4486,
+      "step": 164
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.294034272126286e-06,
+      "loss": 0.4672,
+      "step": 165
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.28356132320175e-06,
+      "loss": 0.4387,
+      "step": 166
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.273017253682916e-06,
+      "loss": 0.5255,
+      "step": 167
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.262402238634514e-06,
+      "loss": 0.4563,
+      "step": 168
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.25171645429919e-06,
+      "loss": 0.4088,
+      "step": 169
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.240960078094578e-06,
+      "loss": 0.4088,
+      "step": 170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.230133288610366e-06,
+      "loss": 0.3959,
+      "step": 171
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.219236265605322e-06,
+      "loss": 0.4247,
+      "step": 172
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.208269190004308e-06,
+      "loss": 0.4845,
+      "step": 173
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.197232243895285e-06,
+      "loss": 0.4452,
+      "step": 174
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.186125610526278e-06,
+      "loss": 0.435,
+      "step": 175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.174949474302343e-06,
+      "loss": 0.4192,
+      "step": 176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.163704020782507e-06,
+      "loss": 0.4277,
+      "step": 177
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.152389436676673e-06,
+      "loss": 0.4671,
+      "step": 178
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.141005909842532e-06,
+      "loss": 0.4434,
+      "step": 179
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.129553629282448e-06,
+      "loss": 0.4359,
+      "step": 180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.118032785140305e-06,
+      "loss": 0.4627,
+      "step": 181
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.10644356869836e-06,
+      "loss": 0.422,
+      "step": 182
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.094786172374066e-06,
+      "loss": 0.5088,
+      "step": 183
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.08306078971688e-06,
+      "loss": 0.5419,
+      "step": 184
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.071267615405039e-06,
+      "loss": 0.4761,
+      "step": 185
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.059406845242343e-06,
+      "loss": 0.4877,
+      "step": 186
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.047478676154886e-06,
+      "loss": 0.5063,
+      "step": 187
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.035483306187806e-06,
+      "loss": 0.4383,
+      "step": 188
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.023420934501981e-06,
+      "loss": 0.4909,
+      "step": 189
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.011291761370728e-06,
+      "loss": 0.4463,
+      "step": 190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.999095988176485e-06,
+      "loss": 0.3915,
+      "step": 191
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.98683381740745e-06,
+      "loss": 0.45,
+      "step": 192
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.97450545265424e-06,
+      "loss": 0.4835,
+      "step": 193
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.962111098606492e-06,
+      "loss": 0.4662,
+      "step": 194
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.949650961049479e-06,
+      "loss": 0.4738,
+      "step": 195
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.937125246860681e-06,
+      "loss": 0.4222,
+      "step": 196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.924534164006367e-06,
+      "loss": 0.4193,
+      "step": 197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.911877921538117e-06,
+      "loss": 0.4405,
+      "step": 198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.899156729589377e-06,
+      "loss": 0.4691,
+      "step": 199
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.886370799371957e-06,
+      "loss": 0.434,
+      "step": 200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.87352034317252e-06,
+      "loss": 0.4156,
+      "step": 201
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.86060557434907e-06,
+      "loss": 0.4075,
+      "step": 202
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.847626707327398e-06,
+      "loss": 0.4729,
+      "step": 203
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.83458395759753e-06,
+      "loss": 0.4216,
+      "step": 204
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.821477541710144e-06,
+      "loss": 0.498,
+      "step": 205
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.808307677272978e-06,
+      "loss": 0.4679,
+      "step": 206
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.795074582947214e-06,
+      "loss": 0.467,
+      "step": 207
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.781778478443852e-06,
+      "loss": 0.476,
+      "step": 208
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.768419584520053e-06,
+      "loss": 0.4929,
+      "step": 209
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.754998122975489e-06,
+      "loss": 0.4823,
+      "step": 210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.741514316648643e-06,
+      "loss": 0.4594,
+      "step": 211
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.72796838941312e-06,
+      "loss": 0.4843,
+      "step": 212
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.714360566173932e-06,
+      "loss": 0.4452,
+      "step": 213
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.700691072863754e-06,
+      "loss": 0.3934,
+      "step": 214
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.686960136439178e-06,
+      "loss": 0.4612,
+      "step": 215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.67316798487695e-06,
+      "loss": 0.4554,
+      "step": 216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.659314847170175e-06,
+      "loss": 0.4372,
+      "step": 217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.64540095332452e-06,
+      "loss": 0.4316,
+      "step": 218
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.631426534354404e-06,
+      "loss": 0.5208,
+      "step": 219
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.617391822279138e-06,
+      "loss": 0.4302,
+      "step": 220
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.6032970501191e-06,
+      "loss": 0.4084,
+      "step": 221
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.589142451891849e-06,
+      "loss": 0.4455,
+      "step": 222
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.57492826260825e-06,
+      "loss": 0.4318,
+      "step": 223
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.560654718268562e-06,
+      "loss": 0.4357,
+      "step": 224
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.546322055858526e-06,
+      "loss": 0.4688,
+      "step": 225
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.531930513345432e-06,
+      "loss": 0.3941,
+      "step": 226
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.517480329674157e-06,
+      "loss": 0.4475,
+      "step": 227
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.502971744763216e-06,
+      "loss": 0.4418,
+      "step": 228
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.48840499950076e-06,
+      "loss": 0.4064,
+      "step": 229
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.473780335740591e-06,
+      "loss": 0.3988,
+      "step": 230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.459097996298137e-06,
+      "loss": 0.4409,
+      "step": 231
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.444358224946421e-06,
+      "loss": 0.4039,
+      "step": 232
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.429561266412024e-06,
+      "loss": 0.4393,
+      "step": 233
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.414707366371006e-06,
+      "loss": 0.3782,
+      "step": 234
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.399796771444843e-06,
+      "loss": 0.4641,
+      "step": 235
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.384829729196318e-06,
+      "loss": 0.4058,
+      "step": 236
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.369806488125418e-06,
+      "loss": 0.4258,
+      "step": 237
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.35472729766521e-06,
+      "loss": 0.4966,
+      "step": 238
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.339592408177697e-06,
+      "loss": 0.4681,
+      "step": 239
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.324402070949658e-06,
+      "loss": 0.4131,
+      "step": 240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.309156538188483e-06,
+      "loss": 0.3929,
+      "step": 241
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.293856063017976e-06,
+      "loss": 0.477,
+      "step": 242
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.278500899474162e-06,
+      "loss": 0.4237,
+      "step": 243
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.263091302501065e-06,
+      "loss": 0.415,
+      "step": 244
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.247627527946474e-06,
+      "loss": 0.4129,
+      "step": 245
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.232109832557696e-06,
+      "loss": 0.4101,
+      "step": 246
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.216538473977297e-06,
+      "loss": 0.4864,
+      "step": 247
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.200913710738812e-06,
+      "loss": 0.4467,
+      "step": 248
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.18523580226247e-06,
+      "loss": 0.4281,
+      "step": 249
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.169505008850878e-06,
+      "loss": 0.4897,
+      "step": 250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.153721591684691e-06,
+      "loss": 0.4859,
+      "step": 251
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.137885812818296e-06,
+      "loss": 0.4346,
+      "step": 252
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.121997935175438e-06,
+      "loss": 0.4544,
+      "step": 253
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.10605822254487e-06,
+      "loss": 0.4578,
+      "step": 254
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.090066939575972e-06,
+      "loss": 0.4292,
+      "step": 255
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.074024351774348e-06,
+      "loss": 0.46,
+      "step": 256
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.057930725497428e-06,
+      "loss": 0.3906,
+      "step": 257
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.041786327950037e-06,
+      "loss": 0.5179,
+      "step": 258
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.025591427179968e-06,
+      "loss": 0.4168,
+      "step": 259
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.009346292073525e-06,
+      "loss": 0.4428,
+      "step": 260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.993051192351056e-06,
+      "loss": 0.4656,
+      "step": 261
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.976706398562484e-06,
+      "loss": 0.4188,
+      "step": 262
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.960312182082805e-06,
+      "loss": 0.4323,
+      "step": 263
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.943868815107594e-06,
+      "loss": 0.3912,
+      "step": 264
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.927376570648469e-06,
+      "loss": 0.411,
+      "step": 265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.910835722528579e-06,
+      "loss": 0.4037,
+      "step": 266
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.894246545378037e-06,
+      "loss": 0.4195,
+      "step": 267
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.877609314629374e-06,
+      "loss": 0.3578,
+      "step": 268
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.860924306512965e-06,
+      "loss": 0.3486,
+      "step": 269
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.844191798052438e-06,
+      "loss": 0.3801,
+      "step": 270
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.82741206706007e-06,
+      "loss": 0.3391,
+      "step": 271
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.81058539213219e-06,
+      "loss": 0.3462,
+      "step": 272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.793712052644535e-06,
+      "loss": 0.3341,
+      "step": 273
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.776792328747628e-06,
+      "loss": 0.2972,
+      "step": 274
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.759826501362118e-06,
+      "loss": 0.3077,
+      "step": 275
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.742814852174112e-06,
+      "loss": 0.2975,
+      "step": 276
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.725757663630506e-06,
+      "loss": 0.3341,
+      "step": 277
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.708655218934293e-06,
+      "loss": 0.3267,
+      "step": 278
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.691507802039861e-06,
+      "loss": 0.3359,
+      "step": 279
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.674315697648277e-06,
+      "loss": 0.3156,
+      "step": 280
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.65707919120256e-06,
+      "loss": 0.3544,
+      "step": 281
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.639798568882947e-06,
+      "loss": 0.3171,
+      "step": 282
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.622474117602132e-06,
+      "loss": 0.2959,
+      "step": 283
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.605106125000512e-06,
+      "loss": 0.3667,
+      "step": 284
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.5876948794414015e-06,
+      "loss": 0.3609,
+      "step": 285
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.570240670006257e-06,
+      "loss": 0.324,
+      "step": 286
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.5527437864898645e-06,
+      "loss": 0.3275,
+      "step": 287
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.535204519395538e-06,
+      "loss": 0.35,
+      "step": 288
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.517623159930289e-06,
+      "loss": 0.3219,
+      "step": 289
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.3012,
+      "step": 290
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.482335332204568e-06,
+      "loss": 0.3422,
+      "step": 291
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.464629449833051e-06,
+      "loss": 0.3246,
+      "step": 292
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.4468826468588e-06,
+      "loss": 0.3232,
+      "step": 293
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.429095217934578e-06,
+      "loss": 0.2921,
+      "step": 294
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.411267458387665e-06,
+      "loss": 0.347,
+      "step": 295
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.393399664214957e-06,
+      "loss": 0.3213,
+      "step": 296
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.375492132078051e-06,
+      "loss": 0.3438,
+      "step": 297
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.3575451592983185e-06,
+      "loss": 0.3568,
+      "step": 298
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.339559043851967e-06,
+      "loss": 0.3194,
+      "step": 299
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.321534084365101e-06,
+      "loss": 0.3396,
+      "step": 300
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.303470580108756e-06,
+      "loss": 0.3043,
+      "step": 301
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.285368830993931e-06,
+      "loss": 0.3391,
+      "step": 302
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.267229137566607e-06,
+      "loss": 0.2639,
+      "step": 303
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.249051801002769e-06,
+      "loss": 0.2813,
+      "step": 304
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.230837123103385e-06,
+      "loss": 0.2928,
+      "step": 305
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.2125854062894184e-06,
+      "loss": 0.3474,
+      "step": 306
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.194296953596786e-06,
+      "loss": 0.3425,
+      "step": 307
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.17597206867134e-06,
+      "loss": 0.2928,
+      "step": 308
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.15761105576382e-06,
+      "loss": 0.3534,
+      "step": 309
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.139214219724806e-06,
+      "loss": 0.3529,
+      "step": 310
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.120781865999655e-06,
+      "loss": 0.3136,
+      "step": 311
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.102314300623425e-06,
+      "loss": 0.3459,
+      "step": 312
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.083811830215805e-06,
+      "loss": 0.2742,
+      "step": 313
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.065274761976011e-06,
+      "loss": 0.3225,
+      "step": 314
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.0467034036776945e-06,
+      "loss": 0.3696,
+      "step": 315
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.028098063663831e-06,
+      "loss": 0.3225,
+      "step": 316
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.009459050841596e-06,
+      "loss": 0.285,
+      "step": 317
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.990786674677246e-06,
+      "loss": 0.3588,
+      "step": 318
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.9720812451909654e-06,
+      "loss": 0.3713,
+      "step": 319
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.953343072951733e-06,
+      "loss": 0.3379,
+      "step": 320
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.934572469072163e-06,
+      "loss": 0.3538,
+      "step": 321
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.915769745203329e-06,
+      "loss": 0.2825,
+      "step": 322
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.896935213529602e-06,
+      "loss": 0.3121,
+      "step": 323
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.878069186763466e-06,
+      "loss": 0.3222,
+      "step": 324
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.859171978140312e-06,
+      "loss": 0.3422,
+      "step": 325
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.8402439014132615e-06,
+      "loss": 0.3003,
+      "step": 326
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.821285270847934e-06,
+      "loss": 0.3473,
+      "step": 327
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.802296401217243e-06,
+      "loss": 0.3388,
+      "step": 328
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.7832776077961615e-06,
+      "loss": 0.324,
+      "step": 329
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.764229206356498e-06,
+      "loss": 0.3472,
+      "step": 330
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.745151513161644e-06,
+      "loss": 0.3219,
+      "step": 331
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.726044844961323e-06,
+      "loss": 0.351,
+      "step": 332
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.706909518986341e-06,
+      "loss": 0.3953,
+      "step": 333
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.687745852943307e-06,
+      "loss": 0.3348,
+      "step": 334
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.668554165009369e-06,
+      "loss": 0.3859,
+      "step": 335
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.649334773826924e-06,
+      "loss": 0.3204,
+      "step": 336
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.630087998498333e-06,
+      "loss": 0.3357,
+      "step": 337
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.610814158580615e-06,
+      "loss": 0.3526,
+      "step": 338
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.591513574080152e-06,
+      "loss": 0.307,
+      "step": 339
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.5721865654473645e-06,
+      "loss": 0.2805,
+      "step": 340
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.552833453571402e-06,
+      "loss": 0.306,
+      "step": 341
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.5334545597748075e-06,
+      "loss": 0.3353,
+      "step": 342
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.514050205808184e-06,
+      "loss": 0.2939,
+      "step": 343
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.494620713844856e-06,
+      "loss": 0.3369,
+      "step": 344
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.475166406475515e-06,
+      "loss": 0.3421,
+      "step": 345
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.455687606702869e-06,
+      "loss": 0.287,
+      "step": 346
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.436184637936273e-06,
+      "loss": 0.3526,
+      "step": 347
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.41665782398637e-06,
+      "loss": 0.3264,
+      "step": 348
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.397107489059699e-06,
+      "loss": 0.288,
+      "step": 349
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.377533957753332e-06,
+      "loss": 0.366,
+      "step": 350
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.357937555049465e-06,
+      "loss": 0.3095,
+      "step": 351
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.338318606310034e-06,
+      "loss": 0.3512,
+      "step": 352
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.31867743727131e-06,
+      "loss": 0.293,
+      "step": 353
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.299014374038493e-06,
+      "loss": 0.3398,
+      "step": 354
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.279329743080295e-06,
+      "loss": 0.3113,
+      "step": 355
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.259623871223514e-06,
+      "loss": 0.3712,
+      "step": 356
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.239897085647624e-06,
+      "loss": 0.3284,
+      "step": 357
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.220149713879323e-06,
+      "loss": 0.2893,
+      "step": 358
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.2003820837871124e-06,
+      "loss": 0.3448,
+      "step": 359
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.180594523575838e-06,
+      "loss": 0.31,
+      "step": 360
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.1607873617812555e-06,
+      "loss": 0.3668,
+      "step": 361
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.1409609272645635e-06,
+      "loss": 0.3411,
+      "step": 362
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.12111554920695e-06,
+      "loss": 0.3761,
+      "step": 363
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.101251557104126e-06,
+      "loss": 0.3403,
+      "step": 364
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.081369280760852e-06,
+      "loss": 0.3567,
+      "step": 365
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.061469050285469e-06,
+      "loss": 0.3355,
+      "step": 366
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.041551196084407e-06,
+      "loss": 0.2983,
+      "step": 367
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.021616048856712e-06,
+      "loss": 0.3589,
+      "step": 368
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.0016639395885424e-06,
+      "loss": 0.3866,
+      "step": 369
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.981695199547684e-06,
+      "loss": 0.3436,
+      "step": 370
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.961710160278042e-06,
+      "loss": 0.3069,
+      "step": 371
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.941709153594146e-06,
+      "loss": 0.3087,
+      "step": 372
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.9216925115756275e-06,
+      "loss": 0.356,
+      "step": 373
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.901660566561721e-06,
+      "loss": 0.3378,
+      "step": 374
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.881613651145732e-06,
+      "loss": 0.3122,
+      "step": 375
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.861552098169527e-06,
+      "loss": 0.3133,
+      "step": 376
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.841476240718e-06,
+      "loss": 0.3335,
+      "step": 377
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.821386412113546e-06,
+      "loss": 0.2725,
+      "step": 378
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.80128294591052e-06,
+      "loss": 0.2854,
+      "step": 379
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.781166175889711e-06,
+      "loss": 0.3153,
+      "step": 380
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.761036436052788e-06,
+      "loss": 0.3268,
+      "step": 381
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.7408940606167615e-06,
+      "loss": 0.3489,
+      "step": 382
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.7207393840084355e-06,
+      "loss": 0.3587,
+      "step": 383
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.700572740858847e-06,
+      "loss": 0.3006,
+      "step": 384
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.680394465997717e-06,
+      "loss": 0.3529,
+      "step": 385
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.660204894447893e-06,
+      "loss": 0.3665,
+      "step": 386
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.640004361419776e-06,
+      "loss": 0.3287,
+      "step": 387
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.619793202305771e-06,
+      "loss": 0.3399,
+      "step": 388
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5995717526747004e-06,
+      "loss": 0.2815,
+      "step": 389
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.579340348266251e-06,
+      "loss": 0.2901,
+      "step": 390
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.559099324985381e-06,
+      "loss": 0.3532,
+      "step": 391
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.538849018896758e-06,
+      "loss": 0.2894,
+      "step": 392
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.518589766219173e-06,
+      "loss": 0.3437,
+      "step": 393
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.498321903319952e-06,
+      "loss": 0.3078,
+      "step": 394
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.4780457667093885e-06,
+      "loss": 0.3242,
+      "step": 395
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.457761693035139e-06,
+      "loss": 0.3112,
+      "step": 396
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.437470019076639e-06,
+      "loss": 0.3443,
+      "step": 397
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.417171081739517e-06,
+      "loss": 0.3604,
+      "step": 398
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.396865218049995e-06,
+      "loss": 0.2279,
+      "step": 399
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.376552765149291e-06,
+      "loss": 0.2948,
+      "step": 400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.356234060288029e-06,
+      "loss": 0.3455,
+      "step": 401
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.335909440820635e-06,
+      "loss": 0.3068,
+      "step": 402
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.315579244199731e-06,
+      "loss": 0.2834,
+      "step": 403
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.295243807970544e-06,
+      "loss": 0.3372,
+      "step": 404
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.27490346976529e-06,
+      "loss": 0.2951,
+      "step": 405
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.254558567297576e-06,
+      "loss": 0.2838,
+      "step": 406
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.23420943835679e-06,
+      "loss": 0.3197,
+      "step": 407
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.21385642080249e-06,
+      "loss": 0.3398,
+      "step": 408
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.1934998525588e-06,
+      "loss": 0.3328,
+      "step": 409
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.173140071608796e-06,
+      "loss": 0.3318,
+      "step": 410
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.152777415988894e-06,
+      "loss": 0.3095,
+      "step": 411
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.132412223783242e-06,
+      "loss": 0.3155,
+      "step": 412
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.112044833118097e-06,
+      "loss": 0.3047,
+      "step": 413
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.091675582156224e-06,
+      "loss": 0.2845,
+      "step": 414
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.0713048090912695e-06,
+      "loss": 0.3433,
+      "step": 415
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.050932852142155e-06,
+      "loss": 0.2861,
+      "step": 416
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.0305600495474586e-06,
+      "loss": 0.3259,
+      "step": 417
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.010186739559795e-06,
+      "loss": 0.3052,
+      "step": 418
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.989813260440207e-06,
+      "loss": 0.3061,
+      "step": 419
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.969439950452543e-06,
+      "loss": 0.3659,
+      "step": 420
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.949067147857846e-06,
+      "loss": 0.3362,
+      "step": 421
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.928695190908731e-06,
+      "loss": 0.2483,
+      "step": 422
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.908324417843779e-06,
+      "loss": 0.3317,
+      "step": 423
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.887955166881904e-06,
+      "loss": 0.305,
+      "step": 424
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.86758777621676e-06,
+      "loss": 0.3278,
+      "step": 425
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.847222584011107e-06,
+      "loss": 0.3219,
+      "step": 426
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.826859928391206e-06,
+      "loss": 0.3153,
+      "step": 427
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.806500147441203e-06,
+      "loss": 0.3681,
+      "step": 428
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.7861435791975124e-06,
+      "loss": 0.3168,
+      "step": 429
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.765790561643213e-06,
+      "loss": 0.3393,
+      "step": 430
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.745441432702425e-06,
+      "loss": 0.2416,
+      "step": 431
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.72509653023471e-06,
+      "loss": 0.3022,
+      "step": 432
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.704756192029458e-06,
+      "loss": 0.3132,
+      "step": 433
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.6844207558002695e-06,
+      "loss": 0.3151,
+      "step": 434
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.664090559179367e-06,
+      "loss": 0.3262,
+      "step": 435
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.6437659397119715e-06,
+      "loss": 0.31,
+      "step": 436
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.623447234850711e-06,
+      "loss": 0.324,
+      "step": 437
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.603134781950007e-06,
+      "loss": 0.3346,
+      "step": 438
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.582828918260484e-06,
+      "loss": 0.3069,
+      "step": 439
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.562529980923362e-06,
+      "loss": 0.3122,
+      "step": 440
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.542238306964863e-06,
+      "loss": 0.3101,
+      "step": 441
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.521954233290612e-06,
+      "loss": 0.2809,
+      "step": 442
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.501678096680049e-06,
+      "loss": 0.3464,
+      "step": 443
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.48141023378083e-06,
+      "loss": 0.311,
+      "step": 444
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.461150981103244e-06,
+      "loss": 0.2865,
+      "step": 445
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.440900675014621e-06,
+      "loss": 0.3592,
+      "step": 446
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.420659651733751e-06,
+      "loss": 0.3151,
+      "step": 447
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4004282473253e-06,
+      "loss": 0.3091,
+      "step": 448
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.380206797694232e-06,
+      "loss": 0.2809,
+      "step": 449
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.359995638580226e-06,
+      "loss": 0.2801,
+      "step": 450
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.33979510555211e-06,
+      "loss": 0.3252,
+      "step": 451
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.319605534002285e-06,
+      "loss": 0.3299,
+      "step": 452
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.299427259141155e-06,
+      "loss": 0.3626,
+      "step": 453
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.279260615991567e-06,
+      "loss": 0.3322,
+      "step": 454
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.259105939383238e-06,
+      "loss": 0.3144,
+      "step": 455
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.238963563947212e-06,
+      "loss": 0.4156,
+      "step": 456
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.21883382411029e-06,
+      "loss": 0.2972,
+      "step": 457
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.19871705408948e-06,
+      "loss": 0.284,
+      "step": 458
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.178613587886455e-06,
+      "loss": 0.3211,
+      "step": 459
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.158523759282e-06,
+      "loss": 0.2976,
+      "step": 460
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1384479018304735e-06,
+      "loss": 0.2577,
+      "step": 461
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1183863488542686e-06,
+      "loss": 0.2976,
+      "step": 462
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.098339433438281e-06,
+      "loss": 0.2663,
+      "step": 463
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.078307488424374e-06,
+      "loss": 0.2594,
+      "step": 464
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.058290846405856e-06,
+      "loss": 0.3512,
+      "step": 465
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.038289839721959e-06,
+      "loss": 0.3166,
+      "step": 466
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.018304800452318e-06,
+      "loss": 0.296,
+      "step": 467
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.998336060411459e-06,
+      "loss": 0.2829,
+      "step": 468
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.978383951143289e-06,
+      "loss": 0.3076,
+      "step": 469
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.9584488039155945e-06,
+      "loss": 0.3318,
+      "step": 470
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.938530949714533e-06,
+      "loss": 0.315,
+      "step": 471
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.91863071923915e-06,
+      "loss": 0.2838,
+      "step": 472
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.898748442895877e-06,
+      "loss": 0.2924,
+      "step": 473
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.878884450793053e-06,
+      "loss": 0.2841,
+      "step": 474
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.859039072735439e-06,
+      "loss": 0.3619,
+      "step": 475
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.839212638218747e-06,
+      "loss": 0.3063,
+      "step": 476
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.819405476424164e-06,
+      "loss": 0.3808,
+      "step": 477
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7996179162128905e-06,
+      "loss": 0.3008,
+      "step": 478
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.779850286120677e-06,
+      "loss": 0.3169,
+      "step": 479
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7601029143523767e-06,
+      "loss": 0.3157,
+      "step": 480
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.740376128776486e-06,
+      "loss": 0.3004,
+      "step": 481
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.720670256919706e-06,
+      "loss": 0.3064,
+      "step": 482
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7009856259615074e-06,
+      "loss": 0.3131,
+      "step": 483
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6813225627286907e-06,
+      "loss": 0.3201,
+      "step": 484
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6616813936899674e-06,
+      "loss": 0.3119,
+      "step": 485
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.642062444950537e-06,
+      "loss": 0.333,
+      "step": 486
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.622466042246669e-06,
+      "loss": 0.3321,
+      "step": 487
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6028925109403014e-06,
+      "loss": 0.3233,
+      "step": 488
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.5833421760136323e-06,
+      "loss": 0.2676,
+      "step": 489
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5638153620637283e-06,
+      "loss": 0.3336,
+      "step": 490
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.544312393297133e-06,
+      "loss": 0.3453,
+      "step": 491
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.524833593524487e-06,
+      "loss": 0.2824,
+      "step": 492
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.505379286155145e-06,
+      "loss": 0.318,
+      "step": 493
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4859497941918173e-06,
+      "loss": 0.3373,
+      "step": 494
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.4665454402251937e-06,
+      "loss": 0.3361,
+      "step": 495
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.4471665464285993e-06,
+      "loss": 0.3103,
+      "step": 496
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.4278134345526363e-06,
+      "loss": 0.3269,
+      "step": 497
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.40848642591985e-06,
+      "loss": 0.3137,
+      "step": 498
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.3891858414193875e-06,
+      "loss": 0.2976,
+      "step": 499
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.3699120015016696e-06,
+      "loss": 0.385,
+      "step": 500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.350665226173078e-06,
+      "loss": 0.2808,
+      "step": 501
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.331445834990633e-06,
+      "loss": 0.3101,
+      "step": 502
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.312254147056694e-06,
+      "loss": 0.3255,
+      "step": 503
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.293090481013661e-06,
+      "loss": 0.288,
+      "step": 504
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.273955155038677e-06,
+      "loss": 0.3123,
+      "step": 505
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.254848486838357e-06,
+      "loss": 0.2844,
+      "step": 506
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.2357707936435013e-06,
+      "loss": 0.256,
+      "step": 507
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.2167223922038393e-06,
+      "loss": 0.2284,
+      "step": 508
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.197703598782759e-06,
+      "loss": 0.2628,
+      "step": 509
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1787147291520675e-06,
+      "loss": 0.2879,
+      "step": 510
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.1597560985867393e-06,
+      "loss": 0.2559,
+      "step": 511
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.1408280218596893e-06,
+      "loss": 0.3208,
+      "step": 512
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.1219308132365365e-06,
+      "loss": 0.3214,
+      "step": 513
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.103064786470399e-06,
+      "loss": 0.2904,
+      "step": 514
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0842302547966716e-06,
+      "loss": 0.3476,
+      "step": 515
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0654275309278382e-06,
+      "loss": 0.3299,
+      "step": 516
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.046656927048267e-06,
+      "loss": 0.32,
+      "step": 517
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0279187548090362e-06,
+      "loss": 0.2823,
+      "step": 518
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0092133253227563e-06,
+      "loss": 0.3282,
+      "step": 519
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.9905409491584047e-06,
+      "loss": 0.3057,
+      "step": 520
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.971901936336172e-06,
+      "loss": 0.3554,
+      "step": 521
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.9532965963223076e-06,
+      "loss": 0.3025,
+      "step": 522
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.934725238023992e-06,
+      "loss": 0.3324,
+      "step": 523
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.916188169784197e-06,
+      "loss": 0.2954,
+      "step": 524
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8976856993765766e-06,
+      "loss": 0.3043,
+      "step": 525
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.879218134000348e-06,
+      "loss": 0.2777,
+      "step": 526
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.860785780275195e-06,
+      "loss": 0.3056,
+      "step": 527
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8423889442361797e-06,
+      "loss": 0.2495,
+      "step": 528
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8240279313286604e-06,
+      "loss": 0.2861,
+      "step": 529
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8057030464032147e-06,
+      "loss": 0.3049,
+      "step": 530
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.787414593710583e-06,
+      "loss": 0.3014,
+      "step": 531
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7691628768966137e-06,
+      "loss": 0.2864,
+      "step": 532
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7509481989972326e-06,
+      "loss": 0.3151,
+      "step": 533
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7327708624333936e-06,
+      "loss": 0.2994,
+      "step": 534
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.714631169006072e-06,
+      "loss": 0.3348,
+      "step": 535
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.696529419891246e-06,
+      "loss": 0.2387,
+      "step": 536
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.678465915634899e-06,
+      "loss": 0.2451,
+      "step": 537
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.6604409561480337e-06,
+      "loss": 0.2019,
+      "step": 538
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.6424548407016836e-06,
+      "loss": 0.29,
+      "step": 539
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.6245078679219503e-06,
+      "loss": 0.2275,
+      "step": 540
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.6066003357850426e-06,
+      "loss": 0.2131,
+      "step": 541
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.5887325416123354e-06,
+      "loss": 0.2056,
+      "step": 542
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.5709047820654236e-06,
+      "loss": 0.1996,
+      "step": 543
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.5531173531412024e-06,
+      "loss": 0.2429,
+      "step": 544
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.5353705501669506e-06,
+      "loss": 0.2148,
+      "step": 545
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.517664667795434e-06,
+      "loss": 0.2234,
+      "step": 546
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.5000000000000015e-06,
+      "loss": 0.1758,
+      "step": 547
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4823768400697125e-06,
+      "loss": 0.2567,
+      "step": 548
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4647954806044633e-06,
+      "loss": 0.2115,
+      "step": 549
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4472562135101376e-06,
+      "loss": 0.2593,
+      "step": 550
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4297593299937456e-06,
+      "loss": 0.2079,
+      "step": 551
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.412305120558599e-06,
+      "loss": 0.2055,
+      "step": 552
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.3948938749994893e-06,
+      "loss": 0.2744,
+      "step": 553
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.3775258823978685e-06,
+      "loss": 0.2173,
+      "step": 554
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.3602014311170524e-06,
+      "loss": 0.2341,
+      "step": 555
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.3429208087974397e-06,
+      "loss": 0.2354,
+      "step": 556
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.3256843023517245e-06,
+      "loss": 0.2244,
+      "step": 557
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.308492197960141e-06,
+      "loss": 0.2225,
+      "step": 558
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.291344781065708e-06,
+      "loss": 0.2189,
+      "step": 559
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.274242336369496e-06,
+      "loss": 0.1943,
+      "step": 560
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2571851478258903e-06,
+      "loss": 0.1896,
+      "step": 561
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2401734986378835e-06,
+      "loss": 0.2224,
+      "step": 562
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.223207671252371e-06,
+      "loss": 0.2318,
+      "step": 563
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2062879473554654e-06,
+      "loss": 0.2047,
+      "step": 564
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.189414607867812e-06,
+      "loss": 0.1969,
+      "step": 565
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.1725879329399313e-06,
+      "loss": 0.186,
+      "step": 566
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.155808201947563e-06,
+      "loss": 0.2344,
+      "step": 567
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.139075693487035e-06,
+      "loss": 0.2176,
+      "step": 568
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.122390685370627e-06,
+      "loss": 0.2317,
+      "step": 569
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.105753454621966e-06,
+      "loss": 0.2116,
+      "step": 570
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.089164277471425e-06,
+      "loss": 0.2394,
+      "step": 571
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.072623429351532e-06,
+      "loss": 0.2567,
+      "step": 572
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.0561311848924082e-06,
+      "loss": 0.2321,
+      "step": 573
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.039687817917196e-06,
+      "loss": 0.2328,
+      "step": 574
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.023293601437518e-06,
+      "loss": 0.2186,
+      "step": 575
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.0069488076489445e-06,
+      "loss": 0.2136,
+      "step": 576
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9906537079264746e-06,
+      "loss": 0.1901,
+      "step": 577
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.974408572820031e-06,
+      "loss": 0.2009,
+      "step": 578
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.958213672049964e-06,
+      "loss": 0.2243,
+      "step": 579
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9420692745025744e-06,
+      "loss": 0.2552,
+      "step": 580
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9259756482256526e-06,
+      "loss": 0.2426,
+      "step": 581
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.909933060424029e-06,
+      "loss": 0.21,
+      "step": 582
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.8939417774551311e-06,
+      "loss": 0.2136,
+      "step": 583
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.8780020648245645e-06,
+      "loss": 0.205,
+      "step": 584
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.862114187181705e-06,
+      "loss": 0.2338,
+      "step": 585
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.8462784083153091e-06,
+      "loss": 0.2091,
+      "step": 586
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.8304949911491244e-06,
+      "loss": 0.241,
+      "step": 587
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.8147641977375313e-06,
+      "loss": 0.1846,
+      "step": 588
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.799086289261191e-06,
+      "loss": 0.2407,
+      "step": 589
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.7834615260227057e-06,
+      "loss": 0.2331,
+      "step": 590
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.7678901674423044e-06,
+      "loss": 0.2928,
+      "step": 591
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.7523724720535274e-06,
+      "loss": 0.2061,
+      "step": 592
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.7369086974989369e-06,
+      "loss": 0.2426,
+      "step": 593
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7214991005258386e-06,
+      "loss": 0.262,
+      "step": 594
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.706143936982026e-06,
+      "loss": 0.2345,
+      "step": 595
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.6908434618115194e-06,
+      "loss": 0.2576,
+      "step": 596
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.6755979290503437e-06,
+      "loss": 0.2054,
+      "step": 597
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.660407591822304e-06,
+      "loss": 0.2313,
+      "step": 598
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.6452727023347909e-06,
+      "loss": 0.1974,
+      "step": 599
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.6301935118745826e-06,
+      "loss": 0.2415,
+      "step": 600
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.6151702708036837e-06,
+      "loss": 0.2455,
+      "step": 601
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.6002032285551577e-06,
+      "loss": 0.2515,
+      "step": 602
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5852926336289926e-06,
+      "loss": 0.2426,
+      "step": 603
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5704387335879762e-06,
+      "loss": 0.2567,
+      "step": 604
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.555641775053579e-06,
+      "loss": 0.2365,
+      "step": 605
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5409020037018652e-06,
+      "loss": 0.2384,
+      "step": 606
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5262196642594086e-06,
+      "loss": 0.2247,
+      "step": 607
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5115950004992397e-06,
+      "loss": 0.2161,
+      "step": 608
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.4970282552367854e-06,
+      "loss": 0.2208,
+      "step": 609
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.4825196703258442e-06,
+      "loss": 0.2173,
+      "step": 610
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.4680694866545708e-06,
+      "loss": 0.2632,
+      "step": 611
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.453677944141474e-06,
+      "loss": 0.2281,
+      "step": 612
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.4393452817314385e-06,
+      "loss": 0.2233,
+      "step": 613
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.4250717373917506e-06,
+      "loss": 0.2058,
+      "step": 614
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.4108575481081522e-06,
+      "loss": 0.2106,
+      "step": 615
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.3967029498809015e-06,
+      "loss": 0.1998,
+      "step": 616
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.3826081777208638e-06,
+      "loss": 0.2285,
+      "step": 617
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.368573465645599e-06,
+      "loss": 0.2716,
+      "step": 618
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.3545990466754805e-06,
+      "loss": 0.1904,
+      "step": 619
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.3406851528298264e-06,
+      "loss": 0.2023,
+      "step": 620
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.3268320151230518e-06,
+      "loss": 0.2207,
+      "step": 621
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.3130398635608237e-06,
+      "loss": 0.2239,
+      "step": 622
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.2993089271362491e-06,
+      "loss": 0.1774,
+      "step": 623
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.2856394338260691e-06,
+      "loss": 0.2267,
+      "step": 624
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.272031610586879e-06,
+      "loss": 0.237,
+      "step": 625
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.2584856833513581e-06,
+      "loss": 0.2514,
+      "step": 626
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.245001877024512e-06,
+      "loss": 0.2379,
+      "step": 627
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.231580415479947e-06,
+      "loss": 0.2141,
+      "step": 628
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.2182215215561498e-06,
+      "loss": 0.2728,
+      "step": 629
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.2049254170527857e-06,
+      "loss": 0.2345,
+      "step": 630
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1916923227270228e-06,
+      "loss": 0.1998,
+      "step": 631
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.178522458289857e-06,
+      "loss": 0.2128,
+      "step": 632
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1654160424024718e-06,
+      "loss": 0.2096,
+      "step": 633
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1523732926726029e-06,
+      "loss": 0.1985,
+      "step": 634
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.139394425650931e-06,
+      "loss": 0.2119,
+      "step": 635
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1264796568274811e-06,
+      "loss": 0.2246,
+      "step": 636
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1136292006280447e-06,
+      "loss": 0.2074,
+      "step": 637
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1008432704106225e-06,
+      "loss": 0.1846,
+      "step": 638
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.088122078461884e-06,
+      "loss": 0.2163,
+      "step": 639
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0754658359936359e-06,
+      "loss": 0.23,
+      "step": 640
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0628747531393202e-06,
+      "loss": 0.1924,
+      "step": 641
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0503490389505244e-06,
+      "loss": 0.1792,
+      "step": 642
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0378889013935096e-06,
+      "loss": 0.2036,
+      "step": 643
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0254945473457622e-06,
+      "loss": 0.2444,
+      "step": 644
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.013166182592551e-06,
+      "loss": 0.2168,
+      "step": 645
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0009040118235175e-06,
+      "loss": 0.2217,
+      "step": 646
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.887082386292718e-07,
+      "loss": 0.2264,
+      "step": 647
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.765790654980195e-07,
+      "loss": 0.1934,
+      "step": 648
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.645166938121946e-07,
+      "loss": 0.2432,
+      "step": 649
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.525213238451148e-07,
+      "loss": 0.2521,
+      "step": 650
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.405931547576591e-07,
+      "loss": 0.2454,
+      "step": 651
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.287323845949609e-07,
+      "loss": 0.2344,
+      "step": 652
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.169392102831215e-07,
+      "loss": 0.1988,
+      "step": 653
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.052138276259348e-07,
+      "loss": 0.2388,
+      "step": 654
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 8.935564313016426e-07,
+      "loss": 0.2442,
+      "step": 655
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.819672148596964e-07,
+      "loss": 0.2125,
+      "step": 656
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.704463707175526e-07,
+      "loss": 0.2377,
+      "step": 657
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.589940901574678e-07,
+      "loss": 0.2541,
+      "step": 658
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.476105633233295e-07,
+      "loss": 0.2462,
+      "step": 659
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.362959792174941e-07,
+      "loss": 0.1772,
+      "step": 660
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.250505256976565e-07,
+      "loss": 0.2402,
+      "step": 661
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.138743894737238e-07,
+      "loss": 0.2112,
+      "step": 662
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.027677561047176e-07,
+      "loss": 0.2082,
+      "step": 663
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.917308099956938e-07,
+      "loss": 0.2669,
+      "step": 664
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.807637343946794e-07,
+      "loss": 0.2423,
+      "step": 665
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.698667113896346e-07,
+      "loss": 0.2418,
+      "step": 666
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.590399219054234e-07,
+      "loss": 0.2142,
+      "step": 667
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.482835457008125e-07,
+      "loss": 0.2148,
+      "step": 668
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.375977613654861e-07,
+      "loss": 0.2323,
+      "step": 669
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.269827463170847e-07,
+      "loss": 0.1871,
+      "step": 670
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.164386767982523e-07,
+      "loss": 0.1946,
+      "step": 671
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.059657278737136e-07,
+      "loss": 0.1994,
+      "step": 672
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.955640734273716e-07,
+      "loss": 0.237,
+      "step": 673
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.852338861594127e-07,
+      "loss": 0.2096,
+      "step": 674
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.749753375834467e-07,
+      "loss": 0.2036,
+      "step": 675
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.647885980236529e-07,
+      "loss": 0.2033,
+      "step": 676
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.546738366119564e-07,
+      "loss": 0.2129,
+      "step": 677
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.446312212852162e-07,
+      "loss": 0.1915,
+      "step": 678
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.346609187824443e-07,
+      "loss": 0.1824,
+      "step": 679
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.247630946420285e-07,
+      "loss": 0.2161,
+      "step": 680
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.14937913198988e-07,
+      "loss": 0.223,
+      "step": 681
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.051855375822458e-07,
+      "loss": 0.2265,
+      "step": 682
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.955061297119175e-07,
+      "loss": 0.2293,
+      "step": 683
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.858998502966273e-07,
+      "loss": 0.2454,
+      "step": 684
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.763668588308352e-07,
+      "loss": 0.2117,
+      "step": 685
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.66907313592191e-07,
+      "loss": 0.222,
+      "step": 686
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.575213716389039e-07,
+      "loss": 0.2464,
+      "step": 687
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.482091888071417e-07,
+      "loss": 0.2087,
+      "step": 688
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.389709197084348e-07,
+      "loss": 0.2354,
+      "step": 689
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.298067177271144e-07,
+      "loss": 0.2039,
+      "step": 690
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.207167350177639e-07,
+      "loss": 0.2394,
+      "step": 691
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.117011225026958e-07,
+      "loss": 0.2411,
+      "step": 692
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.027600298694397e-07,
+      "loss": 0.2379,
+      "step": 693
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.938936055682636e-07,
+      "loss": 0.243,
+      "step": 694
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.851019968097037e-07,
+      "loss": 0.2315,
+      "step": 695
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.763853495621251e-07,
+      "loss": 0.2254,
+      "step": 696
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.6774380854929634e-07,
+      "loss": 0.2711,
+      "step": 697
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.59177517247984e-07,
+      "loss": 0.2188,
+      "step": 698
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.5068661788557345e-07,
+      "loss": 0.2201,
+      "step": 699
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.4227125143770755e-07,
+      "loss": 0.2211,
+      "step": 700
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.3393155762594343e-07,
+      "loss": 0.1843,
+      "step": 701
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.2566767491543706e-07,
+      "loss": 0.2305,
+      "step": 702
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.174797405126396e-07,
+      "loss": 0.2284,
+      "step": 703
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.093678903630216e-07,
+      "loss": 0.2298,
+      "step": 704
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.01332259148815e-07,
+      "loss": 0.2261,
+      "step": 705
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.933729802867803e-07,
+      "loss": 0.2522,
+      "step": 706
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.854901859259863e-07,
+      "loss": 0.2057,
+      "step": 707
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.776840069456189e-07,
+      "loss": 0.1833,
+      "step": 708
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.69954572952807e-07,
+      "loss": 0.2332,
+      "step": 709
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6230201228047326e-07,
+      "loss": 0.24,
+      "step": 710
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.5472645198520064e-07,
+      "loss": 0.2315,
+      "step": 711
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.472280178451226e-07,
+      "loss": 0.2043,
+      "step": 712
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.398068343578348e-07,
+      "loss": 0.2475,
+      "step": 713
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.324630247383337e-07,
+      "loss": 0.1574,
+      "step": 714
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.251967109169607e-07,
+      "loss": 0.2145,
+      "step": 715
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.18008013537387e-07,
+      "loss": 0.2308,
+      "step": 716
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.108970519546034e-07,
+      "loss": 0.235,
+      "step": 717
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.038639442329444e-07,
+      "loss": 0.2476,
+      "step": 718
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.969088071441256e-07,
+      "loss": 0.2223,
+      "step": 719
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.9003175616530264e-07,
+      "loss": 0.2396,
+      "step": 720
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.832329054771593e-07,
+      "loss": 0.212,
+      "step": 721
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.7651236796200464e-07,
+      "loss": 0.2596,
+      "step": 722
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.698702552019045e-07,
+      "loss": 0.2338,
+      "step": 723
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.633066774768289e-07,
+      "loss": 0.1989,
+      "step": 724
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.5682174376281677e-07,
+      "loss": 0.2071,
+      "step": 725
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.50415561730169e-07,
+      "loss": 0.2203,
+      "step": 726
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.4408823774166143e-07,
+      "loss": 0.2033,
+      "step": 727
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.3783987685077947e-07,
+      "loss": 0.2514,
+      "step": 728
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.3167058279997156e-07,
+      "loss": 0.2318,
+      "step": 729
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2558045801892746e-07,
+      "loss": 0.2593,
+      "step": 730
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.1956960362287726e-07,
+      "loss": 0.253,
+      "step": 731
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.136381194109166e-07,
+      "loss": 0.244,
+      "step": 732
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.077861038643436e-07,
+      "loss": 0.2487,
+      "step": 733
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.0201365414502627e-07,
+      "loss": 0.219,
+      "step": 734
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9632086609379041e-07,
+      "loss": 0.2416,
+      "step": 735
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9070783422882822e-07,
+      "loss": 0.2312,
+      "step": 736
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.851746517441272e-07,
+      "loss": 0.2256,
+      "step": 737
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.797214105079248e-07,
+      "loss": 0.2578,
+      "step": 738
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7434820106118077e-07,
+      "loss": 0.1865,
+      "step": 739
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6905511261607667e-07,
+      "loss": 0.2483,
+      "step": 740
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6384223305453417e-07,
+      "loss": 0.2198,
+      "step": 741
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5870964892675255e-07,
+      "loss": 0.221,
+      "step": 742
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.536574454497769e-07,
+      "loss": 0.258,
+      "step": 743
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4868570650607816e-07,
+      "loss": 0.186,
+      "step": 744
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4379451464216488e-07,
+      "loss": 0.1999,
+      "step": 745
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.389839510672103e-07,
+      "loss": 0.2348,
+      "step": 746
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.342540956517041e-07,
+      "loss": 0.2674,
+      "step": 747
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.296050269261262e-07,
+      "loss": 0.2377,
+      "step": 748
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.250368220796433e-07,
+      "loss": 0.2683,
+      "step": 749
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.205495569588283e-07,
+      "loss": 0.2371,
+      "step": 750
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.1614330606639912e-07,
+      "loss": 0.2515,
+      "step": 751
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1181814255998346e-07,
+      "loss": 0.267,
+      "step": 752
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.0757413825090212e-07,
+      "loss": 0.2306,
+      "step": 753
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.0341136360297821e-07,
+      "loss": 0.2179,
+      "step": 754
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.932988773136753e-08,
+      "loss": 0.2173,
+      "step": 755
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.532977840141123e-08,
+      "loss": 0.2556,
+      "step": 756
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.141110202750714e-08,
+      "loss": 0.2201,
+      "step": 757
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.757392367201134e-08,
+      "loss": 0.2439,
+      "step": 758
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.381830704415839e-08,
+      "loss": 0.2196,
+      "step": 759
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.014431449899829e-08,
+      "loss": 0.2468,
+      "step": 760
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.655200703636623e-08,
+      "loss": 0.2216,
+      "step": 761
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.30414442998667e-08,
+      "loss": 0.1994,
+      "step": 762
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.961268457588766e-08,
+      "loss": 0.2205,
+      "step": 763
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.626578479262568e-08,
+      "loss": 0.2378,
+      "step": 764
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.300080051914792e-08,
+      "loss": 0.2357,
+      "step": 765
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.981778596446386e-08,
+      "loss": 0.2281,
+      "step": 766
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.6716793976630544e-08,
+      "loss": 0.1995,
+      "step": 767
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.369787604186993e-08,
+      "loss": 0.2377,
+      "step": 768
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.07610822837179e-08,
+      "loss": 0.1831,
+      "step": 769
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.790646146218991e-08,
+      "loss": 0.2611,
+      "step": 770
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.513406097297224e-08,
+      "loss": 0.2453,
+      "step": 771
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.244392684663534e-08,
+      "loss": 0.2404,
+      "step": 772
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.983610374786895e-08,
+      "loss": 0.2015,
+      "step": 773
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.731063497474152e-08,
+      "loss": 0.249,
+      "step": 774
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.486756245797862e-08,
+      "loss": 0.2136,
+      "step": 775
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.250692676027123e-08,
+      "loss": 0.1933,
+      "step": 776
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.022876707559796e-08,
+      "loss": 0.2765,
+      "step": 777
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.8033121228577266e-08,
+      "loss": 0.2334,
+      "step": 778
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.5920025673837888e-08,
+      "loss": 0.2274,
+      "step": 779
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.3889515495413297e-08,
+      "loss": 0.1901,
+      "step": 780
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.194162440616099e-08,
+      "loss": 0.228,
+      "step": 781
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0076384747201304e-08,
+      "loss": 0.2506,
+      "step": 782
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8293827487380623e-08,
+      "loss": 0.1897,
+      "step": 783
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.659398222275732e-08,
+      "loss": 0.2019,
+      "step": 784
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.497687717611107e-08,
+      "loss": 0.2434,
+      "step": 785
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.3442539196472647e-08,
+      "loss": 0.2148,
+      "step": 786
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.1990993758679848e-08,
+      "loss": 0.219,
+      "step": 787
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0622264962953377e-08,
+      "loss": 0.192,
+      "step": 788
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.336375534497732e-09,
+      "loss": 0.1977,
+      "step": 789
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.133346823122612e-09,
+      "loss": 0.2317,
+      "step": 790
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.01319880288931e-09,
+      "loss": 0.1918,
+      "step": 791
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.975950071779313e-09,
+      "loss": 0.2231,
+      "step": 792
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.021617851385107e-09,
+      "loss": 0.2141,
+      "step": 793
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.150217986623184e-09,
+      "loss": 0.2099,
+      "step": 794
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.361764945473134e-09,
+      "loss": 0.2595,
+      "step": 795
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.656271818735623e-09,
+      "loss": 0.2249,
+      "step": 796
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.033750319817007e-09,
+      "loss": 0.1926,
+      "step": 797
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4942107845317132e-09,
+      "loss": 0.232,
+      "step": 798
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0376621709340396e-09,
+      "loss": 0.2183,
+      "step": 799
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.641120591666105e-10,
+      "loss": 0.2367,
+      "step": 800
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.735666513371428e-10,
+      "loss": 0.2081,
+      "step": 801
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6603077141352786e-10,
+      "loss": 0.2642,
+      "step": 802
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.150786514389715e-11,
+      "loss": 0.2708,
+      "step": 803
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.221,
+      "step": 804
+    },
+    {
+      "epoch": 3.0,
+      "step": 804,
+      "total_flos": 2.1243591603498516e+18,
+      "train_loss": 0.3452750629935395,
+      "train_runtime": 17473.4569,
+      "train_samples_per_second": 5.872,
+      "train_steps_per_second": 0.046
+    }
+  ],
+  "max_steps": 804,
+  "num_train_epochs": 3,
+  "total_flos": 2.1243591603498516e+18,
+  "trial_name": null,
+  "trial_params": null
+}