diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,13975 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.999677523379555,
+  "global_step": 2325,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 7.1173,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 6.8988,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 5.8169,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.714285714285714e-05,
+      "loss": 5.1234,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.142857142857142e-05,
+      "loss": 4.6262,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.571428571428571e-05,
+      "loss": 4.023,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 3.6811,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011428571428571428,
+      "loss": 3.4244,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012857142857142855,
+      "loss": 3.2695,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014285714285714284,
+      "loss": 3.1907,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015714285714285713,
+      "loss": 3.17,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017142857142857143,
+      "loss": 3.0468,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018571428571428572,
+      "loss": 3.0251,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 2.9694,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00021428571428571427,
+      "loss": 2.9467,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022857142857142857,
+      "loss": 2.8464,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00024285714285714286,
+      "loss": 2.8554,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002571428571428571,
+      "loss": 2.8062,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002714285714285714,
+      "loss": 2.7252,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002857142857142857,
+      "loss": 2.7115,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003,
+      "loss": 2.6823,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00031428571428571427,
+      "loss": 2.6615,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00032857142857142856,
+      "loss": 2.6304,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00034285714285714285,
+      "loss": 2.6045,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00035714285714285714,
+      "loss": 2.577,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00037142857142857143,
+      "loss": 2.576,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003857142857142857,
+      "loss": 2.5407,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004,
+      "loss": 2.4894,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004142857142857143,
+      "loss": 2.4883,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00042857142857142855,
+      "loss": 2.4709,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00044285714285714284,
+      "loss": 2.4803,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00045714285714285713,
+      "loss": 2.4547,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004714285714285714,
+      "loss": 2.461,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004857142857142857,
+      "loss": 2.3806,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005,
+      "loss": 2.366,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005142857142857142,
+      "loss": 2.3526,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005285714285714286,
+      "loss": 2.2832,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005428571428571428,
+      "loss": 2.242,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005571428571428572,
+      "loss": 2.2121,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005714285714285714,
+      "loss": 2.1552,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005857142857142858,
+      "loss": 2.1252,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006,
+      "loss": 1.985,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006142857142857143,
+      "loss": 1.9463,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006285714285714285,
+      "loss": 1.9088,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006428571428571429,
+      "loss": 1.8761,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006571428571428571,
+      "loss": 1.8289,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006714285714285714,
+      "loss": 1.8541,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006857142857142857,
+      "loss": 1.8582,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007,
+      "loss": 1.8327,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007142857142857143,
+      "loss": 1.807,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007285714285714286,
+      "loss": 1.8403,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007428571428571429,
+      "loss": 1.7653,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007571428571428572,
+      "loss": 1.791,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007714285714285715,
+      "loss": 1.8201,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007857142857142857,
+      "loss": 1.7917,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008,
+      "loss": 1.7602,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008142857142857143,
+      "loss": 1.7531,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008285714285714286,
+      "loss": 1.7488,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008428571428571429,
+      "loss": 1.7664,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008571428571428571,
+      "loss": 1.7952,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008714285714285715,
+      "loss": 1.7687,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008857142857142857,
+      "loss": 1.8269,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009000000000000001,
+      "loss": 1.8018,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009142857142857143,
+      "loss": 1.7591,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009285714285714287,
+      "loss": 1.797,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009428571428571429,
+      "loss": 1.8064,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009571428571428573,
+      "loss": 1.7807,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009714285714285714,
+      "loss": 1.7491,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009857142857142857,
+      "loss": 1.734,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001,
+      "loss": 1.7504,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999999514771176,
+      "loss": 1.7543,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999980590856454,
+      "loss": 1.7863,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999956329462339,
+      "loss": 1.7687,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999922363576504,
+      "loss": 1.7516,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999878693264872,
+      "loss": 1.7659,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999825318612206,
+      "loss": 1.798,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999762239722098,
+      "loss": 1.7856,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999689456716982,
+      "loss": 1.7287,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999606969738123,
+      "loss": 1.7539,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999951477894562,
+      "loss": 1.7828,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999412884518408,
+      "loss": 1.7727,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999301286654258,
+      "loss": 1.7699,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999179985569766,
+      "loss": 1.7511,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999048981500372,
+      "loss": 1.7084,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998908274700344,
+      "loss": 1.7081,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999875786544278,
+      "loss": 1.7488,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998597754019612,
+      "loss": 1.7749,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998427940741603,
+      "loss": 1.7328,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998248425938346,
+      "loss": 1.7394,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998059209958266,
+      "loss": 1.7348,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997860293168611,
+      "loss": 1.7676,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997651675955465,
+      "loss": 1.7644,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997433358723737,
+      "loss": 1.7636,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997205341897158,
+      "loss": 1.731,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996967625918293,
+      "loss": 1.7328,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996720211248528,
+      "loss": 1.7264,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996463098368073,
+      "loss": 1.7235,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996196287775962,
+      "loss": 1.7651,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995919779990055,
+      "loss": 1.7205,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995633575547026,
+      "loss": 1.7054,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995337675002376,
+      "loss": 1.6914,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999503207893042,
+      "loss": 1.7325,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994716787924299,
+      "loss": 1.6909,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994391802595963,
+      "loss": 1.7086,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999405712357618,
+      "loss": 1.7708,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993712751514537,
+      "loss": 1.7012,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993358687079429,
+      "loss": 1.7022,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992994930958063,
+      "loss": 1.696,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992621483856463,
+      "loss": 1.7431,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992238346499456,
+      "loss": 1.7084,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991845519630679,
+      "loss": 1.7228,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991443004012575,
+      "loss": 1.6836,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991030800426395,
+      "loss": 1.732,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999060890967219,
+      "loss": 1.7,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990177332568813,
+      "loss": 1.7118,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998973606995392,
+      "loss": 1.6913,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989285122683966,
+      "loss": 1.7129,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988824491634198,
+      "loss": 1.71,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988354177698662,
+      "loss": 1.7197,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987874181790202,
+      "loss": 1.6955,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987384504840445,
+      "loss": 1.6827,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986885147799814,
+      "loss": 1.7083,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998637611163752,
+      "loss": 1.6887,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985857397341554,
+      "loss": 1.7355,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985329005918703,
+      "loss": 1.6643,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984790938394525,
+      "loss": 1.6786,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984243195813368,
+      "loss": 1.6817,
+      "step": 127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998368577923835,
+      "loss": 1.7139,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998311868975137,
+      "loss": 1.6888,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982541928453105,
+      "loss": 1.6581,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981955496462995,
+      "loss": 1.6614,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981359394919257,
+      "loss": 1.6963,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998075362497887,
+      "loss": 1.6552,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980138187817586,
+      "loss": 1.701,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979513084629917,
+      "loss": 1.6656,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978878316629133,
+      "loss": 1.6633,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978233885047265,
+      "loss": 1.6561,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00099775797911351,
+      "loss": 1.64,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976916036162182,
+      "loss": 1.607,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00099762426214168,
+      "loss": 1.6542,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975559548205996,
+      "loss": 1.686,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974866817855556,
+      "loss": 1.6945,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974164431710013,
+      "loss": 1.6216,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973452391132639,
+      "loss": 1.6513,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997273069750544,
+      "loss": 1.6505,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997199935222917,
+      "loss": 1.6241,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00099712583567233,
+      "loss": 1.6413,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970507712426045,
+      "loss": 1.6309,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000996974742079434,
+      "loss": 1.6916,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968977483303847,
+      "loss": 1.6606,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000996819790144895,
+      "loss": 1.6607,
+      "step": 151
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967408676742752,
+      "loss": 1.6049,
+      "step": 152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966609810717069,
+      "loss": 1.6439,
+      "step": 153
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965801304922433,
+      "loss": 1.6522,
+      "step": 154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964983160928086,
+      "loss": 1.6565,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964155380321976,
+      "loss": 1.6307,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963317964710753,
+      "loss": 1.6683,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962470915719774,
+      "loss": 1.5944,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961614234993087,
+      "loss": 1.6543,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996074792419344,
+      "loss": 1.6215,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995987198500226,
+      "loss": 1.6369,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958986419119679,
+      "loss": 1.6412,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958091228264502,
+      "loss": 1.5846,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995718641417422,
+      "loss": 1.6469,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956271978605,
+      "loss": 1.6002,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955347923331684,
+      "loss": 1.621,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009954414250147785,
+      "loss": 1.6096,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009953470960865484,
+      "loss": 1.5767,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952518057315624,
+      "loss": 1.6096,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995155554134771,
+      "loss": 1.6146,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950583414829906,
+      "loss": 1.626,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009949601679649026,
+      "loss": 1.6059,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009948610337710538,
+      "loss": 1.6192,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000994760939093855,
+      "loss": 1.5413,
+      "step": 174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009946598841275812,
+      "loss": 1.5814,
+      "step": 175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945578690683718,
+      "loss": 1.6162,
+      "step": 176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009944548941142296,
+      "loss": 1.5797,
+      "step": 177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009943509594650198,
+      "loss": 1.5985,
+      "step": 178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009942460653224712,
+      "loss": 1.6401,
+      "step": 179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009941402118901744,
+      "loss": 1.631,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940333993735816,
+      "loss": 1.5753,
+      "step": 181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993925627980007,
+      "loss": 1.6165,
+      "step": 182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993816897918626,
+      "loss": 1.5631,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993707209400474,
+      "loss": 1.5983,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935965626384476,
+      "loss": 1.5742,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009934849578473025,
+      "loss": 1.5698,
+      "step": 186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009933723952436542,
+      "loss": 1.5726,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009932588750459772,
+      "loss": 1.5834,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009931443974746045,
+      "loss": 1.5847,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009930289627517275,
+      "loss": 1.5981,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992912571101395,
+      "loss": 1.637,
+      "step": 191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009927952227495138,
+      "loss": 1.5689,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009926769179238466,
+      "loss": 1.5357,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009925576568540133,
+      "loss": 1.5935,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009924374397714893,
+      "loss": 1.5572,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009923162669096062,
+      "loss": 1.5897,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009921941385035498,
+      "loss": 1.5569,
+      "step": 197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009920710547903615,
+      "loss": 1.5394,
+      "step": 198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991947016008936,
+      "loss": 1.5698,
+      "step": 199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991822022400022,
+      "loss": 1.5418,
+      "step": 200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009916960742062217,
+      "loss": 1.5874,
+      "step": 201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009915691716719899,
+      "loss": 1.5739,
+      "step": 202
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009914413150436333,
+      "loss": 1.5963,
+      "step": 203
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009913125045693115,
+      "loss": 1.5461,
+      "step": 204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009911827404990342,
+      "loss": 1.5711,
+      "step": 205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009910520230846623,
+      "loss": 1.5446,
+      "step": 206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009909203525799077,
+      "loss": 1.5485,
+      "step": 207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009907877292403317,
+      "loss": 1.545,
+      "step": 208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009906541533233444,
+      "loss": 1.5998,
+      "step": 209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009905196250882058,
+      "loss": 1.5421,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009903841447960239,
+      "loss": 1.5627,
+      "step": 211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990247712709754,
+      "loss": 1.5607,
+      "step": 212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009901103290941996,
+      "loss": 1.5503,
+      "step": 213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009899719942160106,
+      "loss": 1.5478,
+      "step": 214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009898327083436834,
+      "loss": 1.5566,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009896924717475598,
+      "loss": 1.5635,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989551284699827,
+      "loss": 1.5956,
+      "step": 217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989409147474518,
+      "loss": 1.5444,
+      "step": 218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009892660603475082,
+      "loss": 1.5627,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009891220235965178,
+      "loss": 1.5526,
+      "step": 220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009889770375011101,
+      "loss": 1.6079,
+      "step": 221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009888311023426907,
+      "loss": 1.5228,
+      "step": 222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009886842184045075,
+      "loss": 1.5188,
+      "step": 223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009885363859716495,
+      "loss": 1.5733,
+      "step": 224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009883876053310475,
+      "loss": 1.545,
+      "step": 225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009882378767714715,
+      "loss": 1.5447,
+      "step": 226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009880872005835323,
+      "loss": 1.5422,
+      "step": 227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009879355770596797,
+      "loss": 1.507,
+      "step": 228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987783006494202,
+      "loss": 1.56,
+      "step": 229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009876294891832255,
+      "loss": 1.5236,
+      "step": 230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009874750254247144,
+      "loss": 1.5522,
+      "step": 231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009873196155184701,
+      "loss": 1.5562,
+      "step": 232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009871632597661297,
+      "loss": 1.5627,
+      "step": 233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009870059584711668,
+      "loss": 1.5567,
+      "step": 234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009868477119388895,
+      "loss": 1.5321,
+      "step": 235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009866885204764411,
+      "loss": 1.5896,
+      "step": 236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986528384392799,
+      "loss": 1.5596,
+      "step": 237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009863673039987733,
+      "loss": 1.5818,
+      "step": 238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009862052796070079,
+      "loss": 1.5643,
+      "step": 239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009860423115319779,
+      "loss": 1.5093,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009858784000899908,
+      "loss": 1.5234,
+      "step": 241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009857135455991848,
+      "loss": 1.5154,
+      "step": 242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009855477483795285,
+      "loss": 1.5342,
+      "step": 243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009853810087528203,
+      "loss": 1.5161,
+      "step": 244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009852133270426876,
+      "loss": 1.5335,
+      "step": 245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009850447035745866,
+      "loss": 1.4797,
+      "step": 246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984875138675801,
+      "loss": 1.567,
+      "step": 247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009847046326754419,
+      "loss": 1.5262,
+      "step": 248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984533185904447,
+      "loss": 1.5486,
+      "step": 249
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009843607986955798,
+      "loss": 1.5428,
+      "step": 250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009841874713834298,
+      "loss": 1.5295,
+      "step": 251
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009840132043044101,
+      "loss": 1.5239,
+      "step": 252
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009838379977967588,
+      "loss": 1.5252,
+      "step": 253
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009836618522005363,
+      "loss": 1.5256,
+      "step": 254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009834847678576268,
+      "loss": 1.5272,
+      "step": 255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009833067451117357,
+      "loss": 1.5919,
+      "step": 256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009831277843083902,
+      "loss": 1.5384,
+      "step": 257
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009829478857949383,
+      "loss": 1.5364,
+      "step": 258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009827670499205474,
+      "loss": 1.5165,
+      "step": 259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009825852770362045,
+      "loss": 1.5233,
+      "step": 260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982402567494716,
+      "loss": 1.5245,
+      "step": 261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982218921650705,
+      "loss": 1.5204,
+      "step": 262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009820343398606129,
+      "loss": 1.5118,
+      "step": 263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009818488224826972,
+      "loss": 1.4953,
+      "step": 264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009816623698770312,
+      "loss": 1.5121,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009814749824055041,
+      "loss": 1.5329,
+      "step": 266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009812866604318187,
+      "loss": 1.4876,
+      "step": 267
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000981097404321492,
+      "loss": 1.5005,
+      "step": 268
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009809072144418545,
+      "loss": 1.5379,
+      "step": 269
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009807160911620484,
+      "loss": 1.5167,
+      "step": 270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009805240348530276,
+      "loss": 1.5387,
+      "step": 271
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009803310458875573,
+      "loss": 1.5366,
+      "step": 272
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980137124640213,
+      "loss": 1.4875,
+      "step": 273
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979942271487379,
+      "loss": 1.5224,
+      "step": 274
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009797464868072487,
+      "loss": 1.5686,
+      "step": 275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979549770979824,
+      "loss": 1.5311,
+      "step": 276
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009793521243869133,
+      "loss": 1.5191,
+      "step": 277
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009791535474121323,
+      "loss": 1.5046,
+      "step": 278
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009789540404409017,
+      "loss": 1.5412,
+      "step": 279
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009787536038604477,
+      "loss": 1.506,
+      "step": 280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009785522380598008,
+      "loss": 1.5407,
+      "step": 281
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978349943429795,
+      "loss": 1.5146,
+      "step": 282
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978146720363067,
+      "loss": 1.4959,
+      "step": 283
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009779425692540554,
+      "loss": 1.5096,
+      "step": 284
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009777374904990006,
+      "loss": 1.497,
+      "step": 285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009775314844959427,
+      "loss": 1.524,
+      "step": 286
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009773245516447222,
+      "loss": 1.4836,
+      "step": 287
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009771166923469778,
+      "loss": 1.518,
+      "step": 288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009769079070061474,
+      "loss": 1.5355,
+      "step": 289
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009766981960274653,
+      "loss": 1.5135,
+      "step": 290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009764875598179628,
+      "loss": 1.5155,
+      "step": 291
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009762759987864669,
+      "loss": 1.5627,
+      "step": 292
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009760635133435995,
+      "loss": 1.4901,
+      "step": 293
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009758501039017773,
+      "loss": 1.5274,
+      "step": 294
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009756357708752095,
+      "loss": 1.4852,
+      "step": 295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009754205146798987,
+      "loss": 1.4465,
+      "step": 296
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009752043357336385,
+      "loss": 1.5169,
+      "step": 297
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009749872344560145,
+      "loss": 1.5115,
+      "step": 298
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009747692112684012,
+      "loss": 1.5187,
+      "step": 299
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009745502665939638,
+      "loss": 1.5022,
+      "step": 300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000974330400857655,
+      "loss": 1.5568,
+      "step": 301
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009741096144862157,
+      "loss": 1.4908,
+      "step": 302
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009738879079081735,
+      "loss": 1.5195,
+      "step": 303
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009736652815538421,
+      "loss": 1.4803,
+      "step": 304
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009734417358553204,
+      "loss": 1.4662,
+      "step": 305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009732172712464915,
+      "loss": 1.4833,
+      "step": 306
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009729918881630225,
+      "loss": 1.5112,
+      "step": 307
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009727655870423627,
+      "loss": 1.5167,
+      "step": 308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009725383683237435,
+      "loss": 1.5362,
+      "step": 309
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009723102324481769,
+      "loss": 1.5018,
+      "step": 310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009720811798584557,
+      "loss": 1.4848,
+      "step": 311
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009718512109991514,
+      "loss": 1.4303,
+      "step": 312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000971620326316614,
+      "loss": 1.458,
+      "step": 313
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009713885262589712,
+      "loss": 1.5301,
+      "step": 314
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009711558112761274,
+      "loss": 1.5097,
+      "step": 315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009709221818197624,
+      "loss": 1.5116,
+      "step": 316
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009706876383433313,
+      "loss": 1.4851,
+      "step": 317
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009704521813020632,
+      "loss": 1.4798,
+      "step": 318
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009702158111529602,
+      "loss": 1.5185,
+      "step": 319
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009699785283547968,
+      "loss": 1.5167,
+      "step": 320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009697403333681188,
+      "loss": 1.4812,
+      "step": 321
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009695012266552423,
+      "loss": 1.4678,
+      "step": 322
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009692612086802534,
+      "loss": 1.4822,
+      "step": 323
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009690202799090067,
+      "loss": 1.5023,
+      "step": 324
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009687784408091242,
+      "loss": 1.4852,
+      "step": 325
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009685356918499957,
+      "loss": 1.5112,
+      "step": 326
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009682920335027757,
+      "loss": 1.4645,
+      "step": 327
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009680474662403848,
+      "loss": 1.4773,
+      "step": 328
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009678019905375073,
+      "loss": 1.4954,
+      "step": 329
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009675556068705907,
+      "loss": 1.5315,
+      "step": 330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009673083157178448,
+      "loss": 1.5065,
+      "step": 331
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009670601175592409,
+      "loss": 1.5494,
+      "step": 332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009668110128765106,
+      "loss": 1.4699,
+      "step": 333
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009665610021531446,
+      "loss": 1.4673,
+      "step": 334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009663100858743932,
+      "loss": 1.4865,
+      "step": 335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000966058264527263,
+      "loss": 1.5089,
+      "step": 336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009658055386005184,
+      "loss": 1.5155,
+      "step": 337
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009655519085846787,
+      "loss": 1.4499,
+      "step": 338
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009652973749720184,
+      "loss": 1.4929,
+      "step": 339
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009650419382565657,
+      "loss": 1.4757,
+      "step": 340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009647855989341015,
+      "loss": 1.4834,
+      "step": 341
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000964528357502159,
+      "loss": 1.4744,
+      "step": 342
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009642702144600217,
+      "loss": 1.5102,
+      "step": 343
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009640111703087236,
+      "loss": 1.4649,
+      "step": 344
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009637512255510475,
+      "loss": 1.4497,
+      "step": 345
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009634903806915238,
+      "loss": 1.5108,
+      "step": 346
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009632286362364305,
+      "loss": 1.4355,
+      "step": 347
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009629659926937915,
+      "loss": 1.4645,
+      "step": 348
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009627024505733757,
+      "loss": 1.5523,
+      "step": 349
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009624380103866959,
+      "loss": 1.5128,
+      "step": 350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009621726726470079,
+      "loss": 1.5204,
+      "step": 351
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009619064378693104,
+      "loss": 1.4977,
+      "step": 352
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009616393065703418,
+      "loss": 1.4582,
+      "step": 353
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009613712792685818,
+      "loss": 1.4911,
+      "step": 354
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009611023564842487,
+      "loss": 1.4681,
+      "step": 355
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009608325387392986,
+      "loss": 1.4906,
+      "step": 356
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009605618265574251,
+      "loss": 1.4617,
+      "step": 357
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009602902204640574,
+      "loss": 1.4771,
+      "step": 358
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009600177209863602,
+      "loss": 1.4749,
+      "step": 359
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009597443286532317,
+      "loss": 1.4652,
+      "step": 360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009594700439953032,
+      "loss": 1.4709,
+      "step": 361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009591948675449383,
+      "loss": 1.4911,
+      "step": 362
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009589187998362308,
+      "loss": 1.4914,
+      "step": 363
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009586418414050051,
+      "loss": 1.474,
+      "step": 364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009583639927888137,
+      "loss": 1.4514,
+      "step": 365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009580852545269375,
+      "loss": 1.4543,
+      "step": 366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009578056271603837,
+      "loss": 1.4792,
+      "step": 367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009575251112318854,
+      "loss": 1.4627,
+      "step": 368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009572437072859004,
+      "loss": 1.4586,
+      "step": 369
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009569614158686096,
+      "loss": 1.4321,
+      "step": 370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000956678237527917,
+      "loss": 1.4949,
+      "step": 371
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009563941728134476,
+      "loss": 1.4917,
+      "step": 372
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000956109222276547,
+      "loss": 1.4486,
+      "step": 373
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009558233864702803,
+      "loss": 1.4422,
+      "step": 374
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009555366659494303,
+      "loss": 1.4507,
+      "step": 375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009552490612704972,
+      "loss": 1.478,
+      "step": 376
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009549605729916975,
+      "loss": 1.5067,
+      "step": 377
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009546712016729624,
+      "loss": 1.4884,
+      "step": 378
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009543809478759374,
+      "loss": 1.468,
+      "step": 379
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009540898121639799,
+      "loss": 1.4834,
+      "step": 380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009537977951021604,
+      "loss": 1.4527,
+      "step": 381
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009535048972572587,
+      "loss": 1.4829,
+      "step": 382
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009532111191977652,
+      "loss": 1.5097,
+      "step": 383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009529164614938778,
+      "loss": 1.4748,
+      "step": 384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009526209247175024,
+      "loss": 1.5038,
+      "step": 385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009523245094422507,
+      "loss": 1.4678,
+      "step": 386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009520272162434398,
+      "loss": 1.4786,
+      "step": 387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009517290456980905,
+      "loss": 1.4572,
+      "step": 388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009514299983849267,
+      "loss": 1.4911,
+      "step": 389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009511300748843736,
+      "loss": 1.475,
+      "step": 390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009508292757785575,
+      "loss": 1.4773,
+      "step": 391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009505276016513042,
+      "loss": 1.478,
+      "step": 392
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009502250530881374,
+      "loss": 1.49,
+      "step": 393
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009499216306762782,
+      "loss": 1.4892,
+      "step": 394
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000949617335004644,
+      "loss": 1.4383,
+      "step": 395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009493121666638467,
+      "loss": 1.4532,
+      "step": 396
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009490061262461922,
+      "loss": 1.4939,
+      "step": 397
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009486992143456792,
+      "loss": 1.4738,
+      "step": 398
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009483914315579976,
+      "loss": 1.4707,
+      "step": 399
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009480827784805278,
+      "loss": 1.4619,
+      "step": 400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009477732557123393,
+      "loss": 1.4658,
+      "step": 401
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009474628638541893,
+      "loss": 1.4739,
+      "step": 402
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009471516035085224,
+      "loss": 1.4935,
+      "step": 403
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009468394752794684,
+      "loss": 1.4672,
+      "step": 404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000946526479772842,
+      "loss": 1.529,
+      "step": 405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009462126175961407,
+      "loss": 1.459,
+      "step": 406
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009458978893585444,
+      "loss": 1.467,
+      "step": 407
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009455822956709141,
+      "loss": 1.4482,
+      "step": 408
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009452658371457904,
+      "loss": 1.5018,
+      "step": 409
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009449485143973923,
+      "loss": 1.4504,
+      "step": 410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009446303280416168,
+      "loss": 1.4949,
+      "step": 411
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009443112786960362,
+      "loss": 1.4599,
+      "step": 412
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009439913669798985,
+      "loss": 1.4609,
+      "step": 413
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009436705935141251,
+      "loss": 1.4531,
+      "step": 414
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009433489589213102,
+      "loss": 1.4476,
+      "step": 415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009430264638257195,
+      "loss": 1.4805,
+      "step": 416
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009427031088532884,
+      "loss": 1.4624,
+      "step": 417
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009423788946316215,
+      "loss": 1.4436,
+      "step": 418
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009420538217899913,
+      "loss": 1.4542,
+      "step": 419
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009417278909593365,
+      "loss": 1.5004,
+      "step": 420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009414011027722615,
+      "loss": 1.4604,
+      "step": 421
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009410734578630342,
+      "loss": 1.4799,
+      "step": 422
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009407449568675857,
+      "loss": 1.4935,
+      "step": 423
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009404156004235088,
+      "loss": 1.4969,
+      "step": 424
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000940085389170056,
+      "loss": 1.4951,
+      "step": 425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009397543237481398,
+      "loss": 1.4686,
+      "step": 426
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009394224048003302,
+      "loss": 1.4847,
+      "step": 427
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009390896329708533,
+      "loss": 1.4694,
+      "step": 428
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009387560089055913,
+      "loss": 1.4137,
+      "step": 429
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009384215332520805,
+      "loss": 1.4704,
+      "step": 430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009380862066595094,
+      "loss": 1.4431,
+      "step": 431
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009377500297787187,
+      "loss": 1.4837,
+      "step": 432
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009374130032621992,
+      "loss": 1.4322,
+      "step": 433
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000937075127764091,
+      "loss": 1.4637,
+      "step": 434
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009367364039401814,
+      "loss": 1.4537,
+      "step": 435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009363968324479052,
+      "loss": 1.5061,
+      "step": 436
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009360564139463414,
+      "loss": 1.4635,
+      "step": 437
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009357151490962139,
+      "loss": 1.4727,
+      "step": 438
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009353730385598887,
+      "loss": 1.4468,
+      "step": 439
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009350300830013731,
+      "loss": 1.4993,
+      "step": 440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009346862830863152,
+      "loss": 1.5041,
+      "step": 441
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009343416394820014,
+      "loss": 1.4573,
+      "step": 442
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009339961528573557,
+      "loss": 1.4782,
+      "step": 443
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009336498238829383,
+      "loss": 1.4892,
+      "step": 444
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009333026532309443,
+      "loss": 1.4543,
+      "step": 445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000932954641575203,
+      "loss": 1.4524,
+      "step": 446
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009326057895911751,
+      "loss": 1.4515,
+      "step": 447
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009322560979559529,
+      "loss": 1.4793,
+      "step": 448
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009319055673482582,
+      "loss": 1.457,
+      "step": 449
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009315541984484414,
+      "loss": 1.445,
+      "step": 450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009312019919384796,
+      "loss": 1.5042,
+      "step": 451
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009308489485019758,
+      "loss": 1.4777,
+      "step": 452
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009304950688241574,
+      "loss": 1.4706,
+      "step": 453
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009301403535918748,
+      "loss": 1.453,
+      "step": 454
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009297848034936007,
+      "loss": 1.4683,
+      "step": 455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009294284192194271,
+      "loss": 1.5421,
+      "step": 456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009290712014610661,
+      "loss": 1.4508,
+      "step": 457
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009287131509118467,
+      "loss": 1.468,
+      "step": 458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009283542682667152,
+      "loss": 1.4479,
+      "step": 459
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009279945542222321,
+      "loss": 1.4578,
+      "step": 460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009276340094765721,
+      "loss": 1.4802,
+      "step": 461
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009272726347295216,
+      "loss": 1.4959,
+      "step": 462
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009269104306824788,
+      "loss": 1.4186,
+      "step": 463
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000926547398038451,
+      "loss": 1.4291,
+      "step": 464
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009261835375020535,
+      "loss": 1.4522,
+      "step": 465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009258188497795093,
+      "loss": 1.4761,
+      "step": 466
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009254533355786458,
+      "loss": 1.4404,
+      "step": 467
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009250869956088957,
+      "loss": 1.414,
+      "step": 468
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009247198305812934,
+      "loss": 1.4305,
+      "step": 469
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009243518412084753,
+      "loss": 1.417,
+      "step": 470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009239830282046775,
+      "loss": 1.4483,
+      "step": 471
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009236133922857349,
+      "loss": 1.45,
+      "step": 472
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009232429341690796,
+      "loss": 1.4649,
+      "step": 473
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009228716545737392,
+      "loss": 1.406,
+      "step": 474
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009224995542203361,
+      "loss": 1.4501,
+      "step": 475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009221266338310854,
+      "loss": 1.462,
+      "step": 476
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009217528941297942,
+      "loss": 1.4425,
+      "step": 477
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009213783358418595,
+      "loss": 1.5023,
+      "step": 478
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009210029596942672,
+      "loss": 1.4391,
+      "step": 479
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009206267664155906,
+      "loss": 1.4437,
+      "step": 480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009202497567359891,
+      "loss": 1.4162,
+      "step": 481
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009198719313872063,
+      "loss": 1.4219,
+      "step": 482
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009194932911025695,
+      "loss": 1.4904,
+      "step": 483
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009191138366169873,
+      "loss": 1.472,
+      "step": 484
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009187335686669487,
+      "loss": 1.4434,
+      "step": 485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009183524879905216,
+      "loss": 1.432,
+      "step": 486
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009179705953273513,
+      "loss": 1.4515,
+      "step": 487
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000917587891418659,
+      "loss": 1.4361,
+      "step": 488
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009172043770072408,
+      "loss": 1.4579,
+      "step": 489
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009168200528374655,
+      "loss": 1.479,
+      "step": 490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000916434919655274,
+      "loss": 1.4105,
+      "step": 491
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009160489782081768,
+      "loss": 1.4209,
+      "step": 492
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009156622292452538,
+      "loss": 1.4614,
+      "step": 493
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000915274673517152,
+      "loss": 1.4551,
+      "step": 494
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009148863117760842,
+      "loss": 1.4482,
+      "step": 495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009144971447758276,
+      "loss": 1.4411,
+      "step": 496
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009141071732717223,
+      "loss": 1.4802,
+      "step": 497
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009137163980206703,
+      "loss": 1.4685,
+      "step": 498
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000913324819781133,
+      "loss": 1.4995,
+      "step": 499
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009129324393131306,
+      "loss": 1.4936,
+      "step": 500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009125392573782403,
+      "loss": 1.4734,
+      "step": 501
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009121452747395951,
+      "loss": 1.4579,
+      "step": 502
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009117504921618818,
+      "loss": 1.4056,
+      "step": 503
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009113549104113399,
+      "loss": 1.4648,
+      "step": 504
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009109585302557603,
+      "loss": 1.4714,
+      "step": 505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009105613524644831,
+      "loss": 1.4654,
+      "step": 506
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009101633778083968,
+      "loss": 1.4671,
+      "step": 507
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009097646070599366,
+      "loss": 1.4474,
+      "step": 508
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009093650409930827,
+      "loss": 1.4269,
+      "step": 509
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009089646803833589,
+      "loss": 1.4202,
+      "step": 510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009085635260078312,
+      "loss": 1.4159,
+      "step": 511
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009081615786451064,
+      "loss": 1.4553,
+      "step": 512
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009077588390753302,
+      "loss": 1.4482,
+      "step": 513
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009073553080801861,
+      "loss": 1.4099,
+      "step": 514
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009069509864428935,
+      "loss": 1.4736,
+      "step": 515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009065458749482064,
+      "loss": 1.4519,
+      "step": 516
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009061399743824119,
+      "loss": 1.4888,
+      "step": 517
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009057332855333288,
+      "loss": 1.4344,
+      "step": 518
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009053258091903055,
+      "loss": 1.4334,
+      "step": 519
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009049175461442191,
+      "loss": 1.449,
+      "step": 520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009045084971874737,
+      "loss": 1.4493,
+      "step": 521
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009040986631139987,
+      "loss": 1.4099,
+      "step": 522
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000903688044719247,
+      "loss": 1.4406,
+      "step": 523
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009032766428001946,
+      "loss": 1.4455,
+      "step": 524
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009028644581553373,
+      "loss": 1.4806,
+      "step": 525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000902451491584691,
+      "loss": 1.4414,
+      "step": 526
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009020377438897887,
+      "loss": 1.4201,
+      "step": 527
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009016232158736794,
+      "loss": 1.4191,
+      "step": 528
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000901207908340927,
+      "loss": 1.4167,
+      "step": 529
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009007918220976086,
+      "loss": 1.4499,
+      "step": 530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009003749579513117,
+      "loss": 1.4461,
+      "step": 531
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008999573167111348,
+      "loss": 1.4325,
+      "step": 532
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000899538899187684,
+      "loss": 1.4198,
+      "step": 533
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008991197061930722,
+      "loss": 1.428,
+      "step": 534
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008986997385409177,
+      "loss": 1.4911,
+      "step": 535
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000898278997046342,
+      "loss": 1.4776,
+      "step": 536
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008978574825259689,
+      "loss": 1.45,
+      "step": 537
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008974351957979219,
+      "loss": 1.4328,
+      "step": 538
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008970121376818243,
+      "loss": 1.4322,
+      "step": 539
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008965883089987957,
+      "loss": 1.4014,
+      "step": 540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008961637105714518,
+      "loss": 1.4397,
+      "step": 541
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008957383432239022,
+      "loss": 1.4428,
+      "step": 542
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000895312207781749,
+      "loss": 1.4483,
+      "step": 543
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008948853050720849,
+      "loss": 1.4456,
+      "step": 544
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008944576359234917,
+      "loss": 1.4191,
+      "step": 545
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008940292011660394,
+      "loss": 1.4388,
+      "step": 546
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008936000016312831,
+      "loss": 1.4215,
+      "step": 547
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008931700381522631,
+      "loss": 1.4631,
+      "step": 548
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008927393115635021,
+      "loss": 1.4702,
+      "step": 549
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008923078227010037,
+      "loss": 1.4117,
+      "step": 550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008918755724022514,
+      "loss": 1.4404,
+      "step": 551
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008914425615062065,
+      "loss": 1.4234,
+      "step": 552
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008910087908533062,
+      "loss": 1.403,
+      "step": 553
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008905742612854627,
+      "loss": 1.4693,
+      "step": 554
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008901389736460611,
+      "loss": 1.5045,
+      "step": 555
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008897029287799579,
+      "loss": 1.4423,
+      "step": 556
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008892661275334792,
+      "loss": 1.4022,
+      "step": 557
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008888285707544192,
+      "loss": 1.4415,
+      "step": 558
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008883902592920387,
+      "loss": 1.4212,
+      "step": 559
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008879511939970628,
+      "loss": 1.4356,
+      "step": 560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008875113757216803,
+      "loss": 1.434,
+      "step": 561
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008870708053195413,
+      "loss": 1.4051,
+      "step": 562
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008866294836457554,
+      "loss": 1.4099,
+      "step": 563
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008861874115568906,
+      "loss": 1.4327,
+      "step": 564
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008857445899109715,
+      "loss": 1.4277,
+      "step": 565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008853010195674773,
+      "loss": 1.4662,
+      "step": 566
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008848567013873406,
+      "loss": 1.4144,
+      "step": 567
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008844116362329453,
+      "loss": 1.4761,
+      "step": 568
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008839658249681252,
+      "loss": 1.4097,
+      "step": 569
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008835192684581621,
+      "loss": 1.4374,
+      "step": 570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008830719675697842,
+      "loss": 1.3831,
+      "step": 571
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008826239231711652,
+      "loss": 1.5147,
+      "step": 572
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008821751361319207,
+      "loss": 1.4354,
+      "step": 573
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008817256073231088,
+      "loss": 1.4554,
+      "step": 574
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008812753376172265,
+      "loss": 1.4102,
+      "step": 575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008808243278882093,
+      "loss": 1.4173,
+      "step": 576
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000880372579011429,
+      "loss": 1.4133,
+      "step": 577
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008799200918636919,
+      "loss": 1.4257,
+      "step": 578
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008794668673232367,
+      "loss": 1.4383,
+      "step": 579
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008790129062697346,
+      "loss": 1.4094,
+      "step": 580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008785582095842851,
+      "loss": 1.4108,
+      "step": 581
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008781027781494161,
+      "loss": 1.4064,
+      "step": 582
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008776466128490814,
+      "loss": 1.4076,
+      "step": 583
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008771897145686591,
+      "loss": 1.3924,
+      "step": 584
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008767320841949503,
+      "loss": 1.421,
+      "step": 585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008762737226161766,
+      "loss": 1.4026,
+      "step": 586
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008758146307219792,
+      "loss": 1.5166,
+      "step": 587
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008753548094034163,
+      "loss": 1.4277,
+      "step": 588
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008748942595529623,
+      "loss": 1.4425,
+      "step": 589
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008744329820645055,
+      "loss": 1.4267,
+      "step": 590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008739709778333463,
+      "loss": 1.4048,
+      "step": 591
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008735082477561958,
+      "loss": 1.4645,
+      "step": 592
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000873044792731174,
+      "loss": 1.439,
+      "step": 593
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008725806136578077,
+      "loss": 1.4216,
+      "step": 594
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008721157114370293,
+      "loss": 1.4311,
+      "step": 595
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008716500869711746,
+      "loss": 1.4109,
+      "step": 596
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008711837411639811,
+      "loss": 1.4319,
+      "step": 597
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008707166749205865,
+      "loss": 1.4224,
+      "step": 598
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008702488891475273,
+      "loss": 1.4489,
+      "step": 599
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008697803847527354,
+      "loss": 1.3983,
+      "step": 600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008693111626455385,
+      "loss": 1.4099,
+      "step": 601
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000868841223736657,
+      "loss": 1.3821,
+      "step": 602
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008683705689382025,
+      "loss": 1.4478,
+      "step": 603
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000867899199163676,
+      "loss": 1.4156,
+      "step": 604
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008674271153279663,
+      "loss": 1.4341,
+      "step": 605
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000866954318347348,
+      "loss": 1.4043,
+      "step": 606
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008664808091394802,
+      "loss": 1.4604,
+      "step": 607
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008660065886234044,
+      "loss": 1.3484,
+      "step": 608
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000865531657719542,
+      "loss": 1.4181,
+      "step": 609
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008650560173496937,
+      "loss": 1.398,
+      "step": 610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008645796684370375,
+      "loss": 1.4389,
+      "step": 611
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008641026119061262,
+      "loss": 1.4229,
+      "step": 612
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008636248486828857,
+      "loss": 1.4308,
+      "step": 613
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008631463796946144,
+      "loss": 1.3942,
+      "step": 614
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00086266720586998,
+      "loss": 1.4534,
+      "step": 615
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008621873281390183,
+      "loss": 1.4372,
+      "step": 616
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008617067474331312,
+      "loss": 1.411,
+      "step": 617
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008612254646850852,
+      "loss": 1.4055,
+      "step": 618
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008607434808290093,
+      "loss": 1.4135,
+      "step": 619
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008602607968003935,
+      "loss": 1.4332,
+      "step": 620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008597774135360865,
+      "loss": 1.3962,
+      "step": 621
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008592933319742944,
+      "loss": 1.408,
+      "step": 622
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008588085530545781,
+      "loss": 1.4583,
+      "step": 623
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000858323077717853,
+      "loss": 1.4471,
+      "step": 624
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008578369069063852,
+      "loss": 1.383,
+      "step": 625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008573500415637914,
+      "loss": 1.4496,
+      "step": 626
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008568624826350356,
+      "loss": 1.403,
+      "step": 627
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008563742310664285,
+      "loss": 1.3885,
+      "step": 628
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008558852878056252,
+      "loss": 1.3982,
+      "step": 629
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000855395653801623,
+      "loss": 1.4393,
+      "step": 630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008549053300047603,
+      "loss": 1.4237,
+      "step": 631
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008544143173667136,
+      "loss": 1.4626,
+      "step": 632
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008539226168404971,
+      "loss": 1.4824,
+      "step": 633
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008534302293804599,
+      "loss": 1.4394,
+      "step": 634
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008529371559422843,
+      "loss": 1.4761,
+      "step": 635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008524433974829842,
+      "loss": 1.4517,
+      "step": 636
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008519489549609026,
+      "loss": 1.4075,
+      "step": 637
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008514538293357111,
+      "loss": 1.3944,
+      "step": 638
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008509580215684061,
+      "loss": 1.3979,
+      "step": 639
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008504615326213085,
+      "loss": 1.4795,
+      "step": 640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008499643634580616,
+      "loss": 1.3933,
+      "step": 641
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008494665150436287,
+      "loss": 1.3913,
+      "step": 642
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008489679883442909,
+      "loss": 1.4165,
+      "step": 643
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008484687843276469,
+      "loss": 1.4111,
+      "step": 644
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008479689039626087,
+      "loss": 1.4123,
+      "step": 645
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008474683482194024,
+      "loss": 1.415,
+      "step": 646
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000846967118069564,
+      "loss": 1.4057,
+      "step": 647
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008464652144859388,
+      "loss": 1.4105,
+      "step": 648
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008459626384426793,
+      "loss": 1.4263,
+      "step": 649
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008454593909152427,
+      "loss": 1.4602,
+      "step": 650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008449554728803901,
+      "loss": 1.4021,
+      "step": 651
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008444508853161835,
+      "loss": 1.3835,
+      "step": 652
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008439456292019849,
+      "loss": 1.4372,
+      "step": 653
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008434397055184535,
+      "loss": 1.412,
+      "step": 654
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008429331152475443,
+      "loss": 1.4615,
+      "step": 655
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008424258593725059,
+      "loss": 1.4273,
+      "step": 656
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008419179388778793,
+      "loss": 1.3899,
+      "step": 657
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008414093547494951,
+      "loss": 1.3894,
+      "step": 658
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008409001079744719,
+      "loss": 1.3915,
+      "step": 659
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008403901995412146,
+      "loss": 1.4682,
+      "step": 660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008398796304394122,
+      "loss": 1.3928,
+      "step": 661
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008393684016600363,
+      "loss": 1.3675,
+      "step": 662
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008388565141953386,
+      "loss": 1.4022,
+      "step": 663
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000838343969038849,
+      "loss": 1.3849,
+      "step": 664
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008378307671853746,
+      "loss": 1.4088,
+      "step": 665
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008373169096309967,
+      "loss": 1.4598,
+      "step": 666
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000836802397373069,
+      "loss": 1.4367,
+      "step": 667
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008362872314102165,
+      "loss": 1.4767,
+      "step": 668
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008357714127423326,
+      "loss": 1.3612,
+      "step": 669
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008352549423705775,
+      "loss": 1.4153,
+      "step": 670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008347378212973767,
+      "loss": 1.4699,
+      "step": 671
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008342200505264183,
+      "loss": 1.3379,
+      "step": 672
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008337016310626515,
+      "loss": 1.4271,
+      "step": 673
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008331825639122845,
+      "loss": 1.4105,
+      "step": 674
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008326628500827827,
+      "loss": 1.4322,
+      "step": 675
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008321424905828668,
+      "loss": 1.4351,
+      "step": 676
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008316214864225102,
+      "loss": 1.4022,
+      "step": 677
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008310998386129381,
+      "loss": 1.4056,
+      "step": 678
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008305775481666245,
+      "loss": 1.3894,
+      "step": 679
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000830054616097291,
+      "loss": 1.4467,
+      "step": 680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008295310434199048,
+      "loss": 1.4128,
+      "step": 681
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008290068311506756,
+      "loss": 1.4293,
+      "step": 682
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008284819803070552,
+      "loss": 1.3965,
+      "step": 683
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008279564919077347,
+      "loss": 1.4503,
+      "step": 684
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008274303669726426,
+      "loss": 1.3933,
+      "step": 685
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008269036065229426,
+      "loss": 1.4561,
+      "step": 686
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008263762115810325,
+      "loss": 1.3904,
+      "step": 687
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008258481831705407,
+      "loss": 1.4068,
+      "step": 688
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008253195223163261,
+      "loss": 1.4162,
+      "step": 689
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008247902300444743,
+      "loss": 1.453,
+      "step": 690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000824260307382297,
+      "loss": 1.3911,
+      "step": 691
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008237297553583292,
+      "loss": 1.4086,
+      "step": 692
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008231985750023272,
+      "loss": 1.3679,
+      "step": 693
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008226667673452672,
+      "loss": 1.3765,
+      "step": 694
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000822134333419343,
+      "loss": 1.4595,
+      "step": 695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008216012742579635,
+      "loss": 1.4218,
+      "step": 696
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008210675908957513,
+      "loss": 1.3932,
+      "step": 697
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008205332843685409,
+      "loss": 1.4016,
+      "step": 698
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000819998355713376,
+      "loss": 1.3813,
+      "step": 699
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008194628059685076,
+      "loss": 1.4225,
+      "step": 700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008189266361733925,
+      "loss": 1.3598,
+      "step": 701
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008183898473686908,
+      "loss": 1.4328,
+      "step": 702
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008178524405962643,
+      "loss": 1.4406,
+      "step": 703
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008173144168991738,
+      "loss": 1.4115,
+      "step": 704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008167757773216777,
+      "loss": 1.4449,
+      "step": 705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008162365229092299,
+      "loss": 1.4387,
+      "step": 706
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008156966547084775,
+      "loss": 1.4178,
+      "step": 707
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008151561737672591,
+      "loss": 1.4056,
+      "step": 708
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008146150811346023,
+      "loss": 1.3775,
+      "step": 709
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008140733778607219,
+      "loss": 1.4193,
+      "step": 710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008135310649970183,
+      "loss": 1.4316,
+      "step": 711
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008129881435960749,
+      "loss": 1.4647,
+      "step": 712
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008124446147116559,
+      "loss": 1.3902,
+      "step": 713
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000811900479398705,
+      "loss": 1.3717,
+      "step": 714
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008113557387133427,
+      "loss": 1.4064,
+      "step": 715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008108103937128643,
+      "loss": 1.4242,
+      "step": 716
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008102644454557388,
+      "loss": 1.4058,
+      "step": 717
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008097178950016051,
+      "loss": 1.3863,
+      "step": 718
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008091707434112716,
+      "loss": 1.4493,
+      "step": 719
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008086229917467127,
+      "loss": 1.4342,
+      "step": 720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008080746410710685,
+      "loss": 1.4209,
+      "step": 721
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008075256924486412,
+      "loss": 1.4514,
+      "step": 722
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000806976146944893,
+      "loss": 1.4126,
+      "step": 723
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008064260056264459,
+      "loss": 1.3615,
+      "step": 724
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008058752695610772,
+      "loss": 1.4262,
+      "step": 725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008053239398177191,
+      "loss": 1.3708,
+      "step": 726
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008047720174664558,
+      "loss": 1.421,
+      "step": 727
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008042195035785219,
+      "loss": 1.4284,
+      "step": 728
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008036663992263001,
+      "loss": 1.4007,
+      "step": 729
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000803112705483319,
+      "loss": 1.3844,
+      "step": 730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008025584234242514,
+      "loss": 1.4003,
+      "step": 731
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008020035541249117,
+      "loss": 1.4364,
+      "step": 732
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008014480986622542,
+      "loss": 1.3974,
+      "step": 733
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000800892058114371,
+      "loss": 1.4036,
+      "step": 734
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008003354335604897,
+      "loss": 1.3658,
+      "step": 735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007997782260809712,
+      "loss": 1.4309,
+      "step": 736
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007992204367573084,
+      "loss": 1.4108,
+      "step": 737
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007986620666721229,
+      "loss": 1.4364,
+      "step": 738
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007981031169091638,
+      "loss": 1.4366,
+      "step": 739
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007975435885533049,
+      "loss": 1.4365,
+      "step": 740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000796983482690544,
+      "loss": 1.4186,
+      "step": 741
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007964228004079985,
+      "loss": 1.4005,
+      "step": 742
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007958615427939058,
+      "loss": 1.3983,
+      "step": 743
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007952997109376187,
+      "loss": 1.4214,
+      "step": 744
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007947373059296061,
+      "loss": 1.3956,
+      "step": 745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007941743288614477,
+      "loss": 1.3681,
+      "step": 746
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007936107808258348,
+      "loss": 1.3849,
+      "step": 747
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007930466629165661,
+      "loss": 1.3962,
+      "step": 748
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007924819762285469,
+      "loss": 1.4475,
+      "step": 749
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000791916721857786,
+      "loss": 1.3373,
+      "step": 750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007913509009013946,
+      "loss": 1.4257,
+      "step": 751
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000790784514457583,
+      "loss": 1.3692,
+      "step": 752
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007902175636256592,
+      "loss": 1.439,
+      "step": 753
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007896500495060268,
+      "loss": 1.3931,
+      "step": 754
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007890819732001831,
+      "loss": 1.3945,
+      "step": 755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007885133358107154,
+      "loss": 1.4132,
+      "step": 756
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007879441384413012,
+      "loss": 1.3674,
+      "step": 757
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007873743821967041,
+      "loss": 1.4443,
+      "step": 758
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007868040681827728,
+      "loss": 1.4107,
+      "step": 759
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007862331975064387,
+      "loss": 1.3893,
+      "step": 760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000785661771275713,
+      "loss": 1.4412,
+      "step": 761
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007850897905996861,
+      "loss": 1.4239,
+      "step": 762
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007845172565885237,
+      "loss": 1.4535,
+      "step": 763
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007839441703534658,
+      "loss": 1.4217,
+      "step": 764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007833705330068244,
+      "loss": 1.3956,
+      "step": 765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007827963456619811,
+      "loss": 1.4005,
+      "step": 766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007822216094333848,
+      "loss": 1.3798,
+      "step": 767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007816463254365496,
+      "loss": 1.4195,
+      "step": 768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007810704947880531,
+      "loss": 1.3803,
+      "step": 769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007804941186055341,
+      "loss": 1.3989,
+      "step": 770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007799171980076898,
+      "loss": 1.3786,
+      "step": 771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000779339734114274,
+      "loss": 1.4323,
+      "step": 772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007787617280460955,
+      "loss": 1.414,
+      "step": 773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007781831809250151,
+      "loss": 1.3919,
+      "step": 774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007776040938739434,
+      "loss": 1.3995,
+      "step": 775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007770244680168399,
+      "loss": 1.411,
+      "step": 776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007764443044787088,
+      "loss": 1.3809,
+      "step": 777
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007758636043855985,
+      "loss": 1.392,
+      "step": 778
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007752823688645989,
+      "loss": 1.4232,
+      "step": 779
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007747005990438385,
+      "loss": 1.426,
+      "step": 780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007741182960524837,
+      "loss": 1.384,
+      "step": 781
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000773535461020735,
+      "loss": 1.4351,
+      "step": 782
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007729520950798261,
+      "loss": 1.3827,
+      "step": 783
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007723681993620205,
+      "loss": 1.3988,
+      "step": 784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007717837750006106,
+      "loss": 1.4282,
+      "step": 785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007711988231299146,
+      "loss": 1.3853,
+      "step": 786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007706133448852743,
+      "loss": 1.3964,
+      "step": 787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007700273414030535,
+      "loss": 1.3619,
+      "step": 788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007694408138206354,
+      "loss": 1.4317,
+      "step": 789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007688537632764203,
+      "loss": 1.3965,
+      "step": 790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007682661909098234,
+      "loss": 1.4122,
+      "step": 791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007676780978612734,
+      "loss": 1.39,
+      "step": 792
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007670894852722086,
+      "loss": 1.4282,
+      "step": 793
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007665003542850764,
+      "loss": 1.4445,
+      "step": 794
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007659107060433299,
+      "loss": 1.3943,
+      "step": 795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007653205416914267,
+      "loss": 1.3761,
+      "step": 796
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007647298623748258,
+      "loss": 1.3998,
+      "step": 797
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007641386692399854,
+      "loss": 1.4254,
+      "step": 798
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007635469634343617,
+      "loss": 1.4058,
+      "step": 799
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007629547461064053,
+      "loss": 1.3657,
+      "step": 800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007623620184055601,
+      "loss": 1.3837,
+      "step": 801
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00076176878148226,
+      "loss": 1.395,
+      "step": 802
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007611750364879278,
+      "loss": 1.4063,
+      "step": 803
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007605807845749723,
+      "loss": 1.3831,
+      "step": 804
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007599860268967863,
+      "loss": 1.4396,
+      "step": 805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007593907646077436,
+      "loss": 1.4031,
+      "step": 806
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007587949988631981,
+      "loss": 1.4087,
+      "step": 807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000758198730819481,
+      "loss": 1.4322,
+      "step": 808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007576019616338976,
+      "loss": 1.3592,
+      "step": 809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007570046924647264,
+      "loss": 1.3754,
+      "step": 810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007564069244712165,
+      "loss": 1.3449,
+      "step": 811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007558086588135847,
+      "loss": 1.4162,
+      "step": 812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007552098966530142,
+      "loss": 1.3922,
+      "step": 813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007546106391516514,
+      "loss": 1.4167,
+      "step": 814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007540108874726046,
+      "loss": 1.4289,
+      "step": 815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007534106427799406,
+      "loss": 1.4259,
+      "step": 816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007528099062386839,
+      "loss": 1.4026,
+      "step": 817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007522086790148133,
+      "loss": 1.3782,
+      "step": 818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007516069622752595,
+      "loss": 1.4149,
+      "step": 819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007510047571879042,
+      "loss": 1.3901,
+      "step": 820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007504020649215759,
+      "loss": 1.4094,
+      "step": 821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007497988866460497,
+      "loss": 1.4035,
+      "step": 822
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007491952235320435,
+      "loss": 1.4135,
+      "step": 823
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007485910767512162,
+      "loss": 1.3811,
+      "step": 824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007479864474761652,
+      "loss": 1.4112,
+      "step": 825
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007473813368804252,
+      "loss": 1.4116,
+      "step": 826
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007467757461384646,
+      "loss": 1.4359,
+      "step": 827
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007461696764256833,
+      "loss": 1.3982,
+      "step": 828
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007455631289184116,
+      "loss": 1.395,
+      "step": 829
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007449561047939068,
+      "loss": 1.4488,
+      "step": 830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007443486052303512,
+      "loss": 1.3598,
+      "step": 831
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007437406314068502,
+      "loss": 1.4034,
+      "step": 832
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007431321845034292,
+      "loss": 1.3825,
+      "step": 833
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007425232657010325,
+      "loss": 1.4116,
+      "step": 834
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007419138761815195,
+      "loss": 1.3648,
+      "step": 835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007413040171276639,
+      "loss": 1.4124,
+      "step": 836
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007406936897231502,
+      "loss": 1.3517,
+      "step": 837
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007400828951525727,
+      "loss": 1.4003,
+      "step": 838
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007394716346014314,
+      "loss": 1.4251,
+      "step": 839
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007388599092561315,
+      "loss": 1.4202,
+      "step": 840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00073824772030398,
+      "loss": 1.4638,
+      "step": 841
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007376350689331839,
+      "loss": 1.4451,
+      "step": 842
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007370219563328476,
+      "loss": 1.3169,
+      "step": 843
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007364083836929707,
+      "loss": 1.3846,
+      "step": 844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007357943522044455,
+      "loss": 1.3522,
+      "step": 845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007351798630590556,
+      "loss": 1.429,
+      "step": 846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007345649174494719,
+      "loss": 1.4454,
+      "step": 847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000733949516569252,
+      "loss": 1.4256,
+      "step": 848
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007333336616128369,
+      "loss": 1.3625,
+      "step": 849
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007327173537755487,
+      "loss": 1.4021,
+      "step": 850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000732100594253589,
+      "loss": 1.4174,
+      "step": 851
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007314833842440354,
+      "loss": 1.4203,
+      "step": 852
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007308657249448407,
+      "loss": 1.3781,
+      "step": 853
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007302476175548288,
+      "loss": 1.4303,
+      "step": 854
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007296290632736942,
+      "loss": 1.419,
+      "step": 855
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007290100633019983,
+      "loss": 1.3896,
+      "step": 856
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007283906188411673,
+      "loss": 1.4149,
+      "step": 857
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007277707310934908,
+      "loss": 1.3744,
+      "step": 858
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007271504012621182,
+      "loss": 1.4254,
+      "step": 859
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007265296305510573,
+      "loss": 1.3755,
+      "step": 860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007259084201651714,
+      "loss": 1.3976,
+      "step": 861
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007252867713101772,
+      "loss": 1.4413,
+      "step": 862
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007246646851926424,
+      "loss": 1.4065,
+      "step": 863
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007240421630199836,
+      "loss": 1.4142,
+      "step": 864
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007234192060004636,
+      "loss": 1.3928,
+      "step": 865
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007227958153431893,
+      "loss": 1.4206,
+      "step": 866
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007221719922581088,
+      "loss": 1.4032,
+      "step": 867
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007215477379560102,
+      "loss": 1.3544,
+      "step": 868
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007209230536485183,
+      "loss": 1.3734,
+      "step": 869
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000720297940548092,
+      "loss": 1.3706,
+      "step": 870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007196723998680233,
+      "loss": 1.3445,
+      "step": 871
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007190464328224336,
+      "loss": 1.4277,
+      "step": 872
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007184200406262717,
+      "loss": 1.3834,
+      "step": 873
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007177932244953118,
+      "loss": 1.3635,
+      "step": 874
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007171659856461513,
+      "loss": 1.3979,
+      "step": 875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007165383252962073,
+      "loss": 1.3689,
+      "step": 876
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007159102446637154,
+      "loss": 1.3511,
+      "step": 877
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007152817449677272,
+      "loss": 1.3771,
+      "step": 878
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007146528274281069,
+      "loss": 1.4013,
+      "step": 879
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007140234932655306,
+      "loss": 1.4209,
+      "step": 880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007133937437014824,
+      "loss": 1.3452,
+      "step": 881
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007127635799582529,
+      "loss": 1.4122,
+      "step": 882
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007121330032589366,
+      "loss": 1.4266,
+      "step": 883
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007115020148274294,
+      "loss": 1.3838,
+      "step": 884
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007108706158884265,
+      "loss": 1.4267,
+      "step": 885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007102388076674196,
+      "loss": 1.3983,
+      "step": 886
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007096065913906951,
+      "loss": 1.4077,
+      "step": 887
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007089739682853309,
+      "loss": 1.3518,
+      "step": 888
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007083409395791953,
+      "loss": 1.364,
+      "step": 889
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007077075065009433,
+      "loss": 1.3474,
+      "step": 890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007070736702800147,
+      "loss": 1.3842,
+      "step": 891
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007064394321466319,
+      "loss": 1.3814,
+      "step": 892
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007058047933317975,
+      "loss": 1.3678,
+      "step": 893
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007051697550672917,
+      "loss": 1.3888,
+      "step": 894
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00070453431858567,
+      "loss": 1.3965,
+      "step": 895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007038984851202608,
+      "loss": 1.3933,
+      "step": 896
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007032622559051627,
+      "loss": 1.4241,
+      "step": 897
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000702625632175243,
+      "loss": 1.3888,
+      "step": 898
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007019886151661346,
+      "loss": 1.4484,
+      "step": 899
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007013512061142331,
+      "loss": 1.4372,
+      "step": 900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007007134062566958,
+      "loss": 1.3907,
+      "step": 901
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007000752168314381,
+      "loss": 1.3613,
+      "step": 902
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006994366390771316,
+      "loss": 1.4349,
+      "step": 903
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006987976742332016,
+      "loss": 1.3998,
+      "step": 904
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000698158323539825,
+      "loss": 1.3993,
+      "step": 905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006975185882379272,
+      "loss": 1.375,
+      "step": 906
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006968784695691799,
+      "loss": 1.3569,
+      "step": 907
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006962379687759997,
+      "loss": 1.3784,
+      "step": 908
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000695597087101544,
+      "loss": 1.3976,
+      "step": 909
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006949558257897102,
+      "loss": 1.409,
+      "step": 910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000694314186085132,
+      "loss": 1.3955,
+      "step": 911
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006936721692331776,
+      "loss": 1.4057,
+      "step": 912
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006930297764799476,
+      "loss": 1.3714,
+      "step": 913
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006923870090722717,
+      "loss": 1.373,
+      "step": 914
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006917438682577071,
+      "loss": 1.3759,
+      "step": 915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006911003552845356,
+      "loss": 1.3838,
+      "step": 916
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006904564714017614,
+      "loss": 1.3622,
+      "step": 917
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006898122178591086,
+      "loss": 1.3857,
+      "step": 918
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006891675959070187,
+      "loss": 1.4021,
+      "step": 919
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006885226067966483,
+      "loss": 1.3856,
+      "step": 920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006878772517798669,
+      "loss": 1.3763,
+      "step": 921
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006872315321092534,
+      "loss": 1.3887,
+      "step": 922
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006865854490380954,
+      "loss": 1.4284,
+      "step": 923
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006859390038203852,
+      "loss": 1.3752,
+      "step": 924
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006852921977108183,
+      "loss": 1.3561,
+      "step": 925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006846450319647907,
+      "loss": 1.4222,
+      "step": 926
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006839975078383961,
+      "loss": 1.3741,
+      "step": 927
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000683349626588424,
+      "loss": 1.4319,
+      "step": 928
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006827013894723572,
+      "loss": 1.4193,
+      "step": 929
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006820527977483688,
+      "loss": 1.3806,
+      "step": 930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006814038526753205,
+      "loss": 1.3438,
+      "step": 931
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006807545555127598,
+      "loss": 1.3555,
+      "step": 932
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006801049075209175,
+      "loss": 1.4047,
+      "step": 933
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000679454909960705,
+      "loss": 1.4222,
+      "step": 934
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006788045640937128,
+      "loss": 1.3595,
+      "step": 935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006781538711822072,
+      "loss": 1.4115,
+      "step": 936
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006775028324891282,
+      "loss": 1.4037,
+      "step": 937
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000676851449278086,
+      "loss": 1.3772,
+      "step": 938
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000676199722813361,
+      "loss": 1.3444,
+      "step": 939
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006755476543598986,
+      "loss": 1.3985,
+      "step": 940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006748952451833088,
+      "loss": 1.3912,
+      "step": 941
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006742424965498621,
+      "loss": 1.377,
+      "step": 942
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006735894097264889,
+      "loss": 1.3814,
+      "step": 943
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006729359859807748,
+      "loss": 1.4174,
+      "step": 944
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006722822265809605,
+      "loss": 1.3995,
+      "step": 945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006716281327959371,
+      "loss": 1.326,
+      "step": 946
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006709737058952456,
+      "loss": 1.386,
+      "step": 947
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006703189471490731,
+      "loss": 1.3972,
+      "step": 948
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006696638578282506,
+      "loss": 1.3831,
+      "step": 949
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006690084392042513,
+      "loss": 1.355,
+      "step": 950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006683526925491873,
+      "loss": 1.3994,
+      "step": 951
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006676966191358071,
+      "loss": 1.3638,
+      "step": 952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006670402202374935,
+      "loss": 1.3971,
+      "step": 953
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006663834971282614,
+      "loss": 1.4045,
+      "step": 954
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006657264510827549,
+      "loss": 1.4118,
+      "step": 955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006650690833762443,
+      "loss": 1.4066,
+      "step": 956
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006644113952846248,
+      "loss": 1.4037,
+      "step": 957
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006637533880844133,
+      "loss": 1.3622,
+      "step": 958
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006630950630527461,
+      "loss": 1.4023,
+      "step": 959
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006624364214673763,
+      "loss": 1.401,
+      "step": 960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006617774646066712,
+      "loss": 1.3493,
+      "step": 961
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006611181937496107,
+      "loss": 1.4258,
+      "step": 962
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006604586101757832,
+      "loss": 1.4189,
+      "step": 963
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006597987151653849,
+      "loss": 1.3813,
+      "step": 964
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000659138509999216,
+      "loss": 1.3449,
+      "step": 965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006584779959586789,
+      "loss": 1.3636,
+      "step": 966
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006578171743257752,
+      "loss": 1.3476,
+      "step": 967
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006571560463831036,
+      "loss": 1.3841,
+      "step": 968
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000656494613413858,
+      "loss": 1.3783,
+      "step": 969
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006558328767018233,
+      "loss": 1.3802,
+      "step": 970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006551708375313746,
+      "loss": 1.4459,
+      "step": 971
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006545084971874737,
+      "loss": 1.3559,
+      "step": 972
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006538458569556672,
+      "loss": 1.3451,
+      "step": 973
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006531829181220839,
+      "loss": 1.3795,
+      "step": 974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006525196819734313,
+      "loss": 1.3952,
+      "step": 975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006518561497969953,
+      "loss": 1.3958,
+      "step": 976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006511923228806351,
+      "loss": 1.3994,
+      "step": 977
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006505282025127827,
+      "loss": 1.3574,
+      "step": 978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006498637899824398,
+      "loss": 1.356,
+      "step": 979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006491990865791742,
+      "loss": 1.3572,
+      "step": 980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006485340935931196,
+      "loss": 1.4164,
+      "step": 981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006478688123149704,
+      "loss": 1.3541,
+      "step": 982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006472032440359816,
+      "loss": 1.3945,
+      "step": 983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006465373900479649,
+      "loss": 1.3859,
+      "step": 984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006458712516432864,
+      "loss": 1.3988,
+      "step": 985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006452048301148644,
+      "loss": 1.3809,
+      "step": 986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006445381267561661,
+      "loss": 1.3736,
+      "step": 987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000643871142861207,
+      "loss": 1.3855,
+      "step": 988
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006432038797245459,
+      "loss": 1.372,
+      "step": 989
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006425363386412843,
+      "loss": 1.3597,
+      "step": 990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006418685209070625,
+      "loss": 1.409,
+      "step": 991
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006412004278180584,
+      "loss": 1.4347,
+      "step": 992
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006405320606709844,
+      "loss": 1.4346,
+      "step": 993
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000639863420763084,
+      "loss": 1.3832,
+      "step": 994
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006391945093921308,
+      "loss": 1.3542,
+      "step": 995
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006385253278564254,
+      "loss": 1.3937,
+      "step": 996
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000637855877454792,
+      "loss": 1.3837,
+      "step": 997
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006371861594865774,
+      "loss": 1.3631,
+      "step": 998
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006365161752516476,
+      "loss": 1.392,
+      "step": 999
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006358459260503848,
+      "loss": 1.3554,
+      "step": 1000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006351754131836862,
+      "loss": 1.3746,
+      "step": 1001
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006345046379529606,
+      "loss": 1.4114,
+      "step": 1002
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006338336016601257,
+      "loss": 1.3738,
+      "step": 1003
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006331623056076062,
+      "loss": 1.3796,
+      "step": 1004
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000632490751098331,
+      "loss": 1.4249,
+      "step": 1005
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006318189394357301,
+      "loss": 1.3783,
+      "step": 1006
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006311468719237335,
+      "loss": 1.4257,
+      "step": 1007
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000630474549866767,
+      "loss": 1.3555,
+      "step": 1008
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000629801974569751,
+      "loss": 1.3737,
+      "step": 1009
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006291291473380969,
+      "loss": 1.3816,
+      "step": 1010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006284560694777055,
+      "loss": 1.3576,
+      "step": 1011
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006277827422949641,
+      "loss": 1.3877,
+      "step": 1012
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006271091670967436,
+      "loss": 1.3878,
+      "step": 1013
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006264353451903963,
+      "loss": 1.3786,
+      "step": 1014
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006257612778837535,
+      "loss": 1.3742,
+      "step": 1015
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006250869664851227,
+      "loss": 1.3891,
+      "step": 1016
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006244124123032852,
+      "loss": 1.3816,
+      "step": 1017
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000623737616647494,
+      "loss": 1.4112,
+      "step": 1018
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006230625808274697,
+      "loss": 1.3695,
+      "step": 1019
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006223873061533998,
+      "loss": 1.4113,
+      "step": 1020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006217117939359353,
+      "loss": 1.3836,
+      "step": 1021
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006210360454861883,
+      "loss": 1.4027,
+      "step": 1022
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006203600621157293,
+      "loss": 1.3692,
+      "step": 1023
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006196838451365845,
+      "loss": 1.4164,
+      "step": 1024
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006190073958612339,
+      "loss": 1.3482,
+      "step": 1025
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006183307156026085,
+      "loss": 1.3768,
+      "step": 1026
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000617653805674087,
+      "loss": 1.4166,
+      "step": 1027
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006169766673894946,
+      "loss": 1.3582,
+      "step": 1028
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006162993020630989,
+      "loss": 1.4036,
+      "step": 1029
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006156217110096088,
+      "loss": 1.397,
+      "step": 1030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006149438955441715,
+      "loss": 1.4045,
+      "step": 1031
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006142658569823689,
+      "loss": 1.3434,
+      "step": 1032
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006135875966402169,
+      "loss": 1.3955,
+      "step": 1033
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006129091158341609,
+      "loss": 1.381,
+      "step": 1034
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000612230415881075,
+      "loss": 1.4294,
+      "step": 1035
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000611551498098258,
+      "loss": 1.3747,
+      "step": 1036
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006108723638034322,
+      "loss": 1.3732,
+      "step": 1037
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006101930143147395,
+      "loss": 1.3749,
+      "step": 1038
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006095134509507398,
+      "loss": 1.3916,
+      "step": 1039
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006088336750304078,
+      "loss": 1.3708,
+      "step": 1040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006081536878731315,
+      "loss": 1.4141,
+      "step": 1041
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006074734907987079,
+      "loss": 1.3773,
+      "step": 1042
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000606793085127342,
+      "loss": 1.3933,
+      "step": 1043
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006061124721796436,
+      "loss": 1.3402,
+      "step": 1044
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006054316532766249,
+      "loss": 1.3916,
+      "step": 1045
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006047506297396976,
+      "loss": 1.3894,
+      "step": 1046
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006040694028906709,
+      "loss": 1.366,
+      "step": 1047
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000603387974051748,
+      "loss": 1.3801,
+      "step": 1048
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006027063445455249,
+      "loss": 1.3638,
+      "step": 1049
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006020245156949868,
+      "loss": 1.368,
+      "step": 1050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006013424888235053,
+      "loss": 1.3442,
+      "step": 1051
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006006602652548373,
+      "loss": 1.4111,
+      "step": 1052
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005999778463131206,
+      "loss": 1.39,
+      "step": 1053
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005992952333228728,
+      "loss": 1.3776,
+      "step": 1054
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005986124276089877,
+      "loss": 1.34,
+      "step": 1055
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005979294304967335,
+      "loss": 1.3805,
+      "step": 1056
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005972462433117497,
+      "loss": 1.369,
+      "step": 1057
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005965628673800448,
+      "loss": 1.3728,
+      "step": 1058
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005958793040279933,
+      "loss": 1.3931,
+      "step": 1059
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005951955545823342,
+      "loss": 1.3584,
+      "step": 1060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005945116203701672,
+      "loss": 1.3536,
+      "step": 1061
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005938275027189502,
+      "loss": 1.3743,
+      "step": 1062
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005931432029564981,
+      "loss": 1.4114,
+      "step": 1063
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005924587224109787,
+      "loss": 1.3857,
+      "step": 1064
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005917740624109107,
+      "loss": 1.3709,
+      "step": 1065
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005910892242851611,
+      "loss": 1.413,
+      "step": 1066
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005904042093629428,
+      "loss": 1.3704,
+      "step": 1067
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005897190189738116,
+      "loss": 1.3498,
+      "step": 1068
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005890336544476643,
+      "loss": 1.3742,
+      "step": 1069
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000588348117114735,
+      "loss": 1.3685,
+      "step": 1070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000587662408305594,
+      "loss": 1.3455,
+      "step": 1071
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005869765293511436,
+      "loss": 1.348,
+      "step": 1072
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000586290481582617,
+      "loss": 1.3614,
+      "step": 1073
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005856042663315747,
+      "loss": 1.374,
+      "step": 1074
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005849178849299026,
+      "loss": 1.3718,
+      "step": 1075
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005842313387098087,
+      "loss": 1.427,
+      "step": 1076
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000583544629003821,
+      "loss": 1.3589,
+      "step": 1077
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005828577571447848,
+      "loss": 1.3777,
+      "step": 1078
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005821707244658606,
+      "loss": 1.3679,
+      "step": 1079
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005814835323005203,
+      "loss": 1.405,
+      "step": 1080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005807961819825456,
+      "loss": 1.377,
+      "step": 1081
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005801086748460255,
+      "loss": 1.3983,
+      "step": 1082
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005794210122253531,
+      "loss": 1.3484,
+      "step": 1083
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005787331954552232,
+      "loss": 1.3673,
+      "step": 1084
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005780452258706299,
+      "loss": 1.3806,
+      "step": 1085
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005773571048068638,
+      "loss": 1.4177,
+      "step": 1086
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005766688335995099,
+      "loss": 1.4099,
+      "step": 1087
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000575980413584444,
+      "loss": 1.3384,
+      "step": 1088
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000575291846097831,
+      "loss": 1.3788,
+      "step": 1089
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005746031324761225,
+      "loss": 1.3827,
+      "step": 1090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005739142740560528,
+      "loss": 1.4016,
+      "step": 1091
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005732252721746381,
+      "loss": 1.3555,
+      "step": 1092
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005725361281691724,
+      "loss": 1.4066,
+      "step": 1093
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005718468433772262,
+      "loss": 1.4269,
+      "step": 1094
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005711574191366427,
+      "loss": 1.332,
+      "step": 1095
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005704678567855358,
+      "loss": 1.381,
+      "step": 1096
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005697781576622878,
+      "loss": 1.3876,
+      "step": 1097
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005690883231055463,
+      "loss": 1.3401,
+      "step": 1098
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005683983544542217,
+      "loss": 1.3816,
+      "step": 1099
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005677082530474845,
+      "loss": 1.4715,
+      "step": 1100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005670180202247635,
+      "loss": 1.4044,
+      "step": 1101
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005663276573257417,
+      "loss": 1.4231,
+      "step": 1102
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005656371656903556,
+      "loss": 1.3958,
+      "step": 1103
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005649465466587902,
+      "loss": 1.3418,
+      "step": 1104
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005642558015714792,
+      "loss": 1.3837,
+      "step": 1105
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005635649317690998,
+      "loss": 1.3523,
+      "step": 1106
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005628739385925723,
+      "loss": 1.3479,
+      "step": 1107
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005621828233830555,
+      "loss": 1.3213,
+      "step": 1108
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005614915874819457,
+      "loss": 1.3268,
+      "step": 1109
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005608002322308731,
+      "loss": 1.3825,
+      "step": 1110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005601087589716996,
+      "loss": 1.3717,
+      "step": 1111
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005594171690465167,
+      "loss": 1.3212,
+      "step": 1112
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005587254637976413,
+      "loss": 1.3932,
+      "step": 1113
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005580336445676151,
+      "loss": 1.3574,
+      "step": 1114
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005573417126992003,
+      "loss": 1.3653,
+      "step": 1115
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005566496695353782,
+      "loss": 1.3658,
+      "step": 1116
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000555957516419346,
+      "loss": 1.3413,
+      "step": 1117
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005552652546945144,
+      "loss": 1.3846,
+      "step": 1118
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005545728857045043,
+      "loss": 1.3745,
+      "step": 1119
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005538804107931457,
+      "loss": 1.3633,
+      "step": 1120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005531878313044737,
+      "loss": 1.4035,
+      "step": 1121
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005524951485827263,
+      "loss": 1.3601,
+      "step": 1122
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000551802363972342,
+      "loss": 1.3764,
+      "step": 1123
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005511094788179571,
+      "loss": 1.4225,
+      "step": 1124
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000550416494464403,
+      "loss": 1.3281,
+      "step": 1125
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005497234122567035,
+      "loss": 1.369,
+      "step": 1126
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005490302335400727,
+      "loss": 1.3609,
+      "step": 1127
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005483369596599117,
+      "loss": 1.3694,
+      "step": 1128
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005476435919618062,
+      "loss": 1.3559,
+      "step": 1129
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005469501317915242,
+      "loss": 1.3873,
+      "step": 1130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005462565804950133,
+      "loss": 1.3782,
+      "step": 1131
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005455629394183979,
+      "loss": 1.361,
+      "step": 1132
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005448692099079764,
+      "loss": 1.3762,
+      "step": 1133
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005441753933102189,
+      "loss": 1.377,
+      "step": 1134
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005434814909717648,
+      "loss": 1.3818,
+      "step": 1135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005427875042394199,
+      "loss": 1.3451,
+      "step": 1136
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005420934344601536,
+      "loss": 1.3493,
+      "step": 1137
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005413992829810963,
+      "loss": 1.3962,
+      "step": 1138
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005407050511495373,
+      "loss": 1.3818,
+      "step": 1139
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005400107403129222,
+      "loss": 1.3619,
+      "step": 1140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005393163518188491,
+      "loss": 1.3313,
+      "step": 1141
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005386218870150673,
+      "loss": 1.4263,
+      "step": 1142
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005379273472494741,
+      "loss": 1.3733,
+      "step": 1143
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005372327338701127,
+      "loss": 1.3326,
+      "step": 1144
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005365380482251685,
+      "loss": 1.3789,
+      "step": 1145
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005358432916629677,
+      "loss": 1.3882,
+      "step": 1146
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005351484655319738,
+      "loss": 1.3629,
+      "step": 1147
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005344535711807854,
+      "loss": 1.3812,
+      "step": 1148
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005337586099581338,
+      "loss": 1.3668,
+      "step": 1149
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005330635832128796,
+      "loss": 1.3404,
+      "step": 1150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005323684922940112,
+      "loss": 1.3632,
+      "step": 1151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000531673338550641,
+      "loss": 1.384,
+      "step": 1152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005309781233320032,
+      "loss": 1.3602,
+      "step": 1153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005302828479874522,
+      "loss": 1.3872,
+      "step": 1154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005295875138664583,
+      "loss": 1.3448,
+      "step": 1155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005288921223186063,
+      "loss": 1.3414,
+      "step": 1156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005281966746935919,
+      "loss": 1.341,
+      "step": 1157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005275011723412203,
+      "loss": 1.3423,
+      "step": 1158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005268056166114028,
+      "loss": 1.3521,
+      "step": 1159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005261100088541538,
+      "loss": 1.3659,
+      "step": 1160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005254143504195895,
+      "loss": 1.3639,
+      "step": 1161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005247186426579235,
+      "loss": 1.3859,
+      "step": 1162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000524022886919466,
+      "loss": 1.3839,
+      "step": 1163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005233270845546197,
+      "loss": 1.3443,
+      "step": 1164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005226312369138781,
+      "loss": 1.3077,
+      "step": 1165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005219353453478229,
+      "loss": 1.4235,
+      "step": 1166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005212394112071202,
+      "loss": 1.3736,
+      "step": 1167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005205434358425192,
+      "loss": 1.3686,
+      "step": 1168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005198474206048496,
+      "loss": 1.4044,
+      "step": 1169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005191513668450177,
+      "loss": 1.3838,
+      "step": 1170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005184552759140049,
+      "loss": 1.3928,
+      "step": 1171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005177591491628648,
+      "loss": 1.3652,
+      "step": 1172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005170629879427205,
+      "loss": 1.3605,
+      "step": 1173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005163667936047619,
+      "loss": 1.3502,
+      "step": 1174
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005156705675002431,
+      "loss": 1.3681,
+      "step": 1175
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005149743109804803,
+      "loss": 1.4006,
+      "step": 1176
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005142780253968481,
+      "loss": 1.3446,
+      "step": 1177
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005135817121007782,
+      "loss": 1.3494,
+      "step": 1178
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005128853724437553,
+      "loss": 1.3721,
+      "step": 1179
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005121890077773161,
+      "loss": 1.3835,
+      "step": 1180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005114926194530453,
+      "loss": 1.3681,
+      "step": 1181
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005107962088225735,
+      "loss": 1.3675,
+      "step": 1182
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005100997772375748,
+      "loss": 1.3939,
+      "step": 1183
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005094033260497642,
+      "loss": 1.3037,
+      "step": 1184
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005087068566108941,
+      "loss": 1.3764,
+      "step": 1185
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005080103702727526,
+      "loss": 1.3735,
+      "step": 1186
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005073138683871612,
+      "loss": 1.3579,
+      "step": 1187
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005066173523059708,
+      "loss": 1.3986,
+      "step": 1188
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005059208233810599,
+      "loss": 1.3637,
+      "step": 1189
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005052242829643323,
+      "loss": 1.3712,
+      "step": 1190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005045277324077141,
+      "loss": 1.343,
+      "step": 1191
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005038311730631509,
+      "loss": 1.3437,
+      "step": 1192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005031346062826052,
+      "loss": 1.313,
+      "step": 1193
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005024380334180541,
+      "loss": 1.357,
+      "step": 1194
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005017414558214869,
+      "loss": 1.3337,
+      "step": 1195
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005010448748449012,
+      "loss": 1.3379,
+      "step": 1196
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005003482918403022,
+      "loss": 1.3877,
+      "step": 1197
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004996517081596979,
+      "loss": 1.4138,
+      "step": 1198
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004989551251550989,
+      "loss": 1.3619,
+      "step": 1199
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004982585441785132,
+      "loss": 1.3595,
+      "step": 1200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004975619665819458,
+      "loss": 1.3424,
+      "step": 1201
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004968653937173949,
+      "loss": 1.4022,
+      "step": 1202
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004961688269368492,
+      "loss": 1.3681,
+      "step": 1203
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004954722675922858,
+      "loss": 1.3291,
+      "step": 1204
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004947757170356677,
+      "loss": 1.3385,
+      "step": 1205
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004940791766189401,
+      "loss": 1.3387,
+      "step": 1206
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004933826476940293,
+      "loss": 1.3881,
+      "step": 1207
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004926861316128389,
+      "loss": 1.3976,
+      "step": 1208
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004919896297272474,
+      "loss": 1.3493,
+      "step": 1209
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004912931433891061,
+      "loss": 1.3404,
+      "step": 1210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004905966739502359,
+      "loss": 1.3684,
+      "step": 1211
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004899002227624251,
+      "loss": 1.3787,
+      "step": 1212
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004892037911774266,
+      "loss": 1.376,
+      "step": 1213
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004885073805469548,
+      "loss": 1.3465,
+      "step": 1214
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048781099222268377,
+      "loss": 1.3479,
+      "step": 1215
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004871146275562447,
+      "loss": 1.3569,
+      "step": 1216
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004864182878992219,
+      "loss": 1.3668,
+      "step": 1217
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004857219746031519,
+      "loss": 1.3802,
+      "step": 1218
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048502568901951976,
+      "loss": 1.3861,
+      "step": 1219
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048432943249975685,
+      "loss": 1.3446,
+      "step": 1220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004836332063952383,
+      "loss": 1.3781,
+      "step": 1221
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048293701205727956,
+      "loss": 1.3968,
+      "step": 1222
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004822408508371352,
+      "loss": 1.3637,
+      "step": 1223
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048154472408599514,
+      "loss": 1.3678,
+      "step": 1224
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048084863315498236,
+      "loss": 1.3433,
+      "step": 1225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004801525793951503,
+      "loss": 1.4195,
+      "step": 1226
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047945656415748083,
+      "loss": 1.3719,
+      "step": 1227
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047876058879287996,
+      "loss": 1.3975,
+      "step": 1228
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004780646546521773,
+      "loss": 1.359,
+      "step": 1229
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004773687630861219,
+      "loss": 1.3269,
+      "step": 1230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047667291544538034,
+      "loss": 1.3767,
+      "step": 1231
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004759771130805343,
+      "loss": 1.3512,
+      "step": 1232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004752813573420766,
+      "loss": 1.3666,
+      "step": 1233
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004745856495804105,
+      "loss": 1.3426,
+      "step": 1234
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004738899911458462,
+      "loss": 1.3336,
+      "step": 1235
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047319438338859726,
+      "loss": 1.3752,
+      "step": 1236
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047249882765877974,
+      "loss": 1.3356,
+      "step": 1237
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004718033253064082,
+      "loss": 1.344,
+      "step": 1238
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004711078776813938,
+      "loss": 1.3459,
+      "step": 1239
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004704124861335418,
+      "loss": 1.3667,
+      "step": 1240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046971715201254787,
+      "loss": 1.3573,
+      "step": 1241
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004690218766679968,
+      "loss": 1.3154,
+      "step": 1242
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046832666144935924,
+      "loss": 1.3798,
+      "step": 1243
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046763150770598883,
+      "loss": 1.3917,
+      "step": 1244
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046693641678712027,
+      "loss": 1.3786,
+      "step": 1245
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046624139004186637,
+      "loss": 1.3525,
+      "step": 1246
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004655464288192147,
+      "loss": 1.3178,
+      "step": 1247
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046485153446802643,
+      "loss": 1.3636,
+      "step": 1248
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004641567083370324,
+      "loss": 1.3289,
+      "step": 1249
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046346195177483145,
+      "loss": 1.3811,
+      "step": 1250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046276726612988744,
+      "loss": 1.3914,
+      "step": 1251
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004620726527505259,
+      "loss": 1.3843,
+      "step": 1252
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004613781129849327,
+      "loss": 1.3458,
+      "step": 1253
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046068364818115104,
+      "loss": 1.3779,
+      "step": 1254
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004599892596870779,
+      "loss": 1.3688,
+      "step": 1255
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045929494885046257,
+      "loss": 1.3925,
+      "step": 1256
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004586007170189038,
+      "loss": 1.3535,
+      "step": 1257
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004579065655398465,
+      "loss": 1.4215,
+      "step": 1258
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004572124957605803,
+      "loss": 1.3036,
+      "step": 1259
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045651850902823525,
+      "loss": 1.3641,
+      "step": 1260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004558246066897812,
+      "loss": 1.3622,
+      "step": 1261
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045513079009202386,
+      "loss": 1.3795,
+      "step": 1262
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004544370605816022,
+      "loss": 1.4075,
+      "step": 1263
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004537434195049867,
+      "loss": 1.4056,
+      "step": 1264
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045304986820847594,
+      "loss": 1.3193,
+      "step": 1265
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000452356408038194,
+      "loss": 1.3941,
+      "step": 1266
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004516630403400885,
+      "loss": 1.326,
+      "step": 1267
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004509697664599274,
+      "loss": 1.3321,
+      "step": 1268
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045027658774329647,
+      "loss": 1.3327,
+      "step": 1269
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004495835055355972,
+      "loss": 1.3399,
+      "step": 1270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004488905211820429,
+      "loss": 1.3173,
+      "step": 1271
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044819763602765795,
+      "loss": 1.3521,
+      "step": 1272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004475048514172738,
+      "loss": 1.3681,
+      "step": 1273
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004468121686955264,
+      "loss": 1.3297,
+      "step": 1274
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044611958920685423,
+      "loss": 1.384,
+      "step": 1275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004454271142954958,
+      "loss": 1.3543,
+      "step": 1276
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044473474530548575,
+      "loss": 1.4003,
+      "step": 1277
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044404248358065416,
+      "loss": 1.382,
+      "step": 1278
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044335033046462194,
+      "loss": 1.3321,
+      "step": 1279
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004426582873007998,
+      "loss": 1.3569,
+      "step": 1280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004419663554323851,
+      "loss": 1.3209,
+      "step": 1281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044127453620235874,
+      "loss": 1.3499,
+      "step": 1282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044058283095348334,
+      "loss": 1.3638,
+      "step": 1283
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043989124102830043,
+      "loss": 1.3601,
+      "step": 1284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000439199767769127,
+      "loss": 1.324,
+      "step": 1285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004385084125180544,
+      "loss": 1.3771,
+      "step": 1286
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004378171766169446,
+      "loss": 1.3581,
+      "step": 1287
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043712606140742773,
+      "loss": 1.3712,
+      "step": 1288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004364350682309003,
+      "loss": 1.3831,
+      "step": 1289
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043574419842852086,
+      "loss": 1.3809,
+      "step": 1290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004350534533412097,
+      "loss": 1.3928,
+      "step": 1291
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004343628343096447,
+      "loss": 1.3564,
+      "step": 1292
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043367234267425824,
+      "loss": 1.339,
+      "step": 1293
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043298197977523656,
+      "loss": 1.386,
+      "step": 1294
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004322917469525155,
+      "loss": 1.3558,
+      "step": 1295
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004316016455457784,
+      "loss": 1.3292,
+      "step": 1296
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004309116768944539,
+      "loss": 1.3758,
+      "step": 1297
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043022184233771233,
+      "loss": 1.3354,
+      "step": 1298
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004295321432144643,
+      "loss": 1.3918,
+      "step": 1299
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004288425808633575,
+      "loss": 1.3981,
+      "step": 1300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042815315662277385,
+      "loss": 1.3571,
+      "step": 1301
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004274638718308275,
+      "loss": 1.338,
+      "step": 1302
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004267747278253621,
+      "loss": 1.3641,
+      "step": 1303
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004260857259439473,
+      "loss": 1.3626,
+      "step": 1304
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042539686752387764,
+      "loss": 1.3387,
+      "step": 1305
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042470815390216904,
+      "loss": 1.2966,
+      "step": 1306
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004240195864155562,
+      "loss": 1.3284,
+      "step": 1307
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004233311664004903,
+      "loss": 1.3496,
+      "step": 1308
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004226428951931362,
+      "loss": 1.3651,
+      "step": 1309
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042195477412937014,
+      "loss": 1.3498,
+      "step": 1310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042126680454477694,
+      "loss": 1.3512,
+      "step": 1311
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042057898777464697,
+      "loss": 1.3902,
+      "step": 1312
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004198913251539744,
+      "loss": 1.3648,
+      "step": 1313
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00041920381801745444,
+      "loss": 1.3407,
+      "step": 1314
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004185164676994798,
+      "loss": 1.4013,
+      "step": 1315
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041782927553413965,
+      "loss": 1.3388,
+      "step": 1316
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004171422428552152,
+      "loss": 1.3866,
+      "step": 1317
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004164553709961791,
+      "loss": 1.363,
+      "step": 1318
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004157686612901915,
+      "loss": 1.3691,
+      "step": 1319
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004150821150700975,
+      "loss": 1.3625,
+      "step": 1320
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041439573366842524,
+      "loss": 1.3373,
+      "step": 1321
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041370951841738325,
+      "loss": 1.395,
+      "step": 1322
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004130234706488566,
+      "loss": 1.3456,
+      "step": 1323
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004123375916944061,
+      "loss": 1.3017,
+      "step": 1324
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041165188288526513,
+      "loss": 1.377,
+      "step": 1325
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004109663455523358,
+      "loss": 1.3684,
+      "step": 1326
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041028098102618846,
+      "loss": 1.3491,
+      "step": 1327
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004095957906370572,
+      "loss": 1.3364,
+      "step": 1328
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004089107757148389,
+      "loss": 1.3804,
+      "step": 1329
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040822593758908944,
+      "loss": 1.3917,
+      "step": 1330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040754127758902136,
+      "loss": 1.3547,
+      "step": 1331
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004068567970435019,
+      "loss": 1.4136,
+      "step": 1332
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004061724972810499,
+      "loss": 1.3748,
+      "step": 1333
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040548837962983296,
+      "loss": 1.3526,
+      "step": 1334
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040480444541766573,
+      "loss": 1.3857,
+      "step": 1335
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004041206959720068,
+      "loss": 1.3964,
+      "step": 1336
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004034371326199554,
+      "loss": 1.3238,
+      "step": 1337
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004027537566882504,
+      "loss": 1.3736,
+      "step": 1338
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040207056950326657,
+      "loss": 1.3653,
+      "step": 1339
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004013875723910123,
+      "loss": 1.3355,
+      "step": 1340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040070476667712743,
+      "loss": 1.3157,
+      "step": 1341
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004000221536868795,
+      "loss": 1.3536,
+      "step": 1342
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039933973474516273,
+      "loss": 1.3519,
+      "step": 1343
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003986575111764948,
+      "loss": 1.3412,
+      "step": 1344
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039797548430501337,
+      "loss": 1.3394,
+      "step": 1345
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039729365545447515,
+      "loss": 1.3783,
+      "step": 1346
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039661202594825203,
+      "loss": 1.3898,
+      "step": 1347
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003959305971093292,
+      "loss": 1.3485,
+      "step": 1348
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003952493702603025,
+      "loss": 1.3434,
+      "step": 1349
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039456834672337514,
+      "loss": 1.3505,
+      "step": 1350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003938875278203564,
+      "loss": 1.4175,
+      "step": 1351
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039320691487265816,
+      "loss": 1.3668,
+      "step": 1352
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003925265092012922,
+      "loss": 1.3489,
+      "step": 1353
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039184631212686857,
+      "loss": 1.3413,
+      "step": 1354
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003911663249695923,
+      "loss": 1.3904,
+      "step": 1355
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003904865490492603,
+      "loss": 1.3338,
+      "step": 1356
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003898069856852607,
+      "loss": 1.4014,
+      "step": 1357
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003891276361965679,
+      "loss": 1.3769,
+      "step": 1358
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038844850190174196,
+      "loss": 1.3471,
+      "step": 1359
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038776958411892526,
+      "loss": 1.3753,
+      "step": 1360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038709088416583927,
+      "loss": 1.3574,
+      "step": 1361
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003864124033597831,
+      "loss": 1.3056,
+      "step": 1362
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003857341430176311,
+      "loss": 1.3344,
+      "step": 1363
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003850561044558286,
+      "loss": 1.3514,
+      "step": 1364
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003843782889903912,
+      "loss": 1.3429,
+      "step": 1365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003837006979369012,
+      "loss": 1.3635,
+      "step": 1366
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003830233326105055,
+      "loss": 1.3222,
+      "step": 1367
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038234619432591315,
+      "loss": 1.3617,
+      "step": 1368
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038166928439739166,
+      "loss": 1.2886,
+      "step": 1369
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003809926041387661,
+      "loss": 1.3318,
+      "step": 1370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038031615486341567,
+      "loss": 1.3239,
+      "step": 1371
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037963993788427087,
+      "loss": 1.3729,
+      "step": 1372
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037896395451381164,
+      "loss": 1.4059,
+      "step": 1373
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003782882060640648,
+      "loss": 1.3737,
+      "step": 1374
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003776126938466003,
+      "loss": 1.3375,
+      "step": 1375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003769374191725305,
+      "loss": 1.3465,
+      "step": 1376
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037626238335250614,
+      "loss": 1.3398,
+      "step": 1377
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003755875876967146,
+      "loss": 1.3558,
+      "step": 1378
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003749130335148775,
+      "loss": 1.3217,
+      "step": 1379
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003742387221162466,
+      "loss": 1.3346,
+      "step": 1380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003735646548096038,
+      "loss": 1.3489,
+      "step": 1381
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037289083290325663,
+      "loss": 1.3312,
+      "step": 1382
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000372217257705036,
+      "loss": 1.3598,
+      "step": 1383
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037154393052229453,
+      "loss": 1.3325,
+      "step": 1384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037087085266190325,
+      "loss": 1.3364,
+      "step": 1385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003701980254302492,
+      "loss": 1.3334,
+      "step": 1386
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036952545013323326,
+      "loss": 1.3447,
+      "step": 1387
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003688531280762667,
+      "loss": 1.3441,
+      "step": 1388
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036818106056427,
+      "loss": 1.3697,
+      "step": 1389
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036750924890166926,
+      "loss": 1.3758,
+      "step": 1390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003668376943923938,
+      "loss": 1.3486,
+      "step": 1391
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036616639833987423,
+      "loss": 1.4006,
+      "step": 1392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036549536204703956,
+      "loss": 1.3823,
+      "step": 1393
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036482458681631383,
+      "loss": 1.3966,
+      "step": 1394
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036415407394961536,
+      "loss": 1.3508,
+      "step": 1395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003634838247483526,
+      "loss": 1.3552,
+      "step": 1396
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003628138405134226,
+      "loss": 1.3398,
+      "step": 1397
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003621441225452082,
+      "loss": 1.3536,
+      "step": 1398
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003614746721435747,
+      "loss": 1.3696,
+      "step": 1399
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003608054906078691,
+      "loss": 1.386,
+      "step": 1400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036013657923691613,
+      "loss": 1.3649,
+      "step": 1401
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003594679393290157,
+      "loss": 1.3309,
+      "step": 1402
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035879957218194145,
+      "loss": 1.3588,
+      "step": 1403
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003581314790929376,
+      "loss": 1.34,
+      "step": 1404
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035746366135871587,
+      "loss": 1.384,
+      "step": 1405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035679612027545425,
+      "loss": 1.361,
+      "step": 1406
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035612885713879315,
+      "loss": 1.4044,
+      "step": 1407
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035546187324383396,
+      "loss": 1.3564,
+      "step": 1408
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035479516988513594,
+      "loss": 1.3766,
+      "step": 1409
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003541287483567137,
+      "loss": 1.3728,
+      "step": 1410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035346260995203503,
+      "loss": 1.3614,
+      "step": 1411
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035279675596401844,
+      "loss": 1.3561,
+      "step": 1412
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003521311876850297,
+      "loss": 1.3665,
+      "step": 1413
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035146590640688057,
+      "loss": 1.3556,
+      "step": 1414
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035080091342082586,
+      "loss": 1.3356,
+      "step": 1415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035013621001756033,
+      "loss": 1.3339,
+      "step": 1416
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003494717974872173,
+      "loss": 1.301,
+      "step": 1417
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034880767711936494,
+      "loss": 1.3343,
+      "step": 1418
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034814385020300476,
+      "loss": 1.337,
+      "step": 1419
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003474803180265688,
+      "loss": 1.3522,
+      "step": 1420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034681708187791625,
+      "loss": 1.3523,
+      "step": 1421
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003461541430443327,
+      "loss": 1.3963,
+      "step": 1422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034549150281252633,
+      "loss": 1.3188,
+      "step": 1423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003448291624686254,
+      "loss": 1.3474,
+      "step": 1424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034416712329817683,
+      "loss": 1.3497,
+      "step": 1425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003435053865861421,
+      "loss": 1.2956,
+      "step": 1426
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034284395361689634,
+      "loss": 1.3121,
+      "step": 1427
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000342182825674225,
+      "loss": 1.3952,
+      "step": 1428
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003415220040413213,
+      "loss": 1.3117,
+      "step": 1429
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000340861490000784,
+      "loss": 1.3669,
+      "step": 1430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00034020128483461525,
+      "loss": 1.3079,
+      "step": 1431
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003395413898242169,
+      "loss": 1.3336,
+      "step": 1432
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033888180625038934,
+      "loss": 1.3875,
+      "step": 1433
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003382225353933288,
+      "loss": 1.3475,
+      "step": 1434
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003375635785326238,
+      "loss": 1.3214,
+      "step": 1435
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000336904936947254,
+      "loss": 1.3405,
+      "step": 1436
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003362466119155867,
+      "loss": 1.3137,
+      "step": 1437
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003355886047153752,
+      "loss": 1.3466,
+      "step": 1438
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033493091662375584,
+      "loss": 1.3597,
+      "step": 1439
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033427354891724527,
+      "loss": 1.3616,
+      "step": 1440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003336165028717385,
+      "loss": 1.4011,
+      "step": 1441
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033295977976250654,
+      "loss": 1.3753,
+      "step": 1442
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000332303380864193,
+      "loss": 1.3383,
+      "step": 1443
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003316473074508128,
+      "loss": 1.3354,
+      "step": 1444
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003309915607957487,
+      "loss": 1.346,
+      "step": 1445
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003303361421717494,
+      "loss": 1.3443,
+      "step": 1446
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003296810528509271,
+      "loss": 1.3514,
+      "step": 1447
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003290262941047545,
+      "loss": 1.3163,
+      "step": 1448
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003283718672040629,
+      "loss": 1.3475,
+      "step": 1449
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032771777341903974,
+      "loss": 1.3893,
+      "step": 1450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032706401401922525,
+      "loss": 1.3562,
+      "step": 1451
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032641059027351113,
+      "loss": 1.3383,
+      "step": 1452
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003257575034501379,
+      "loss": 1.3059,
+      "step": 1453
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032510475481669135,
+      "loss": 1.3135,
+      "step": 1454
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032445234564010153,
+      "loss": 1.3728,
+      "step": 1455
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003238002771866391,
+      "loss": 1.3108,
+      "step": 1456
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032314855072191397,
+      "loss": 1.3578,
+      "step": 1457
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032249716751087213,
+      "loss": 1.3797,
+      "step": 1458
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032184612881779273,
+      "loss": 1.3594,
+      "step": 1459
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032119543590628706,
+      "loss": 1.366,
+      "step": 1460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032054509003929507,
+      "loss": 1.3325,
+      "step": 1461
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031989509247908266,
+      "loss": 1.39,
+      "step": 1462
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031924544448724016,
+      "loss": 1.3349,
+      "step": 1463
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031859614732467957,
+      "loss": 1.3657,
+      "step": 1464
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003179472022516313,
+      "loss": 1.3595,
+      "step": 1465
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031729861052764296,
+      "loss": 1.3614,
+      "step": 1466
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031665037341157605,
+      "loss": 1.3626,
+      "step": 1467
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003160024921616039,
+      "loss": 1.3785,
+      "step": 1468
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031535496803520945,
+      "loss": 1.395,
+      "step": 1469
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003147078022891817,
+      "loss": 1.3314,
+      "step": 1470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003140609961796147,
+      "loss": 1.3396,
+      "step": 1471
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031341455096190477,
+      "loss": 1.3311,
+      "step": 1472
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031276846789074665,
+      "loss": 1.3109,
+      "step": 1473
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031212274822013333,
+      "loss": 1.3847,
+      "step": 1474
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031147739320335166,
+      "loss": 1.3506,
+      "step": 1475
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003108324040929812,
+      "loss": 1.3246,
+      "step": 1476
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003101877821408915,
+      "loss": 1.3573,
+      "step": 1477
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030954352859823865,
+      "loss": 1.3119,
+      "step": 1478
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003088996447154644,
+      "loss": 1.357,
+      "step": 1479
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000308256131742293,
+      "loss": 1.3555,
+      "step": 1480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003076129909277283,
+      "loss": 1.3401,
+      "step": 1481
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030697022352005245,
+      "loss": 1.3297,
+      "step": 1482
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003063278307668225,
+      "loss": 1.3239,
+      "step": 1483
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030568581391486816,
+      "loss": 1.3106,
+      "step": 1484
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030504417421028994,
+      "loss": 1.3389,
+      "step": 1485
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000304402912898456,
+      "loss": 1.3272,
+      "step": 1486
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003037620312240004,
+      "loss": 1.3207,
+      "step": 1487
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030312153043082025,
+      "loss": 1.341,
+      "step": 1488
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030248141176207296,
+      "loss": 1.3532,
+      "step": 1489
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030184167646017494,
+      "loss": 1.3164,
+      "step": 1490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003012023257667984,
+      "loss": 1.3735,
+      "step": 1491
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003005633609228685,
+      "loss": 1.3466,
+      "step": 1492
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002999247831685621,
+      "loss": 1.3711,
+      "step": 1493
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029928659374330437,
+      "loss": 1.3814,
+      "step": 1494
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029864879388576696,
+      "loss": 1.3671,
+      "step": 1495
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002980113848338657,
+      "loss": 1.3581,
+      "step": 1496
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002973743678247571,
+      "loss": 1.3689,
+      "step": 1497
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002967377440948374,
+      "loss": 1.3524,
+      "step": 1498
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002961015148797395,
+      "loss": 1.3167,
+      "step": 1499
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029546568141433004,
+      "loss": 1.3675,
+      "step": 1500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002948302449327083,
+      "loss": 1.3843,
+      "step": 1501
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029419520666820265,
+      "loss": 1.3541,
+      "step": 1502
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002935605678533682,
+      "loss": 1.3986,
+      "step": 1503
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002929263297199855,
+      "loss": 1.3705,
+      "step": 1504
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002922924934990568,
+      "loss": 1.3407,
+      "step": 1505
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002916590604208046,
+      "loss": 1.3662,
+      "step": 1506
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029102603171466914,
+      "loss": 1.3255,
+      "step": 1507
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002903934086093051,
+      "loss": 1.3614,
+      "step": 1508
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002897611923325804,
+      "loss": 1.3688,
+      "step": 1509
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028912938411157366,
+      "loss": 1.3421,
+      "step": 1510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028849798517257066,
+      "loss": 1.3648,
+      "step": 1511
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028786699674106345,
+      "loss": 1.3756,
+      "step": 1512
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002872364200417472,
+      "loss": 1.3786,
+      "step": 1513
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028660625629851774,
+      "loss": 1.3681,
+      "step": 1514
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002859765067344695,
+      "loss": 1.3352,
+      "step": 1515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002853471725718931,
+      "loss": 1.3123,
+      "step": 1516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002847182550322729,
+      "loss": 1.3345,
+      "step": 1517
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028408975533628474,
+      "loss": 1.3455,
+      "step": 1518
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002834616747037928,
+      "loss": 1.3502,
+      "step": 1519
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002828340143538488,
+      "loss": 1.3495,
+      "step": 1520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002822067755046883,
+      "loss": 1.3542,
+      "step": 1521
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028157995937372847,
+      "loss": 1.3719,
+      "step": 1522
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002809535671775667,
+      "loss": 1.3066,
+      "step": 1523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002803276001319769,
+      "loss": 1.2758,
+      "step": 1524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027970205945190786,
+      "loss": 1.3673,
+      "step": 1525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027907694635148183,
+      "loss": 1.3485,
+      "step": 1526
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027845226204398975,
+      "loss": 1.3529,
+      "step": 1527
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002778280077418911,
+      "loss": 1.3332,
+      "step": 1528
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002772041846568109,
+      "loss": 1.3259,
+      "step": 1529
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027658079399953643,
+      "loss": 1.3444,
+      "step": 1530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002759578369800164,
+      "loss": 1.3143,
+      "step": 1531
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002753353148073578,
+      "loss": 1.3567,
+      "step": 1532
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000274713228689823,
+      "loss": 1.3577,
+      "step": 1533
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002740915798348287,
+      "loss": 1.3389,
+      "step": 1534
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027347036944894273,
+      "loss": 1.3669,
+      "step": 1535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027284959873788173,
+      "loss": 1.307,
+      "step": 1536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002722292689065093,
+      "loss": 1.3583,
+      "step": 1537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002716093811588327,
+      "loss": 1.3335,
+      "step": 1538
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002709899366980018,
+      "loss": 1.3434,
+      "step": 1539
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002703709367263059,
+      "loss": 1.3196,
+      "step": 1540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002697523824451712,
+      "loss": 1.3387,
+      "step": 1541
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002691342750551594,
+      "loss": 1.3276,
+      "step": 1542
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002685166157559648,
+      "loss": 1.3179,
+      "step": 1543
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000267899405746411,
+      "loss": 1.3232,
+      "step": 1544
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026728264622445137,
+      "loss": 1.365,
+      "step": 1545
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026666633838716316,
+      "loss": 1.3689,
+      "step": 1546
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026605048343074795,
+      "loss": 1.3387,
+      "step": 1547
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026543508255052816,
+      "loss": 1.3021,
+      "step": 1548
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002648201369409445,
+      "loss": 1.3449,
+      "step": 1549
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026420564779555444,
+      "loss": 1.3224,
+      "step": 1550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026359161630702953,
+      "loss": 1.3251,
+      "step": 1551
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002629780436671525,
+      "loss": 1.3478,
+      "step": 1552
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026236493106681616,
+      "loss": 1.3369,
+      "step": 1553
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026175227969602,
+      "loss": 1.33,
+      "step": 1554
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026114009074386846,
+      "loss": 1.337,
+      "step": 1555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002605283653985687,
+      "loss": 1.3153,
+      "step": 1556
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002599171048474274,
+      "loss": 1.3541,
+      "step": 1557
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002593063102768497,
+      "loss": 1.3094,
+      "step": 1558
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025869598287233634,
+      "loss": 1.3649,
+      "step": 1559
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002580861238184806,
+      "loss": 1.3035,
+      "step": 1560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002574767342989676,
+      "loss": 1.3211,
+      "step": 1561
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002568678154965707,
+      "loss": 1.3697,
+      "step": 1562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025625936859314976,
+      "loss": 1.3912,
+      "step": 1563
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025565139476964885,
+      "loss": 1.3477,
+      "step": 1564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002550438952060932,
+      "loss": 1.3634,
+      "step": 1565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025443687108158835,
+      "loss": 1.3135,
+      "step": 1566
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002538303235743168,
+      "loss": 1.3652,
+      "step": 1567
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002532242538615356,
+      "loss": 1.3583,
+      "step": 1568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002526186631195747,
+      "loss": 1.3329,
+      "step": 1569
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025201355252383483,
+      "loss": 1.2922,
+      "step": 1570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000251408923248784,
+      "loss": 1.3958,
+      "step": 1571
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002508047764679566,
+      "loss": 1.3593,
+      "step": 1572
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002502011133539502,
+      "loss": 1.3223,
+      "step": 1573
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000249597935078424,
+      "loss": 1.3046,
+      "step": 1574
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000248995242812096,
+      "loss": 1.3443,
+      "step": 1575
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002483930377247405,
+      "loss": 1.367,
+      "step": 1576
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024779132098518675,
+      "loss": 1.351,
+      "step": 1577
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024719009376131614,
+      "loss": 1.3417,
+      "step": 1578
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024658935722005943,
+      "loss": 1.3551,
+      "step": 1579
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024598911252739556,
+      "loss": 1.3308,
+      "step": 1580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002453893608483486,
+      "loss": 1.3451,
+      "step": 1581
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024479010334698575,
+      "loss": 1.3874,
+      "step": 1582
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002441913411864153,
+      "loss": 1.3174,
+      "step": 1583
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024359307552878357,
+      "loss": 1.3357,
+      "step": 1584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024299530753527353,
+      "loss": 1.3278,
+      "step": 1585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002423980383661026,
+      "loss": 1.2737,
+      "step": 1586
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024180126918051909,
+      "loss": 1.3322,
+      "step": 1587
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024120500113680176,
+      "loss": 1.3469,
+      "step": 1588
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024060923539225655,
+      "loss": 1.366,
+      "step": 1589
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002400139731032139,
+      "loss": 1.3691,
+      "step": 1590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023941921542502764,
+      "loss": 1.3635,
+      "step": 1591
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023882496351207216,
+      "loss": 1.3498,
+      "step": 1592
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023823121851774,
+      "loss": 1.3336,
+      "step": 1593
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023763798159444005,
+      "loss": 1.3809,
+      "step": 1594
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023704525389359472,
+      "loss": 1.3236,
+      "step": 1595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023645303656563826,
+      "loss": 1.3199,
+      "step": 1596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023586133076001464,
+      "loss": 1.3812,
+      "step": 1597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023527013762517435,
+      "loss": 1.2782,
+      "step": 1598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023467945830857328,
+      "loss": 1.2865,
+      "step": 1599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002340892939566701,
+      "loss": 1.3421,
+      "step": 1600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023349964571492366,
+      "loss": 1.3918,
+      "step": 1601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023291051472779158,
+      "loss": 1.3274,
+      "step": 1602
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023232190213872672,
+      "loss": 1.3497,
+      "step": 1603
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002317338090901765,
+      "loss": 1.325,
+      "step": 1604
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023114623672357992,
+      "loss": 1.3099,
+      "step": 1605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023055918617936473,
+      "loss": 1.3705,
+      "step": 1606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002299726585969466,
+      "loss": 1.3499,
+      "step": 1607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022938665511472594,
+      "loss": 1.3532,
+      "step": 1608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002288011768700856,
+      "loss": 1.3263,
+      "step": 1609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022821622499938948,
+      "loss": 1.3238,
+      "step": 1610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002276318006379795,
+      "loss": 1.3547,
+      "step": 1611
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022704790492017392,
+      "loss": 1.3281,
+      "step": 1612
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000226464538979265,
+      "loss": 1.3617,
+      "step": 1613
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022588170394751634,
+      "loss": 1.3625,
+      "step": 1614
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002252994009561614,
+      "loss": 1.3225,
+      "step": 1615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022471763113540133,
+      "loss": 1.4172,
+      "step": 1616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002241363956144016,
+      "loss": 1.331,
+      "step": 1617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002235556955212913,
+      "loss": 1.3307,
+      "step": 1618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022297553198316017,
+      "loss": 1.372,
+      "step": 1619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022239590612605642,
+      "loss": 1.3205,
+      "step": 1620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022181681907498503,
+      "loss": 1.3373,
+      "step": 1621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002212382719539045,
+      "loss": 1.3416,
+      "step": 1622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022066026588572597,
+      "loss": 1.3408,
+      "step": 1623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002200828019923104,
+      "loss": 1.3563,
+      "step": 1624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021950588139446598,
+      "loss": 1.3243,
+      "step": 1625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021892950521194693,
+      "loss": 1.3434,
+      "step": 1626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021835367456345074,
+      "loss": 1.3395,
+      "step": 1627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021777839056661552,
+      "loss": 1.3674,
+      "step": 1628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021720365433801903,
+      "loss": 1.3585,
+      "step": 1629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002166294669931756,
+      "loss": 1.3985,
+      "step": 1630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002160558296465342,
+      "loss": 1.3059,
+      "step": 1631
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002154827434114765,
+      "loss": 1.2744,
+      "step": 1632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021491020940031402,
+      "loss": 1.3485,
+      "step": 1633
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021433822872428694,
+      "loss": 1.3318,
+      "step": 1634
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002137668024935615,
+      "loss": 1.3423,
+      "step": 1635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021319593181722723,
+      "loss": 1.3527,
+      "step": 1636
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021262561780329597,
+      "loss": 1.3534,
+      "step": 1637
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021205586155869887,
+      "loss": 1.3434,
+      "step": 1638
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002114866641892846,
+      "loss": 1.3231,
+      "step": 1639
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002109180267998169,
+      "loss": 1.3019,
+      "step": 1640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021034995049397316,
+      "loss": 1.3433,
+      "step": 1641
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020978243637434092,
+      "loss": 1.3363,
+      "step": 1642
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020921548554241732,
+      "loss": 1.3401,
+      "step": 1643
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020864909909860558,
+      "loss": 1.2983,
+      "step": 1644
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000208083278142214,
+      "loss": 1.3451,
+      "step": 1645
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020751802377145335,
+      "loss": 1.3926,
+      "step": 1646
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000206953337083434,
+      "loss": 1.3228,
+      "step": 1647
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002063892191741653,
+      "loss": 1.3501,
+      "step": 1648
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020582567113855234,
+      "loss": 1.2894,
+      "step": 1649
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020526269407039393,
+      "loss": 1.3439,
+      "step": 1650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020470028906238125,
+      "loss": 1.3468,
+      "step": 1651
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020413845720609436,
+      "loss": 1.3369,
+      "step": 1652
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002035771995920015,
+      "loss": 1.3497,
+      "step": 1653
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020301651730945624,
+      "loss": 1.346,
+      "step": 1654
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020245641144669513,
+      "loss": 1.3705,
+      "step": 1655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020189688309083642,
+      "loss": 1.261,
+      "step": 1656
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020133793332787714,
+      "loss": 1.2669,
+      "step": 1657
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020077956324269158,
+      "loss": 1.3685,
+      "step": 1658
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020022177391902863,
+      "loss": 1.2741,
+      "step": 1659
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019966456643951043,
+      "loss": 1.3104,
+      "step": 1660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019910794188562902,
+      "loss": 1.3184,
+      "step": 1661
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001985519013377459,
+      "loss": 1.3568,
+      "step": 1662
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001979964458750884,
+      "loss": 1.3233,
+      "step": 1663
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019744157657574868,
+      "loss": 1.3307,
+      "step": 1664
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019688729451668114,
+      "loss": 1.2968,
+      "step": 1665
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019633360077370005,
+      "loss": 1.3481,
+      "step": 1666
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019578049642147817,
+      "loss": 1.3659,
+      "step": 1667
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019522798253354424,
+      "loss": 1.3099,
+      "step": 1668
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001946760601822809,
+      "loss": 1.3444,
+      "step": 1669
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001941247304389227,
+      "loss": 1.3266,
+      "step": 1670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019357399437355417,
+      "loss": 1.3425,
+      "step": 1671
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001930238530551069,
+      "loss": 1.3679,
+      "step": 1672
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001924743075513591,
+      "loss": 1.3387,
+      "step": 1673
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019192535892893154,
+      "loss": 1.3512,
+      "step": 1674
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001913770082532873,
+      "loss": 1.3238,
+      "step": 1675
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001908292565887285,
+      "loss": 1.3391,
+      "step": 1676
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019028210499839483,
+      "loss": 1.3376,
+      "step": 1677
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001897355545442611,
+      "loss": 1.3394,
+      "step": 1678
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001891896062871356,
+      "loss": 1.3392,
+      "step": 1679
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018864426128665734,
+      "loss": 1.3603,
+      "step": 1680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001880995206012952,
+      "loss": 1.3199,
+      "step": 1681
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001875553852883442,
+      "loss": 1.3411,
+      "step": 1682
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001870118564039252,
+      "loss": 1.3176,
+      "step": 1683
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018646893500298185,
+      "loss": 1.3275,
+      "step": 1684
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018592662213927824,
+      "loss": 1.3133,
+      "step": 1685
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018538491886539789,
+      "loss": 1.3228,
+      "step": 1686
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000184843826232741,
+      "loss": 1.3082,
+      "step": 1687
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018430334529152248,
+      "loss": 1.3312,
+      "step": 1688
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018376347709077008,
+      "loss": 1.3584,
+      "step": 1689
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018322422267832246,
+      "loss": 1.3333,
+      "step": 1690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018268558310082633,
+      "loss": 1.3413,
+      "step": 1691
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018214755940373594,
+      "loss": 1.2949,
+      "step": 1692
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018161015263130925,
+      "loss": 1.2698,
+      "step": 1693
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018107336382660756,
+      "loss": 1.3733,
+      "step": 1694
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018053719403149243,
+      "loss": 1.3649,
+      "step": 1695
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018000164428662397,
+      "loss": 1.294,
+      "step": 1696
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017946671563145894,
+      "loss": 1.3375,
+      "step": 1697
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017893240910424874,
+      "loss": 1.3529,
+      "step": 1698
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017839872574203664,
+      "loss": 1.317,
+      "step": 1699
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001778656665806572,
+      "loss": 1.3161,
+      "step": 1700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017733323265473289,
+      "loss": 1.3283,
+      "step": 1701
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017680142499767294,
+      "loss": 1.3537,
+      "step": 1702
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017627024464167106,
+      "loss": 1.3272,
+      "step": 1703
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001757396926177031,
+      "loss": 1.3182,
+      "step": 1704
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017520976995552578,
+      "loss": 1.3417,
+      "step": 1705
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000174680477683674,
+      "loss": 1.3674,
+      "step": 1706
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001741518168294593,
+      "loss": 1.3585,
+      "step": 1707
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001736237884189676,
+      "loss": 1.3579,
+      "step": 1708
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017309639347705747,
+      "loss": 1.3297,
+      "step": 1709
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001725696330273575,
+      "loss": 1.278,
+      "step": 1710
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017204350809226542,
+      "loss": 1.3088,
+      "step": 1711
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001715180196929449,
+      "loss": 1.3087,
+      "step": 1712
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017099316884932448,
+      "loss": 1.3239,
+      "step": 1713
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017046895658009525,
+      "loss": 1.3538,
+      "step": 1714
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016994538390270887,
+      "loss": 1.3181,
+      "step": 1715
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016942245183337545,
+      "loss": 1.3272,
+      "step": 1716
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016890016138706204,
+      "loss": 1.3049,
+      "step": 1717
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016837851357748984,
+      "loss": 1.3704,
+      "step": 1718
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001678575094171333,
+      "loss": 1.3154,
+      "step": 1719
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016733714991721738,
+      "loss": 1.3446,
+      "step": 1720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001668174360877156,
+      "loss": 1.3325,
+      "step": 1721
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016629836893734874,
+      "loss": 1.2697,
+      "step": 1722
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016577994947358183,
+      "loss": 1.3214,
+      "step": 1723
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001652621787026234,
+      "loss": 1.3126,
+      "step": 1724
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001647450576294225,
+      "loss": 1.3476,
+      "step": 1725
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016422858725766748,
+      "loss": 1.321,
+      "step": 1726
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016371276858978345,
+      "loss": 1.3128,
+      "step": 1727
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016319760262693107,
+      "loss": 1.3844,
+      "step": 1728
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001626830903690034,
+      "loss": 1.3697,
+      "step": 1729
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016216923281462553,
+      "loss": 1.3296,
+      "step": 1730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016165603096115105,
+      "loss": 1.2708,
+      "step": 1731
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016114348580466154,
+      "loss": 1.3271,
+      "step": 1732
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001606315983399637,
+      "loss": 1.2477,
+      "step": 1733
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016012036956058773,
+      "loss": 1.316,
+      "step": 1734
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015960980045878536,
+      "loss": 1.3053,
+      "step": 1735
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001590998920255282,
+      "loss": 1.3588,
+      "step": 1736
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015859064525050492,
+      "loss": 1.3335,
+      "step": 1737
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015808206112212064,
+      "loss": 1.334,
+      "step": 1738
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001575741406274942,
+      "loss": 1.2943,
+      "step": 1739
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015706688475245585,
+      "loss": 1.3225,
+      "step": 1740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015656029448154668,
+      "loss": 1.3272,
+      "step": 1741
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001560543707980152,
+      "loss": 1.3352,
+      "step": 1742
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015554911468381634,
+      "loss": 1.283,
+      "step": 1743
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015504452711960997,
+      "loss": 1.3228,
+      "step": 1744
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015454060908475732,
+      "loss": 1.353,
+      "step": 1745
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015403736155732072,
+      "loss": 1.3756,
+      "step": 1746
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015353478551406125,
+      "loss": 1.3305,
+      "step": 1747
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000153032881930436,
+      "loss": 1.3315,
+      "step": 1748
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015253165178059775,
+      "loss": 1.3277,
+      "step": 1749
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015203109603739135,
+      "loss": 1.3191,
+      "step": 1750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015153121567235335,
+      "loss": 1.3316,
+      "step": 1751
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001510320116557091,
+      "loss": 1.3542,
+      "step": 1752
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015053348495637137,
+      "loss": 1.355,
+      "step": 1753
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015003563654193824,
+      "loss": 1.3435,
+      "step": 1754
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014953846737869153,
+      "loss": 1.3423,
+      "step": 1755
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014904197843159406,
+      "loss": 1.3409,
+      "step": 1756
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014854617066428899,
+      "loss": 1.4267,
+      "step": 1757
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014805104503909743,
+      "loss": 1.3351,
+      "step": 1758
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014755660251701596,
+      "loss": 1.3102,
+      "step": 1759
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014706284405771586,
+      "loss": 1.2909,
+      "step": 1760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014656977061954025,
+      "loss": 1.3389,
+      "step": 1761
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014607738315950287,
+      "loss": 1.356,
+      "step": 1762
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014558568263328654,
+      "loss": 1.3282,
+      "step": 1763
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014509466999523986,
+      "loss": 1.3164,
+      "step": 1764
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014460434619837693,
+      "loss": 1.3444,
+      "step": 1765
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014411471219437488,
+      "loss": 1.3013,
+      "step": 1766
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014362576893357155,
+      "loss": 1.3499,
+      "step": 1767
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001431375173649645,
+      "loss": 1.3459,
+      "step": 1768
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014264995843620883,
+      "loss": 1.34,
+      "step": 1769
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001421630930936148,
+      "loss": 1.3274,
+      "step": 1770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000141676922282147,
+      "loss": 1.3056,
+      "step": 1771
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001411914469454218,
+      "loss": 1.3184,
+      "step": 1772
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014070666802570565,
+      "loss": 1.3746,
+      "step": 1773
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014022258646391355,
+      "loss": 1.3495,
+      "step": 1774
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013973920319960653,
+      "loss": 1.3755,
+      "step": 1775
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013925651917099063,
+      "loss": 1.3339,
+      "step": 1776
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013877453531491496,
+      "loss": 1.3577,
+      "step": 1777
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013829325256686892,
+      "loss": 1.3563,
+      "step": 1778
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013781267186098196,
+      "loss": 1.364,
+      "step": 1779
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013733279413001998,
+      "loss": 1.3396,
+      "step": 1780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013685362030538552,
+      "loss": 1.3264,
+      "step": 1781
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001363751513171143,
+      "loss": 1.371,
+      "step": 1782
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001358973880938739,
+      "loss": 1.3357,
+      "step": 1783
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013542033156296247,
+      "loss": 1.387,
+      "step": 1784
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013494398265030634,
+      "loss": 1.3391,
+      "step": 1785
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013446834228045813,
+      "loss": 1.3417,
+      "step": 1786
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013399341137659571,
+      "loss": 1.3246,
+      "step": 1787
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013351919086051985,
+      "loss": 1.3047,
+      "step": 1788
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001330456816526521,
+      "loss": 1.3551,
+      "step": 1789
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001325728846720339,
+      "loss": 1.3637,
+      "step": 1790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013210080083632408,
+      "loss": 1.34,
+      "step": 1791
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013162943106179747,
+      "loss": 1.3051,
+      "step": 1792
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000131158776263343,
+      "loss": 1.3699,
+      "step": 1793
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013068883735446146,
+      "loss": 1.3661,
+      "step": 1794
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013021961524726456,
+      "loss": 1.2887,
+      "step": 1795
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012975111085247287,
+      "loss": 1.3725,
+      "step": 1796
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012928332507941342,
+      "loss": 1.2992,
+      "step": 1797
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000128816258836019,
+      "loss": 1.3187,
+      "step": 1798
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001283499130288255,
+      "loss": 1.3346,
+      "step": 1799
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012788428856297068,
+      "loss": 1.2883,
+      "step": 1800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012741938634219235,
+      "loss": 1.3606,
+      "step": 1801
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012695520726882604,
+      "loss": 1.3444,
+      "step": 1802
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012649175224380415,
+      "loss": 1.3306,
+      "step": 1803
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012602902216665384,
+      "loss": 1.3462,
+      "step": 1804
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012556701793549457,
+      "loss": 1.3006,
+      "step": 1805
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001251057404470377,
+      "loss": 1.3002,
+      "step": 1806
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012464519059658381,
+      "loss": 1.2938,
+      "step": 1807
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012418536927802094,
+      "loss": 1.3046,
+      "step": 1808
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001237262773838234,
+      "loss": 1.3451,
+      "step": 1809
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001232679158050497,
+      "loss": 1.3805,
+      "step": 1810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001228102854313408,
+      "loss": 1.3472,
+      "step": 1811
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001223533871509187,
+      "loss": 1.3132,
+      "step": 1812
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012189722185058394,
+      "loss": 1.3097,
+      "step": 1813
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001214417904157149,
+      "loss": 1.3661,
+      "step": 1814
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001209870937302655,
+      "loss": 1.3316,
+      "step": 1815
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001205331326767633,
+      "loss": 1.3444,
+      "step": 1816
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012007990813630831,
+      "loss": 1.3584,
+      "step": 1817
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011962742098857099,
+      "loss": 1.3219,
+      "step": 1818
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011917567211179059,
+      "loss": 1.3592,
+      "step": 1819
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011872466238277357,
+      "loss": 1.3212,
+      "step": 1820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011827439267689127,
+      "loss": 1.3294,
+      "step": 1821
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011782486386807922,
+      "loss": 1.3681,
+      "step": 1822
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011737607682883494,
+      "loss": 1.3134,
+      "step": 1823
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011692803243021577,
+      "loss": 1.2975,
+      "step": 1824
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011648073154183797,
+      "loss": 1.3425,
+      "step": 1825
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011603417503187496,
+      "loss": 1.3293,
+      "step": 1826
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011558836376705479,
+      "loss": 1.3041,
+      "step": 1827
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011514329861265943,
+      "loss": 1.3732,
+      "step": 1828
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011469898043252269,
+      "loss": 1.3371,
+      "step": 1829
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011425541008902851,
+      "loss": 1.3633,
+      "step": 1830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001138125884431095,
+      "loss": 1.3711,
+      "step": 1831
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011337051635424473,
+      "loss": 1.3536,
+      "step": 1832
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011292919468045875,
+      "loss": 1.3468,
+      "step": 1833
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011248862427831974,
+      "loss": 1.3267,
+      "step": 1834
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011204880600293727,
+      "loss": 1.3226,
+      "step": 1835
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011160974070796143,
+      "loss": 1.2947,
+      "step": 1836
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011117142924558077,
+      "loss": 1.3793,
+      "step": 1837
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011073387246652073,
+      "loss": 1.3991,
+      "step": 1838
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011029707122004218,
+      "loss": 1.2913,
+      "step": 1839
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010986102635393891,
+      "loss": 1.3139,
+      "step": 1840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010942573871453731,
+      "loss": 1.3471,
+      "step": 1841
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010899120914669397,
+      "loss": 1.3462,
+      "step": 1842
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010855743849379363,
+      "loss": 1.3098,
+      "step": 1843
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010812442759774855,
+      "loss": 1.357,
+      "step": 1844
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010769217729899632,
+      "loss": 1.3331,
+      "step": 1845
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010726068843649794,
+      "loss": 1.3455,
+      "step": 1846
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010682996184773685,
+      "loss": 1.3376,
+      "step": 1847
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010639999836871684,
+      "loss": 1.3383,
+      "step": 1848
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010597079883396065,
+      "loss": 1.3163,
+      "step": 1849
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010554236407650836,
+      "loss": 1.2749,
+      "step": 1850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010511469492791526,
+      "loss": 1.3657,
+      "step": 1851
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010468779221825104,
+      "loss": 1.3148,
+      "step": 1852
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010426165677609789,
+      "loss": 1.3248,
+      "step": 1853
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010383628942854828,
+      "loss": 1.3511,
+      "step": 1854
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001034116910012044,
+      "loss": 1.3375,
+      "step": 1855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001029878623181758,
+      "loss": 1.3091,
+      "step": 1856
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010256480420207809,
+      "loss": 1.3305,
+      "step": 1857
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001021425174740313,
+      "loss": 1.3587,
+      "step": 1858
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010172100295365793,
+      "loss": 1.2946,
+      "step": 1859
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010130026145908222,
+      "loss": 1.3593,
+      "step": 1860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010088029380692781,
+      "loss": 1.3453,
+      "step": 1861
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010046110081231601,
+      "loss": 1.2973,
+      "step": 1862
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010004268328886523,
+      "loss": 1.2812,
+      "step": 1863
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.962504204868838e-05,
+      "loss": 1.3607,
+      "step": 1864
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.920817790239162e-05,
+      "loss": 1.3593,
+      "step": 1865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.879209165907299e-05,
+      "loss": 1.3213,
+      "step": 1866
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.837678412632067e-05,
+      "loss": 1.3584,
+      "step": 1867
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.796225611021143e-05,
+      "loss": 1.3122,
+      "step": 1868
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.75485084153091e-05,
+      "loss": 1.3448,
+      "step": 1869
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.713554184466267e-05,
+      "loss": 1.3001,
+      "step": 1870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.672335719980546e-05,
+      "loss": 1.3491,
+      "step": 1871
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.631195528075304e-05,
+      "loss": 1.3252,
+      "step": 1872
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.590133688600144e-05,
+      "loss": 1.2988,
+      "step": 1873
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.549150281252633e-05,
+      "loss": 1.3384,
+      "step": 1874
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.508245385578085e-05,
+      "loss": 1.3528,
+      "step": 1875
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.467419080969447e-05,
+      "loss": 1.3123,
+      "step": 1876
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.426671446667129e-05,
+      "loss": 1.3591,
+      "step": 1877
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.38600256175881e-05,
+      "loss": 1.3524,
+      "step": 1878
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.345412505179363e-05,
+      "loss": 1.3273,
+      "step": 1879
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.304901355710666e-05,
+      "loss": 1.3068,
+      "step": 1880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.264469191981395e-05,
+      "loss": 1.3287,
+      "step": 1881
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.224116092466978e-05,
+      "loss": 1.2927,
+      "step": 1882
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.183842135489374e-05,
+      "loss": 1.3704,
+      "step": 1883
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.143647399216886e-05,
+      "loss": 1.3857,
+      "step": 1884
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.103531961664119e-05,
+      "loss": 1.3303,
+      "step": 1885
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.063495900691738e-05,
+      "loss": 1.3279,
+      "step": 1886
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.023539294006333e-05,
+      "loss": 1.3398,
+      "step": 1887
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.983662219160321e-05,
+      "loss": 1.3469,
+      "step": 1888
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.94386475355169e-05,
+      "loss": 1.3536,
+      "step": 1889
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.904146974423971e-05,
+      "loss": 1.334,
+      "step": 1890
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.864508958866018e-05,
+      "loss": 1.3611,
+      "step": 1891
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.824950783811836e-05,
+      "loss": 1.3016,
+      "step": 1892
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.7854725260405e-05,
+      "loss": 1.3201,
+      "step": 1893
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.746074262175973e-05,
+      "loss": 1.3025,
+      "step": 1894
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.706756068686949e-05,
+      "loss": 1.2743,
+      "step": 1895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.6675180218867e-05,
+      "loss": 1.2917,
+      "step": 1896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.628360197932977e-05,
+      "loss": 1.3409,
+      "step": 1897
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.589282672827764e-05,
+      "loss": 1.344,
+      "step": 1898
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.550285522417256e-05,
+      "loss": 1.3353,
+      "step": 1899
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.511368822391591e-05,
+      "loss": 1.2754,
+      "step": 1900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.472532648284808e-05,
+      "loss": 1.3256,
+      "step": 1901
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.433777075474635e-05,
+      "loss": 1.3315,
+      "step": 1902
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.395102179182335e-05,
+      "loss": 1.327,
+      "step": 1903
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.356508034472621e-05,
+      "loss": 1.348,
+      "step": 1904
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.317994716253452e-05,
+      "loss": 1.3183,
+      "step": 1905
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.279562299275922e-05,
+      "loss": 1.3002,
+      "step": 1906
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.241210858134107e-05,
+      "loss": 1.3528,
+      "step": 1907
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.20294046726488e-05,
+      "loss": 1.3446,
+      "step": 1908
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.164751200947845e-05,
+      "loss": 1.3377,
+      "step": 1909
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.126643133305139e-05,
+      "loss": 1.3255,
+      "step": 1910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.088616338301275e-05,
+      "loss": 1.3417,
+      "step": 1911
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.050670889743051e-05,
+      "loss": 1.3087,
+      "step": 1912
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.012806861279364e-05,
+      "loss": 1.3374,
+      "step": 1913
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.975024326401092e-05,
+      "loss": 1.3586,
+      "step": 1914
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.937323358440934e-05,
+      "loss": 1.3045,
+      "step": 1915
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.89970403057329e-05,
+      "loss": 1.3471,
+      "step": 1916
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.862166415814054e-05,
+      "loss": 1.3275,
+      "step": 1917
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.824710587020595e-05,
+      "loss": 1.2802,
+      "step": 1918
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.787336616891467e-05,
+      "loss": 1.304,
+      "step": 1919
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.750044577966404e-05,
+      "loss": 1.3032,
+      "step": 1920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.712834542626096e-05,
+      "loss": 1.3529,
+      "step": 1921
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.675706583092057e-05,
+      "loss": 1.3255,
+      "step": 1922
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.638660771426508e-05,
+      "loss": 1.3721,
+      "step": 1923
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.601697179532252e-05,
+      "loss": 1.3141,
+      "step": 1924
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.56481587915247e-05,
+      "loss": 1.3499,
+      "step": 1925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.528016941870657e-05,
+      "loss": 1.3543,
+      "step": 1926
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.491300439110438e-05,
+      "loss": 1.3075,
+      "step": 1927
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.45466644213541e-05,
+      "loss": 1.3855,
+      "step": 1928
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.418115022049082e-05,
+      "loss": 1.3517,
+      "step": 1929
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.381646249794649e-05,
+      "loss": 1.3328,
+      "step": 1930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.345260196154913e-05,
+      "loss": 1.3827,
+      "step": 1931
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.30895693175212e-05,
+      "loss": 1.3408,
+      "step": 1932
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.272736527047836e-05,
+      "loss": 1.3236,
+      "step": 1933
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.236599052342796e-05,
+      "loss": 1.3771,
+      "step": 1934
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.20054457777679e-05,
+      "loss": 1.329,
+      "step": 1935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.164573173328481e-05,
+      "loss": 1.3168,
+      "step": 1936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.128684908815337e-05,
+      "loss": 1.297,
+      "step": 1937
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.092879853893408e-05,
+      "loss": 1.3433,
+      "step": 1938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.057158078057297e-05,
+      "loss": 1.3127,
+      "step": 1939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.021519650639951e-05,
+      "loss": 1.3415,
+      "step": 1940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.98596464081252e-05,
+      "loss": 1.3755,
+      "step": 1941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.950493117584272e-05,
+      "loss": 1.3319,
+      "step": 1942
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.915105149802431e-05,
+      "loss": 1.3826,
+      "step": 1943
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.879800806152047e-05,
+      "loss": 1.3713,
+      "step": 1944
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.84458015515586e-05,
+      "loss": 1.3076,
+      "step": 1945
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.809443265174181e-05,
+      "loss": 1.3372,
+      "step": 1946
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.774390204404713e-05,
+      "loss": 1.285,
+      "step": 1947
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.739421040882505e-05,
+      "loss": 1.3149,
+      "step": 1948
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.704535842479715e-05,
+      "loss": 1.3515,
+      "step": 1949
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.66973467690557e-05,
+      "loss": 1.3359,
+      "step": 1950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.635017611706184e-05,
+      "loss": 1.3497,
+      "step": 1951
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.60038471426444e-05,
+      "loss": 1.3271,
+      "step": 1952
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.565836051799862e-05,
+      "loss": 1.359,
+      "step": 1953
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.531371691368482e-05,
+      "loss": 1.3294,
+      "step": 1954
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.496991699862686e-05,
+      "loss": 1.3185,
+      "step": 1955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.462696144011149e-05,
+      "loss": 1.3411,
+      "step": 1956
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.428485090378616e-05,
+      "loss": 1.3065,
+      "step": 1957
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.394358605365857e-05,
+      "loss": 1.2815,
+      "step": 1958
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.360316755209494e-05,
+      "loss": 1.2674,
+      "step": 1959
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.326359605981863e-05,
+      "loss": 1.3217,
+      "step": 1960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.292487223590904e-05,
+      "loss": 1.3185,
+      "step": 1961
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.258699673780083e-05,
+      "loss": 1.3525,
+      "step": 1962
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.224997022128137e-05,
+      "loss": 1.4006,
+      "step": 1963
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.191379334049063e-05,
+      "loss": 1.3024,
+      "step": 1964
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.157846674791961e-05,
+      "loss": 1.3694,
+      "step": 1965
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.124399109440865e-05,
+      "loss": 1.2992,
+      "step": 1966
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.091036702914687e-05,
+      "loss": 1.3317,
+      "step": 1967
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.057759519967004e-05,
+      "loss": 1.2752,
+      "step": 1968
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0245676251860225e-05,
+      "loss": 1.3361,
+      "step": 1969
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9914610829944035e-05,
+      "loss": 1.3219,
+      "step": 1970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9584399576491365e-05,
+      "loss": 1.3452,
+      "step": 1971
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.925504313241431e-05,
+      "loss": 1.3001,
+      "step": 1972
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.892654213696591e-05,
+      "loss": 1.3205,
+      "step": 1973
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.859889722773859e-05,
+      "loss": 1.3457,
+      "step": 1974
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8272109040663435e-05,
+      "loss": 1.3436,
+      "step": 1975
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7946178210008806e-05,
+      "loss": 1.3081,
+      "step": 1976
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7621105368378556e-05,
+      "loss": 1.3651,
+      "step": 1977
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.729689114671177e-05,
+      "loss": 1.3237,
+      "step": 1978
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.697353617428064e-05,
+      "loss": 1.2984,
+      "step": 1979
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.665104107868968e-05,
+      "loss": 1.356,
+      "step": 1980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6329406485874924e-05,
+      "loss": 1.3137,
+      "step": 1981
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6008633020101564e-05,
+      "loss": 1.3623,
+      "step": 1982
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5688721303963796e-05,
+      "loss": 1.3167,
+      "step": 1983
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.536967195838333e-05,
+      "loss": 1.3128,
+      "step": 1984
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5051485602607643e-05,
+      "loss": 1.3091,
+      "step": 1985
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.473416285420973e-05,
+      "loss": 1.3367,
+      "step": 1986
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.441770432908594e-05,
+      "loss": 1.3648,
+      "step": 1987
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.410211064145565e-05,
+      "loss": 1.304,
+      "step": 1988
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3787382403859384e-05,
+      "loss": 1.3071,
+      "step": 1989
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.347352022715801e-05,
+      "loss": 1.3262,
+      "step": 1990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.316052472053151e-05,
+      "loss": 1.3545,
+      "step": 1991
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.284839649147766e-05,
+      "loss": 1.2922,
+      "step": 1992
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.253713614581074e-05,
+      "loss": 1.3163,
+      "step": 1993
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.222674428766078e-05,
+      "loss": 1.2823,
+      "step": 1994
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.191722151947226e-05,
+      "loss": 1.3634,
+      "step": 1995
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1608568442002404e-05,
+      "loss": 1.2651,
+      "step": 1996
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.130078565432089e-05,
+      "loss": 1.3324,
+      "step": 1997
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0993873753807816e-05,
+      "loss": 1.3052,
+      "step": 1998
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.068783333615334e-05,
+      "loss": 1.3226,
+      "step": 1999
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.03826649953561e-05,
+      "loss": 1.3313,
+      "step": 2000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.007836932372178e-05,
+      "loss": 1.3013,
+      "step": 2001
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.977494691186263e-05,
+      "loss": 1.318,
+      "step": 2002
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.947239834869588e-05,
+      "loss": 1.2968,
+      "step": 2003
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.917072422144248e-05,
+      "loss": 1.3544,
+      "step": 2004
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.886992511562654e-05,
+      "loss": 1.3447,
+      "step": 2005
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8570001615073524e-05,
+      "loss": 1.3112,
+      "step": 2006
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.827095430190964e-05,
+      "loss": 1.3495,
+      "step": 2007
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.797278375656028e-05,
+      "loss": 1.358,
+      "step": 2008
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7675490557749336e-05,
+      "loss": 1.3072,
+      "step": 2009
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.737907528249763e-05,
+      "loss": 1.3203,
+      "step": 2010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.708353850612229e-05,
+      "loss": 1.2764,
+      "step": 2011
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.678888080223492e-05,
+      "loss": 1.3452,
+      "step": 2012
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.649510274274127e-05,
+      "loss": 1.3463,
+      "step": 2013
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.620220489783977e-05,
+      "loss": 1.3544,
+      "step": 2014
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.591018783602008e-05,
+      "loss": 1.2931,
+      "step": 2015
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.561905212406286e-05,
+      "loss": 1.3832,
+      "step": 2016
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.532879832703757e-05,
+      "loss": 1.305,
+      "step": 2017
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.503942700830249e-05,
+      "loss": 1.3275,
+      "step": 2018
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4750938729502866e-05,
+      "loss": 1.3409,
+      "step": 2019
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.446333405056979e-05,
+      "loss": 1.3648,
+      "step": 2020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.417661352971969e-05,
+      "loss": 1.3768,
+      "step": 2021
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.38907777234529e-05,
+      "loss": 1.335,
+      "step": 2022
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.360582718655237e-05,
+      "loss": 1.3511,
+      "step": 2023
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3321762472083024e-05,
+      "loss": 1.2816,
+      "step": 2024
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3038584131390444e-05,
+      "loss": 1.299,
+      "step": 2025
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.27562927140997e-05,
+      "loss": 1.37,
+      "step": 2026
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.247488876811456e-05,
+      "loss": 1.2892,
+      "step": 2027
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.21943728396163e-05,
+      "loss": 1.3596,
+      "step": 2028
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.191474547306251e-05,
+      "loss": 1.305,
+      "step": 2029
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.163600721118638e-05,
+      "loss": 1.3314,
+      "step": 2030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.135815859499503e-05,
+      "loss": 1.3408,
+      "step": 2031
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.108120016376921e-05,
+      "loss": 1.3402,
+      "step": 2032
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0805132455061845e-05,
+      "loss": 1.3192,
+      "step": 2033
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.052995600469678e-05,
+      "loss": 1.3053,
+      "step": 2034
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.025567134676844e-05,
+      "loss": 1.3321,
+      "step": 2035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.998227901363982e-05,
+      "loss": 1.3434,
+      "step": 2036
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.970977953594251e-05,
+      "loss": 1.3244,
+      "step": 2037
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9438173442575e-05,
+      "loss": 1.3697,
+      "step": 2038
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.916746126070142e-05,
+      "loss": 1.3661,
+      "step": 2039
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.889764351575131e-05,
+      "loss": 1.3138,
+      "step": 2040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.862872073141821e-05,
+      "loss": 1.3247,
+      "step": 2041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.83606934296582e-05,
+      "loss": 1.3911,
+      "step": 2042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8093562130689764e-05,
+      "loss": 1.3438,
+      "step": 2043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.78273273529921e-05,
+      "loss": 1.3252,
+      "step": 2044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.756198961330426e-05,
+      "loss": 1.357,
+      "step": 2045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.729754942662439e-05,
+      "loss": 1.3096,
+      "step": 2046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7034007306208486e-05,
+      "loss": 1.2729,
+      "step": 2047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.677136376356949e-05,
+      "loss": 1.3058,
+      "step": 2048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.650961930847635e-05,
+      "loss": 1.3202,
+      "step": 2049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.624877444895269e-05,
+      "loss": 1.3171,
+      "step": 2050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5988829691276425e-05,
+      "loss": 1.3025,
+      "step": 2051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.572978553997836e-05,
+      "loss": 1.3076,
+      "step": 2052
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.547164249784107e-05,
+      "loss": 1.2808,
+      "step": 2053
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.521440106589846e-05,
+      "loss": 1.3108,
+      "step": 2054
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.49580617434343e-05,
+      "loss": 1.3516,
+      "step": 2055
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.47026250279816e-05,
+      "loss": 1.3614,
+      "step": 2056
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.444809141532135e-05,
+      "loss": 1.3712,
+      "step": 2057
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.419446139948168e-05,
+      "loss": 1.3678,
+      "step": 2058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.394173547273699e-05,
+      "loss": 1.3262,
+      "step": 2059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.368991412560696e-05,
+      "loss": 1.3359,
+      "step": 2060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.343899784685539e-05,
+      "loss": 1.2297,
+      "step": 2061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.318898712348956e-05,
+      "loss": 1.3367,
+      "step": 2062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.293988244075918e-05,
+      "loss": 1.3917,
+      "step": 2063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2691684282155186e-05,
+      "loss": 1.3451,
+      "step": 2064
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.244439312940933e-05,
+      "loss": 1.3791,
+      "step": 2065
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2198009462492685e-05,
+      "loss": 1.2912,
+      "step": 2066
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1952533759615156e-05,
+      "loss": 1.3469,
+      "step": 2067
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.170796649722435e-05,
+      "loss": 1.3415,
+      "step": 2068
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.146430815000445e-05,
+      "loss": 1.2783,
+      "step": 2069
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.122155919087572e-05,
+      "loss": 1.3048,
+      "step": 2070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.097972009099342e-05,
+      "loss": 1.338,
+      "step": 2071
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.073879131974661e-05,
+      "loss": 1.2533,
+      "step": 2072
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0498773344757757e-05,
+      "loss": 1.3184,
+      "step": 2073
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0259666631881288e-05,
+      "loss": 1.3345,
+      "step": 2074
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.002147164520319e-05,
+      "loss": 1.3067,
+      "step": 2075
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9784188847039817e-05,
+      "loss": 1.2922,
+      "step": 2076
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9547818697936823e-05,
+      "loss": 1.3254,
+      "step": 2077
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9312361656668673e-05,
+      "loss": 1.3065,
+      "step": 2078
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9077818180237692e-05,
+      "loss": 1.3249,
+      "step": 2079
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.884418872387273e-05,
+      "loss": 1.3347,
+      "step": 2080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.861147374102879e-05,
+      "loss": 1.367,
+      "step": 2081
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8379673683386065e-05,
+      "loss": 1.3235,
+      "step": 2082
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8148789000848663e-05,
+      "loss": 1.3335,
+      "step": 2083
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7918820141544287e-05,
+      "loss": 1.3047,
+      "step": 2084
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7689767551823063e-05,
+      "loss": 1.3094,
+      "step": 2085
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.746163167625654e-05,
+      "loss": 1.303,
+      "step": 2086
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7234412957637345e-05,
+      "loss": 1.3016,
+      "step": 2087
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.700811183697749e-05,
+      "loss": 1.3244,
+      "step": 2088
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6782728753508455e-05,
+      "loss": 1.3546,
+      "step": 2089
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.655826414467971e-05,
+      "loss": 1.3075,
+      "step": 2090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.633471844615798e-05,
+      "loss": 1.3152,
+      "step": 2091
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.611209209182652e-05,
+      "loss": 1.3144,
+      "step": 2092
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5890385513784308e-05,
+      "loss": 1.2964,
+      "step": 2093
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5669599142344957e-05,
+      "loss": 1.3391,
+      "step": 2094
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5449733406036236e-05,
+      "loss": 1.3327,
+      "step": 2095
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.523078873159873e-05,
+      "loss": 1.3297,
+      "step": 2096
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5012765543985626e-05,
+      "loss": 1.3158,
+      "step": 2097
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.479566426636154e-05,
+      "loss": 1.3634,
+      "step": 2098
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.457948532010146e-05,
+      "loss": 1.2575,
+      "step": 2099
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4364229124790527e-05,
+      "loss": 1.2926,
+      "step": 2100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4149896098222825e-05,
+      "loss": 1.3147,
+      "step": 2101
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.393648665640047e-05,
+      "loss": 1.3302,
+      "step": 2102
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3724001213533243e-05,
+      "loss": 1.3082,
+      "step": 2103
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.35124401820373e-05,
+      "loss": 1.29,
+      "step": 2104
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.330180397253473e-05,
+      "loss": 1.31,
+      "step": 2105
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.309209299385262e-05,
+      "loss": 1.3407,
+      "step": 2106
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.28833076530221e-05,
+      "loss": 1.3286,
+      "step": 2107
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2675448355277905e-05,
+      "loss": 1.3152,
+      "step": 2108
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2468515504057375e-05,
+      "loss": 1.3515,
+      "step": 2109
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2262509500999507e-05,
+      "loss": 1.3237,
+      "step": 2110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2057430745944575e-05,
+      "loss": 1.2734,
+      "step": 2111
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1853279636933122e-05,
+      "loss": 1.3393,
+      "step": 2112
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1650056570205078e-05,
+      "loss": 1.3325,
+      "step": 2113
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.144776194019932e-05,
+      "loss": 1.3241,
+      "step": 2114
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1246396139552425e-05,
+      "loss": 1.3302,
+      "step": 2115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1045959559098438e-05,
+      "loss": 1.3829,
+      "step": 2116
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0846452587867882e-05,
+      "loss": 1.3175,
+      "step": 2117
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0647875613086677e-05,
+      "loss": 1.3044,
+      "step": 2118
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.045022902017607e-05,
+      "loss": 1.2935,
+      "step": 2119
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.025351319275137e-05,
+      "loss": 1.326,
+      "step": 2120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0057728512621222e-05,
+      "loss": 1.3242,
+      "step": 2121
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.986287535978709e-05,
+      "loss": 1.3222,
+      "step": 2122
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9668954112442572e-05,
+      "loss": 1.3022,
+      "step": 2123
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.947596514697236e-05,
+      "loss": 1.3752,
+      "step": 2124
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.928390883795167e-05,
+      "loss": 1.3069,
+      "step": 2125
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.909278555814542e-05,
+      "loss": 1.3177,
+      "step": 2126
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8902595678507828e-05,
+      "loss": 1.3327,
+      "step": 2127
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8713339568181386e-05,
+      "loss": 1.3348,
+      "step": 2128
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8525017594495986e-05,
+      "loss": 1.3921,
+      "step": 2129
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8337630122968795e-05,
+      "loss": 1.3523,
+      "step": 2130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.815117751730294e-05,
+      "loss": 1.3174,
+      "step": 2131
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7965660139387163e-05,
+      "loss": 1.3313,
+      "step": 2132
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7781078349295098e-05,
+      "loss": 1.3121,
+      "step": 2133
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7597432505284118e-05,
+      "loss": 1.3489,
+      "step": 2134
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7414722963795427e-05,
+      "loss": 1.3551,
+      "step": 2135
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7232950079452747e-05,
+      "loss": 1.2762,
+      "step": 2136
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7052114205061754e-05,
+      "loss": 1.3387,
+      "step": 2137
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.687221569160968e-05,
+      "loss": 1.3699,
+      "step": 2138
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6693254888264288e-05,
+      "loss": 1.3277,
+      "step": 2139
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6515232142373216e-05,
+      "loss": 1.3921,
+      "step": 2140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6338147799463686e-05,
+      "loss": 1.3524,
+      "step": 2141
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.616200220324132e-05,
+      "loss": 1.3479,
+      "step": 2142
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5986795695589806e-05,
+      "loss": 1.3641,
+      "step": 2143
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5812528616570232e-05,
+      "loss": 1.3334,
+      "step": 2144
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.563920130442009e-05,
+      "loss": 1.3076,
+      "step": 2145
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5466814095553117e-05,
+      "loss": 1.3057,
+      "step": 2146
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5295367324558274e-05,
+      "loss": 1.3452,
+      "step": 2147
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5124861324199102e-05,
+      "loss": 1.3618,
+      "step": 2148
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.495529642541349e-05,
+      "loss": 1.322,
+      "step": 2149
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4786672957312397e-05,
+      "loss": 1.3499,
+      "step": 2150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4618991247179747e-05,
+      "loss": 1.3464,
+      "step": 2151
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4452251620471536e-05,
+      "loss": 1.3382,
+      "step": 2152
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.428645440081533e-05,
+      "loss": 1.3414,
+      "step": 2153
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4121599910009331e-05,
+      "loss": 1.2938,
+      "step": 2154
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3957688468022311e-05,
+      "loss": 1.2698,
+      "step": 2155
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3794720392992344e-05,
+      "loss": 1.2845,
+      "step": 2156
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.363269600122674e-05,
+      "loss": 1.3567,
+      "step": 2157
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3471615607201116e-05,
+      "loss": 1.3103,
+      "step": 2158
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3311479523558822e-05,
+      "loss": 1.2733,
+      "step": 2159
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3152288061110518e-05,
+      "loss": 1.3575,
+      "step": 2160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2994041528833267e-05,
+      "loss": 1.3483,
+      "step": 2161
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2836740233870215e-05,
+      "loss": 1.3583,
+      "step": 2162
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2680384481529917e-05,
+      "loss": 1.3017,
+      "step": 2163
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2524974575285508e-05,
+      "loss": 1.3264,
+      "step": 2164
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2370510816774539e-05,
+      "loss": 1.3145,
+      "step": 2165
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2216993505798136e-05,
+      "loss": 1.3372,
+      "step": 2166
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2064422940320285e-05,
+      "loss": 1.2893,
+      "step": 2167
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1912799416467556e-05,
+      "loss": 1.318,
+      "step": 2168
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1762123228528432e-05,
+      "loss": 1.2895,
+      "step": 2169
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1612394668952531e-05,
+      "loss": 1.2993,
+      "step": 2170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1463614028350389e-05,
+      "loss": 1.3242,
+      "step": 2171
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1315781595492625e-05,
+      "loss": 1.3034,
+      "step": 2172
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1168897657309384e-05,
+      "loss": 1.314,
+      "step": 2173
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1022962498890065e-05,
+      "loss": 1.3333,
+      "step": 2174
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0877976403482314e-05,
+      "loss": 1.2814,
+      "step": 2175
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0733939652491975e-05,
+      "loss": 1.3333,
+      "step": 2176
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0590852525482197e-05,
+      "loss": 1.3616,
+      "step": 2177
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0448715300172828e-05,
+      "loss": 1.3063,
+      "step": 2178
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0307528252440246e-05,
+      "loss": 1.2814,
+      "step": 2179
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.016729165631669e-05,
+      "loss": 1.3164,
+      "step": 2180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0028005783989381e-05,
+      "loss": 1.3118,
+      "step": 2181
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.889670905800397e-06,
+      "loss": 1.3063,
+      "step": 2182
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.752287290246076e-06,
+      "loss": 1.4262,
+      "step": 2183
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.615855203976287e-06,
+      "loss": 1.2813,
+      "step": 2184
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.480374911794265e-06,
+      "loss": 1.3214,
+      "step": 2185
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.345846676655723e-06,
+      "loss": 1.3648,
+      "step": 2186
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.212270759668517e-06,
+      "loss": 1.2785,
+      "step": 2187
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.079647420092318e-06,
+      "loss": 1.2947,
+      "step": 2188
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.947976915337663e-06,
+      "loss": 1.3082,
+      "step": 2189
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.817259500965957e-06,
+      "loss": 1.3212,
+      "step": 2190
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.687495430688585e-06,
+      "loss": 1.3645,
+      "step": 2191
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.55868495636658e-06,
+      "loss": 1.3364,
+      "step": 2192
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.430828328010231e-06,
+      "loss": 1.3109,
+      "step": 2193
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.303925793778367e-06,
+      "loss": 1.3279,
+      "step": 2194
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.17797759997807e-06,
+      "loss": 1.3266,
+      "step": 2195
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.052983991064189e-06,
+      "loss": 1.2834,
+      "step": 2196
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.928945209638606e-06,
+      "loss": 1.3593,
+      "step": 2197
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.805861496450074e-06,
+      "loss": 1.2991,
+      "step": 2198
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.683733090393829e-06,
+      "loss": 1.3557,
+      "step": 2199
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.562560228510651e-06,
+      "loss": 1.3099,
+      "step": 2200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.442343145986796e-06,
+      "loss": 1.3107,
+      "step": 2201
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.323082076153509e-06,
+      "loss": 1.3749,
+      "step": 2202
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.20477725048635e-06,
+      "loss": 1.2833,
+      "step": 2203
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.087428898604975e-06,
+      "loss": 1.2821,
+      "step": 2204
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.971037248272582e-06,
+      "loss": 1.324,
+      "step": 2205
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.855602525395521e-06,
+      "loss": 1.3375,
+      "step": 2206
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.741124954022848e-06,
+      "loss": 1.316,
+      "step": 2207
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.627604756345829e-06,
+      "loss": 1.3104,
+      "step": 2208
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.5150421526974926e-06,
+      "loss": 1.3092,
+      "step": 2209
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.403437361552411e-06,
+      "loss": 1.341,
+      "step": 2210
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.2927905995259215e-06,
+      "loss": 1.3381,
+      "step": 2211
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.1831020813740705e-06,
+      "loss": 1.3629,
+      "step": 2212
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.074372019993002e-06,
+      "loss": 1.3635,
+      "step": 2213
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.9666006264184615e-06,
+      "loss": 1.3704,
+      "step": 2214
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.859788109825792e-06,
+      "loss": 1.3557,
+      "step": 2215
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.753934677528772e-06,
+      "loss": 1.3104,
+      "step": 2216
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.649040534980166e-06,
+      "loss": 1.3249,
+      "step": 2217
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.5451058857705625e-06,
+      "loss": 1.3327,
+      "step": 2218
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.442130931628264e-06,
+      "loss": 1.2958,
+      "step": 2219
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.340115872418949e-06,
+      "loss": 1.2972,
+      "step": 2220
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.239060906145288e-06,
+      "loss": 1.3418,
+      "step": 2221
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.138966228946273e-06,
+      "loss": 1.2852,
+      "step": 2222
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.039832035097336e-06,
+      "loss": 1.3257,
+      "step": 2223
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.941658517009395e-06,
+      "loss": 1.3437,
+      "step": 2224
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8444458652290305e-06,
+      "loss": 1.3341,
+      "step": 2225
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.748194268437811e-06,
+      "loss": 1.3444,
+      "step": 2226
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.652903913451801e-06,
+      "loss": 1.2695,
+      "step": 2227
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5585749852216084e-06,
+      "loss": 1.2809,
+      "step": 2228
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.465207666831728e-06,
+      "loss": 1.28,
+      "step": 2229
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3728021395000875e-06,
+      "loss": 1.2711,
+      "step": 2230
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.281358582578054e-06,
+      "loss": 1.3189,
+      "step": 2231
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.190877173549879e-06,
+      "loss": 1.3543,
+      "step": 2232
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.101358088032192e-06,
+      "loss": 1.2989,
+      "step": 2233
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.012801499774066e-06,
+      "loss": 1.317,
+      "step": 2234
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.925207580656231e-06,
+      "loss": 1.2997,
+      "step": 2235
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.838576500691193e-06,
+      "loss": 1.3172,
+      "step": 2236
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.752908428022506e-06,
+      "loss": 1.3463,
+      "step": 2237
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.668203528924552e-06,
+      "loss": 1.3544,
+      "step": 2238
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.584461967802433e-06,
+      "loss": 1.3657,
+      "step": 2239
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5016839071914663e-06,
+      "loss": 1.3433,
+      "step": 2240
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4198695077567453e-06,
+      "loss": 1.3207,
+      "step": 2241
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3390189282932472e-06,
+      "loss": 1.3268,
+      "step": 2242
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2591323257248896e-06,
+      "loss": 1.3196,
+      "step": 2243
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.180209855104976e-06,
+      "loss": 1.2956,
+      "step": 2244
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1022516696152524e-06,
+      "loss": 1.3634,
+      "step": 2245
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0252579205660157e-06,
+      "loss": 1.2879,
+      "step": 2246
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.949228757395561e-06,
+      "loss": 1.3009,
+      "step": 2247
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8741643276700702e-06,
+      "loss": 1.3271,
+      "step": 2248
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.800064777083222e-06,
+      "loss": 1.3597,
+      "step": 2249
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7269302494559723e-06,
+      "loss": 1.3653,
+      "step": 2250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.654760886736274e-06,
+      "loss": 1.288,
+      "step": 2251
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5835568289987452e-06,
+      "loss": 1.3181,
+      "step": 2252
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5133182144444487e-06,
+      "loss": 1.3607,
+      "step": 2253
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.444045179400445e-06,
+      "loss": 1.3045,
+      "step": 2254
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3757378583200153e-06,
+      "loss": 1.329,
+      "step": 2255
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3083963837817745e-06,
+      "loss": 1.2859,
+      "step": 2256
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.242020886489893e-06,
+      "loss": 1.2887,
+      "step": 2257
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.176611495273484e-06,
+      "loss": 1.3437,
+      "step": 2258
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1121683370867727e-06,
+      "loss": 1.3058,
+      "step": 2259
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.048691537008318e-06,
+      "loss": 1.3433,
+      "step": 2260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.986181218241345e-06,
+      "loss": 1.3494,
+      "step": 2261
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9246375021130247e-06,
+      "loss": 1.2808,
+      "step": 2262
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8640605080744722e-06,
+      "loss": 1.3705,
+      "step": 2263
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8044503537005264e-06,
+      "loss": 1.2917,
+      "step": 2264
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7458071546895272e-06,
+      "loss": 1.3444,
+      "step": 2265
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6881310248628156e-06,
+      "loss": 1.3136,
+      "step": 2266
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6314220761649567e-06,
+      "loss": 1.3728,
+      "step": 2267
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575680418663239e-06,
+      "loss": 1.2901,
+      "step": 2268
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5209061605474529e-06,
+      "loss": 1.3692,
+      "step": 2269
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4670994081297795e-06,
+      "loss": 1.3076,
+      "step": 2270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4142602658446245e-06,
+      "loss": 1.3033,
+      "step": 2271
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.362388836248174e-06,
+      "loss": 1.317,
+      "step": 2272
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3114852200186157e-06,
+      "loss": 1.3722,
+      "step": 2273
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.261549515955529e-06,
+      "loss": 1.2918,
+      "step": 2274
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.21258182097983e-06,
+      "loss": 1.3039,
+      "step": 2275
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.164582230133715e-06,
+      "loss": 1.3383,
+      "step": 2276
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1175508365802723e-06,
+      "loss": 1.2932,
+      "step": 2277
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.071487731603482e-06,
+      "loss": 1.3641,
+      "step": 2278
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0263930046079395e-06,
+      "loss": 1.3276,
+      "step": 2279
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.822667431186872e-07,
+      "loss": 1.3502,
+      "step": 2280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.391090327811048e-07,
+      "loss": 1.3294,
+      "step": 2281
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.969199573605202e-07,
+      "loss": 1.3218,
+      "step": 2282
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.556995987424876e-07,
+      "loss": 1.3247,
+      "step": 2283
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.15448036932176e-07,
+      "loss": 1.2941,
+      "step": 2284
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.761653500544252e-07,
+      "loss": 1.3247,
+      "step": 2285
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.378516143536907e-07,
+      "loss": 1.3211,
+      "step": 2286
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.00506904193654e-07,
+      "loss": 1.3101,
+      "step": 2287
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.641312920571685e-07,
+      "loss": 1.3067,
+      "step": 2288
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.287248485463137e-07,
+      "loss": 1.262,
+      "step": 2289
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.942876423818966e-07,
+      "loss": 1.3287,
+      "step": 2290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.608197404036731e-07,
+      "loss": 1.3088,
+      "step": 2291
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.283212075700705e-07,
+      "loss": 1.3226,
+      "step": 2292
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.967921069578551e-07,
+      "loss": 1.3295,
+      "step": 2293
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.662324997624645e-07,
+      "loss": 1.2992,
+      "step": 2294
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.366424452973972e-07,
+      "loss": 1.3419,
+      "step": 2295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0802200099454567e-07,
+      "loss": 1.3774,
+      "step": 2296
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.80371222403697e-07,
+      "loss": 1.2798,
+      "step": 2297
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.536901631926992e-07,
+      "loss": 1.3432,
+      "step": 2298
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.279788751471835e-07,
+      "loss": 1.2715,
+      "step": 2299
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.032374081706757e-07,
+      "loss": 1.3374,
+      "step": 2300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7946581028420734e-07,
+      "loss": 1.3597,
+      "step": 2301
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.56664127626427e-07,
+      "loss": 1.3183,
+      "step": 2302
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3483240445343334e-07,
+      "loss": 1.3341,
+      "step": 2303
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1397068313888656e-07,
+      "loss": 1.3637,
+      "step": 2304
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9407900417345304e-07,
+      "loss": 1.3287,
+      "step": 2305
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7515740616536047e-07,
+      "loss": 1.2896,
+      "step": 2306
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.572059258396763e-07,
+      "loss": 1.2867,
+      "step": 2307
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4022459803880726e-07,
+      "loss": 1.3516,
+      "step": 2308
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2421345572205534e-07,
+      "loss": 1.3054,
+      "step": 2309
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0917252996567318e-07,
+      "loss": 1.3305,
+      "step": 2310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.510184996280869e-08,
+      "loss": 1.3251,
+      "step": 2311
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.200144302344947e-08,
+      "loss": 1.3593,
+      "step": 2312
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.987133457442286e-08,
+      "loss": 1.3412,
+      "step": 2313
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.87115481592293e-08,
+      "loss": 1.2787,
+      "step": 2314
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.852210543809799e-08,
+      "loss": 1.3171,
+      "step": 2315
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.930302618776471e-08,
+      "loss": 1.3371,
+      "step": 2316
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.105432830180499e-08,
+      "loss": 1.3298,
+      "step": 2317
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.377602779018995e-08,
+      "loss": 1.355,
+      "step": 2318
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.746813877950837e-08,
+      "loss": 1.319,
+      "step": 2319
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2130673512744661e-08,
+      "loss": 1.3523,
+      "step": 2320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.76364234961191e-09,
+      "loss": 1.2766,
+      "step": 2321
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.367053766052287e-09,
+      "loss": 1.3177,
+      "step": 2322
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9409143545701115e-09,
+      "loss": 1.3115,
+      "step": 2323
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.852288240653202e-10,
+      "loss": 1.3288,
+      "step": 2324
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.3263,
+      "step": 2325
+    },
+    {
+      "epoch": 1.0,
+      "step": 2325,
+      "total_flos": 2.2405156004744397e+18,
+      "train_loss": 1.4318823972312353,
+      "train_runtime": 112383.2063,
+      "train_samples_per_second": 5.298,
+      "train_steps_per_second": 0.021
+    }
+  ],
+  "max_steps": 2325,
+  "num_train_epochs": 1,
+  "total_flos": 2.2405156004744397e+18,
+  "trial_name": null,
+  "trial_params": null
+}