diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,13110 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999139982226299,
+  "eval_steps": 500,
+  "global_step": 2180,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5151515151515153e-05,
+      "loss": 7.9734,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0303030303030306e-05,
+      "loss": 8.0248,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.545454545454546e-05,
+      "loss": 7.6175,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.060606060606061e-05,
+      "loss": 6.5899,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.575757575757576e-05,
+      "loss": 6.1489,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.090909090909092e-05,
+      "loss": 5.6747,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010606060606060606,
+      "loss": 5.3295,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012121212121212122,
+      "loss": 5.2052,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013636363636363637,
+      "loss": 5.0315,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015151515151515152,
+      "loss": 4.8296,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00016666666666666666,
+      "loss": 4.7823,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018181818181818183,
+      "loss": 4.6469,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019696969696969695,
+      "loss": 4.5349,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00021212121212121213,
+      "loss": 4.5199,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022727272727272727,
+      "loss": 4.4334,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00024242424242424245,
+      "loss": 4.3388,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00025757575757575756,
+      "loss": 4.2643,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00027272727272727274,
+      "loss": 4.2319,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002878787878787879,
+      "loss": 4.1872,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00030303030303030303,
+      "loss": 3.9986,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003181818181818182,
+      "loss": 4.0341,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003333333333333333,
+      "loss": 4.0688,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003484848484848485,
+      "loss": 3.904,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00036363636363636367,
+      "loss": 3.844,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003787878787878788,
+      "loss": 3.9177,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003939393939393939,
+      "loss": 3.7793,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00040909090909090913,
+      "loss": 3.7186,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00042424242424242425,
+      "loss": 3.707,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004393939393939394,
+      "loss": 3.7125,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00045454545454545455,
+      "loss": 3.6874,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004696969696969697,
+      "loss": 3.569,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004848484848484849,
+      "loss": 3.5869,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005,
+      "loss": 3.5748,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005151515151515151,
+      "loss": 3.4655,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005303030303030302,
+      "loss": 3.5675,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005454545454545455,
+      "loss": 3.4905,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005606060606060606,
+      "loss": 3.4641,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005757575757575758,
+      "loss": 3.4974,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005909090909090909,
+      "loss": 3.4201,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006060606060606061,
+      "loss": 3.3907,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006212121212121212,
+      "loss": 3.3513,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006363636363636364,
+      "loss": 3.3905,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006515151515151515,
+      "loss": 3.4031,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006666666666666666,
+      "loss": 3.409,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006818181818181818,
+      "loss": 3.3561,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000696969696969697,
+      "loss": 3.2597,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007121212121212122,
+      "loss": 3.2802,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007272727272727273,
+      "loss": 3.3601,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007424242424242425,
+      "loss": 3.3011,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007575757575757576,
+      "loss": 3.2537,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007727272727272727,
+      "loss": 3.2074,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007878787878787878,
+      "loss": 3.397,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000803030303030303,
+      "loss": 3.2949,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008181818181818183,
+      "loss": 3.3631,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008333333333333334,
+      "loss": 3.259,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008484848484848485,
+      "loss": 3.2194,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008636363636363636,
+      "loss": 3.2748,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008787878787878789,
+      "loss": 3.3091,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000893939393939394,
+      "loss": 3.1038,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009090909090909091,
+      "loss": 3.1655,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009242424242424242,
+      "loss": 3.2564,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009393939393939394,
+      "loss": 3.1626,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009545454545454546,
+      "loss": 3.2046,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009696969696969698,
+      "loss": 3.1067,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000984848484848485,
+      "loss": 3.0746,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001,
+      "loss": 3.0809,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999994478847943,
+      "loss": 3.0726,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999977915403962,
+      "loss": 3.132,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999950309704639,
+      "loss": 3.1215,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999991166181094,
+      "loss": 3.0711,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999861971808216,
+      "loss": 3.047,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999801239806208,
+      "loss": 3.0411,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999729465939035,
+      "loss": 2.9498,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999646650365212,
+      "loss": 3.0257,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999552793267634,
+      "loss": 2.9457,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999447894853577,
+      "loss": 3.0471,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999331955354708,
+      "loss": 3.0867,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999204975027073,
+      "loss": 2.9383,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999066954151103,
+      "loss": 2.9426,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998917893031614,
+      "loss": 2.9541,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998757791997801,
+      "loss": 2.9519,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998586651403238,
+      "loss": 2.9227,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998404471625885,
+      "loss": 2.8714,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998211253068078,
+      "loss": 3.0129,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998006996156535,
+      "loss": 2.8848,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997791701342347,
+      "loss": 2.937,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997565369100983,
+      "loss": 2.9108,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997327999932291,
+      "loss": 2.8742,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999707959436049,
+      "loss": 2.8652,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996820152934176,
+      "loss": 2.8029,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999654967622631,
+      "loss": 2.8865,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996268164834238,
+      "loss": 2.8393,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999597561937966,
+      "loss": 2.8188,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995672040508656,
+      "loss": 2.7489,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995357428891662,
+      "loss": 2.7584,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995031785223491,
+      "loss": 2.78,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999469511022331,
+      "loss": 2.8005,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994347404634657,
+      "loss": 2.7437,
+      "step": 98
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993988669225423,
+      "loss": 2.7538,
+      "step": 99
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999361890478786,
+      "loss": 2.703,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993238112138583,
+      "loss": 2.7379,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992846292118554,
+      "loss": 2.7077,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999244344559309,
+      "loss": 2.7192,
+      "step": 103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992029573451869,
+      "loss": 2.6641,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991604676608905,
+      "loss": 2.6977,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991168756002568,
+      "loss": 2.8006,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990721812595574,
+      "loss": 2.6327,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990263847374976,
+      "loss": 2.7293,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989794861352173,
+      "loss": 2.6865,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989314855562905,
+      "loss": 2.7535,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988823831067245,
+      "loss": 2.6698,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988321788949597,
+      "loss": 2.6648,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987808730318709,
+      "loss": 2.6878,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987284656307644,
+      "loss": 2.6561,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986749568073802,
+      "loss": 2.6538,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986203466798905,
+      "loss": 2.6771,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985646353688996,
+      "loss": 2.6709,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985078229974437,
+      "loss": 2.6465,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984499096909905,
+      "loss": 2.5135,
+      "step": 119
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983908955774397,
+      "loss": 2.6579,
+      "step": 120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983307807871211,
+      "loss": 2.6684,
+      "step": 121
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982695654527965,
+      "loss": 2.6518,
+      "step": 122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982072497096571,
+      "loss": 2.6189,
+      "step": 123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998143833695325,
+      "loss": 2.5255,
+      "step": 124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980793175498517,
+      "loss": 2.6111,
+      "step": 125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998013701415719,
+      "loss": 2.6522,
+      "step": 126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979469854378372,
+      "loss": 2.5641,
+      "step": 127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997879169763546,
+      "loss": 2.6077,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997810254542614,
+      "loss": 2.4588,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977402399272374,
+      "loss": 2.7433,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976691260720407,
+      "loss": 2.552,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975969131340763,
+      "loss": 2.5703,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975236012728236,
+      "loss": 2.5557,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974491906501886,
+      "loss": 2.5417,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973736814305049,
+      "loss": 2.5373,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972970737805312,
+      "loss": 2.5589,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972193678694525,
+      "loss": 2.4782,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971405638688794,
+      "loss": 2.5525,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970606619528475,
+      "loss": 2.4994,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000996979662297817,
+      "loss": 2.5082,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968975650826721,
+      "loss": 2.6383,
+      "step": 141
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996814370488722,
+      "loss": 2.5395,
+      "step": 142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996730078699698,
+      "loss": 2.4704,
+      "step": 143
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966446899017558,
+      "loss": 2.5295,
+      "step": 144
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965582042834728,
+      "loss": 2.5049,
+      "step": 145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964706220358492,
+      "loss": 2.4152,
+      "step": 146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996381943352307,
+      "loss": 2.4497,
+      "step": 147
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962921684286896,
+      "loss": 2.4645,
+      "step": 148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962012974632614,
+      "loss": 2.5849,
+      "step": 149
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961093306567075,
+      "loss": 2.437,
+      "step": 150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960162682121328,
+      "loss": 2.461,
+      "step": 151
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959221103350623,
+      "loss": 2.4718,
+      "step": 152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958268572334394,
+      "loss": 2.4848,
+      "step": 153
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957305091176274,
+      "loss": 2.4762,
+      "step": 154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956330662004075,
+      "loss": 2.4293,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955345286969779,
+      "loss": 2.3827,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009954348968249551,
+      "loss": 2.484,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009953341708043724,
+      "loss": 2.4871,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952323508576793,
+      "loss": 2.4256,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951294372097406,
+      "loss": 2.4524,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950254300878378,
+      "loss": 2.3711,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000994920329721666,
+      "loss": 2.4768,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009948141363433356,
+      "loss": 2.4727,
+      "step": 163
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00099470685018737,
+      "loss": 2.4217,
+      "step": 164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945984714907073,
+      "loss": 2.4901,
+      "step": 165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994489000492697,
+      "loss": 2.3917,
+      "step": 166
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009943784374351016,
+      "loss": 2.3589,
+      "step": 167
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009942667825620951,
+      "loss": 2.3783,
+      "step": 168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009941540361202634,
+      "loss": 2.3844,
+      "step": 169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940401983586022,
+      "loss": 2.3857,
+      "step": 170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993925269528518,
+      "loss": 2.4101,
+      "step": 171
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009938092498838265,
+      "loss": 2.4538,
+      "step": 172
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009936921396807524,
+      "loss": 2.4388,
+      "step": 173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935739391779292,
+      "loss": 2.4166,
+      "step": 174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993454648636398,
+      "loss": 2.2859,
+      "step": 175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009933342683196074,
+      "loss": 2.4281,
+      "step": 176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009932127984934125,
+      "loss": 2.3489,
+      "step": 177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009930902394260745,
+      "loss": 2.4384,
+      "step": 178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009929665913882607,
+      "loss": 2.3836,
+      "step": 179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009928418546530425,
+      "loss": 2.3754,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009927160294958964,
+      "loss": 2.4096,
+      "step": 181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992589116194702,
+      "loss": 2.3367,
+      "step": 182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992461115029743,
+      "loss": 2.4653,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992332026283704,
+      "loss": 2.3603,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009922018502416736,
+      "loss": 2.3874,
+      "step": 185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009920705871911395,
+      "loss": 2.3015,
+      "step": 186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009919382374219915,
+      "loss": 2.3273,
+      "step": 187
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009918048012265187,
+      "loss": 2.3248,
+      "step": 188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009916702788994097,
+      "loss": 2.3784,
+      "step": 189
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009915346707377519,
+      "loss": 2.3489,
+      "step": 190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009913979770410305,
+      "loss": 2.2651,
+      "step": 191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009912601981111285,
+      "loss": 2.282,
+      "step": 192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009911213342523248,
+      "loss": 2.4068,
+      "step": 193
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990981385771295,
+      "loss": 2.3632,
+      "step": 194
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00099084035297711,
+      "loss": 2.4206,
+      "step": 195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990698236181235,
+      "loss": 2.3565,
+      "step": 196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009905550356975293,
+      "loss": 2.3854,
+      "step": 197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009904107518422457,
+      "loss": 2.2664,
+      "step": 198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009902653849340295,
+      "loss": 2.3395,
+      "step": 199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009901189352939177,
+      "loss": 2.4009,
+      "step": 200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009899714032453387,
+      "loss": 2.3877,
+      "step": 201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989822789114111,
+      "loss": 2.3297,
+      "step": 202
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009896730932284434,
+      "loss": 2.3869,
+      "step": 203
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009895223159189332,
+      "loss": 2.3589,
+      "step": 204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009893704575185663,
+      "loss": 2.4283,
+      "step": 205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989217518362716,
+      "loss": 2.3717,
+      "step": 206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009890634987891425,
+      "loss": 2.3894,
+      "step": 207
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009889083991379917,
+      "loss": 2.3358,
+      "step": 208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009887522197517954,
+      "loss": 2.2587,
+      "step": 209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009885949609754693,
+      "loss": 2.3845,
+      "step": 210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988436623156314,
+      "loss": 2.3835,
+      "step": 211
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009882772066440114,
+      "loss": 2.3506,
+      "step": 212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009881167117906276,
+      "loss": 2.4136,
+      "step": 213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009879551389506084,
+      "loss": 2.2729,
+      "step": 214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009877924884807814,
+      "loss": 2.3589,
+      "step": 215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987628760740354,
+      "loss": 2.3277,
+      "step": 216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009874639560909118,
+      "loss": 2.3005,
+      "step": 217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009872980748964202,
+      "loss": 2.3194,
+      "step": 218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987131117523221,
+      "loss": 2.3912,
+      "step": 219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986963084340033,
+      "loss": 2.3419,
+      "step": 220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009867939757179508,
+      "loss": 2.4162,
+      "step": 221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009866237920304443,
+      "loss": 2.3772,
+      "step": 222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009864525336533577,
+      "loss": 2.3452,
+      "step": 223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986280200964908,
+      "loss": 2.2865,
+      "step": 224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009861067943456856,
+      "loss": 2.3081,
+      "step": 225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000985932314178652,
+      "loss": 2.3334,
+      "step": 226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00098575676084914,
+      "loss": 2.315,
+      "step": 227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009855801347448518,
+      "loss": 2.3146,
+      "step": 228
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009854024362558596,
+      "loss": 2.3538,
+      "step": 229
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009852236657746035,
+      "loss": 2.3639,
+      "step": 230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009850438236958911,
+      "loss": 2.3281,
+      "step": 231
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009848629104168966,
+      "loss": 2.3355,
+      "step": 232
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00098468092633716,
+      "loss": 2.3,
+      "step": 233
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009844978718585855,
+      "loss": 2.3636,
+      "step": 234
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009843137473854423,
+      "loss": 2.2318,
+      "step": 235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984128553324362,
+      "loss": 2.3306,
+      "step": 236
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009839422900843383,
+      "loss": 2.2325,
+      "step": 237
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009837549580767261,
+      "loss": 2.2689,
+      "step": 238
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009835665577152411,
+      "loss": 2.3035,
+      "step": 239
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983377089415958,
+      "loss": 2.3346,
+      "step": 240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009831865535973102,
+      "loss": 2.3154,
+      "step": 241
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009829949506800885,
+      "loss": 2.2577,
+      "step": 242
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009828022810874405,
+      "loss": 2.4117,
+      "step": 243
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009826085452448693,
+      "loss": 2.3014,
+      "step": 244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982413743580233,
+      "loss": 2.2751,
+      "step": 245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009822178765237436,
+      "loss": 2.253,
+      "step": 246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009820209445079654,
+      "loss": 2.3355,
+      "step": 247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009818229479678158,
+      "loss": 2.3033,
+      "step": 248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009816238873405615,
+      "loss": 2.4057,
+      "step": 249
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009814237630658207,
+      "loss": 2.2303,
+      "step": 250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00098122257558556,
+      "loss": 2.3482,
+      "step": 251
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009810203253440937,
+      "loss": 2.3479,
+      "step": 252
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009808170127880837,
+      "loss": 2.3441,
+      "step": 253
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980612638366538,
+      "loss": 2.1713,
+      "step": 254
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009804072025308096,
+      "loss": 2.3833,
+      "step": 255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980200705734595,
+      "loss": 2.3403,
+      "step": 256
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009799931484339344,
+      "loss": 2.3305,
+      "step": 257
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009797845310872103,
+      "loss": 2.3547,
+      "step": 258
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009795748541551457,
+      "loss": 2.2834,
+      "step": 259
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979364118100804,
+      "loss": 2.3105,
+      "step": 260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009791523233895875,
+      "loss": 2.2897,
+      "step": 261
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009789394704892364,
+      "loss": 2.2471,
+      "step": 262
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009787255598698282,
+      "loss": 2.4202,
+      "step": 263
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009785105920037758,
+      "loss": 2.2965,
+      "step": 264
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009782945673658275,
+      "loss": 2.293,
+      "step": 265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009780774864330654,
+      "loss": 2.3282,
+      "step": 266
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977859349684904,
+      "loss": 2.3063,
+      "step": 267
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00097764015760309,
+      "loss": 2.2917,
+      "step": 268
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009774199106717004,
+      "loss": 2.3557,
+      "step": 269
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009771986093771417,
+      "loss": 2.3071,
+      "step": 270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009769762542081496,
+      "loss": 2.2727,
+      "step": 271
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000976752845655786,
+      "loss": 2.2913,
+      "step": 272
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009765283842134411,
+      "loss": 2.2409,
+      "step": 273
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009763028703768282,
+      "loss": 2.2369,
+      "step": 274
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009760763046439862,
+      "loss": 2.2042,
+      "step": 275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009758486875152766,
+      "loss": 2.2836,
+      "step": 276
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009756200194933829,
+      "loss": 2.2508,
+      "step": 277
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009753903010833094,
+      "loss": 2.3305,
+      "step": 278
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009751595327923803,
+      "loss": 2.3194,
+      "step": 279
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009749277151302382,
+      "loss": 2.2897,
+      "step": 280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009746948486088435,
+      "loss": 2.2681,
+      "step": 281
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009744609337424727,
+      "loss": 2.2713,
+      "step": 282
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009742259710477177,
+      "loss": 2.254,
+      "step": 283
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009739899610434841,
+      "loss": 2.2027,
+      "step": 284
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009737529042509913,
+      "loss": 2.3847,
+      "step": 285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009735148011937693,
+      "loss": 2.2861,
+      "step": 286
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00097327565239766,
+      "loss": 2.2668,
+      "step": 287
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009730354583908136,
+      "loss": 2.343,
+      "step": 288
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009727942197036895,
+      "loss": 2.2095,
+      "step": 289
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009725519368690539,
+      "loss": 2.3055,
+      "step": 290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009723086104219787,
+      "loss": 2.3363,
+      "step": 291
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009720642408998409,
+      "loss": 2.3139,
+      "step": 292
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009718188288423211,
+      "loss": 2.2223,
+      "step": 293
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009715723747914022,
+      "loss": 2.3221,
+      "step": 294
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009713248792913685,
+      "loss": 2.2285,
+      "step": 295
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009710763428888037,
+      "loss": 2.2287,
+      "step": 296
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009708267661325909,
+      "loss": 2.3259,
+      "step": 297
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009705761495739107,
+      "loss": 2.2508,
+      "step": 298
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009703244937662399,
+      "loss": 2.2585,
+      "step": 299
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009700717992653505,
+      "loss": 2.2574,
+      "step": 300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009698180666293083,
+      "loss": 2.2775,
+      "step": 301
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000969563296418472,
+      "loss": 2.2763,
+      "step": 302
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009693074891954914,
+      "loss": 2.2975,
+      "step": 303
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009690506455253072,
+      "loss": 2.3468,
+      "step": 304
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009687927659751481,
+      "loss": 2.2694,
+      "step": 305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009685338511145312,
+      "loss": 2.2605,
+      "step": 306
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009682739015152598,
+      "loss": 2.1775,
+      "step": 307
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009680129177514226,
+      "loss": 2.2694,
+      "step": 308
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009677509003993915,
+      "loss": 2.211,
+      "step": 309
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009674878500378221,
+      "loss": 2.2258,
+      "step": 310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009672237672476505,
+      "loss": 2.2544,
+      "step": 311
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009669586526120935,
+      "loss": 2.3965,
+      "step": 312
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009666925067166459,
+      "loss": 2.2323,
+      "step": 313
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000966425330149081,
+      "loss": 2.1742,
+      "step": 314
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009661571234994475,
+      "loss": 2.2555,
+      "step": 315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009658878873600691,
+      "loss": 2.3356,
+      "step": 316
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009656176223255438,
+      "loss": 2.2288,
+      "step": 317
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000965346328992741,
+      "loss": 2.2742,
+      "step": 318
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009650740079608014,
+      "loss": 2.2763,
+      "step": 319
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009648006598311353,
+      "loss": 2.2961,
+      "step": 320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009645262852074214,
+      "loss": 2.2465,
+      "step": 321
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009642508846956053,
+      "loss": 2.1229,
+      "step": 322
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009639744589038983,
+      "loss": 2.2165,
+      "step": 323
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009636970084427759,
+      "loss": 2.2164,
+      "step": 324
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009634185339249766,
+      "loss": 2.203,
+      "step": 325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009631390359655003,
+      "loss": 2.2709,
+      "step": 326
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009628585151816074,
+      "loss": 2.3176,
+      "step": 327
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009625769721928172,
+      "loss": 2.38,
+      "step": 328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009622944076209061,
+      "loss": 2.2577,
+      "step": 329
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009620108220899071,
+      "loss": 2.3109,
+      "step": 330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009617262162261075,
+      "loss": 2.3278,
+      "step": 331
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009614405906580486,
+      "loss": 2.2745,
+      "step": 332
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000961153946016523,
+      "loss": 2.296,
+      "step": 333
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000960866282934574,
+      "loss": 2.2488,
+      "step": 334
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009605776020474945,
+      "loss": 2.3646,
+      "step": 335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009602879039928249,
+      "loss": 2.2333,
+      "step": 336
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009599971894103521,
+      "loss": 2.1751,
+      "step": 337
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009597054589421077,
+      "loss": 2.2626,
+      "step": 338
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009594127132323669,
+      "loss": 2.2773,
+      "step": 339
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009591189529276474,
+      "loss": 2.2701,
+      "step": 340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009588241786767072,
+      "loss": 2.3077,
+      "step": 341
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009585283911305436,
+      "loss": 2.3417,
+      "step": 342
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000958231590942392,
+      "loss": 2.2807,
+      "step": 343
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009579337787677238,
+      "loss": 2.2483,
+      "step": 344
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009576349552642456,
+      "loss": 2.1691,
+      "step": 345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009573351210918975,
+      "loss": 2.3125,
+      "step": 346
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009570342769128514,
+      "loss": 2.1757,
+      "step": 347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009567324233915099,
+      "loss": 2.3652,
+      "step": 348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009564295611945047,
+      "loss": 2.271,
+      "step": 349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000956125690990695,
+      "loss": 2.2641,
+      "step": 350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009558208134511665,
+      "loss": 2.3747,
+      "step": 351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009555149292492289,
+      "loss": 2.3159,
+      "step": 352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009552080390604159,
+      "loss": 2.3564,
+      "step": 353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009549001435624823,
+      "loss": 2.2651,
+      "step": 354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009545912434354029,
+      "loss": 2.2761,
+      "step": 355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009542813393613721,
+      "loss": 2.2919,
+      "step": 356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009539704320248006,
+      "loss": 2.2851,
+      "step": 357
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009536585221123151,
+      "loss": 2.2484,
+      "step": 358
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009533456103127565,
+      "loss": 2.2884,
+      "step": 359
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000953031697317178,
+      "loss": 2.256,
+      "step": 360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009527167838188445,
+      "loss": 2.2046,
+      "step": 361
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009524008705132299,
+      "loss": 2.2102,
+      "step": 362
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009520839580980166,
+      "loss": 2.2896,
+      "step": 363
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009517660472730929,
+      "loss": 2.2256,
+      "step": 364
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009514471387405526,
+      "loss": 2.246,
+      "step": 365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009511272332046926,
+      "loss": 2.1495,
+      "step": 366
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009508063313720119,
+      "loss": 2.2502,
+      "step": 367
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009504844339512095,
+      "loss": 2.3385,
+      "step": 368
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009501615416531835,
+      "loss": 2.2837,
+      "step": 369
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009498376551910285,
+      "loss": 2.2455,
+      "step": 370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009495127752800352,
+      "loss": 2.3203,
+      "step": 371
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009491869026376882,
+      "loss": 2.2772,
+      "step": 372
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009488600379836648,
+      "loss": 2.3466,
+      "step": 373
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009485321820398321,
+      "loss": 2.3501,
+      "step": 374
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009482033355302475,
+      "loss": 2.3031,
+      "step": 375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009478734991811556,
+      "loss": 2.17,
+      "step": 376
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009475426737209871,
+      "loss": 2.3416,
+      "step": 377
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000947210859880357,
+      "loss": 2.2592,
+      "step": 378
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009468780583920631,
+      "loss": 2.1369,
+      "step": 379
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009465442699910846,
+      "loss": 2.2661,
+      "step": 380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009462094954145801,
+      "loss": 2.2571,
+      "step": 381
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009458737354018859,
+      "loss": 2.2153,
+      "step": 382
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000945536990694515,
+      "loss": 2.2434,
+      "step": 383
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009451992620361551,
+      "loss": 2.3154,
+      "step": 384
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009448605501726664,
+      "loss": 2.2469,
+      "step": 385
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000944520855852081,
+      "loss": 2.2281,
+      "step": 386
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009441801798246002,
+      "loss": 2.281,
+      "step": 387
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009438385228425939,
+      "loss": 2.2887,
+      "step": 388
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009434958856605982,
+      "loss": 2.2794,
+      "step": 389
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009431522690353137,
+      "loss": 2.2506,
+      "step": 390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009428076737256044,
+      "loss": 2.2134,
+      "step": 391
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009424621004924954,
+      "loss": 2.2473,
+      "step": 392
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009421155500991719,
+      "loss": 2.2347,
+      "step": 393
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009417680233109767,
+      "loss": 2.3209,
+      "step": 394
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000941419520895409,
+      "loss": 2.2,
+      "step": 395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009410700436221229,
+      "loss": 2.247,
+      "step": 396
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009407195922629252,
+      "loss": 2.2077,
+      "step": 397
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000940368167591774,
+      "loss": 2.1964,
+      "step": 398
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009400157703847769,
+      "loss": 2.1988,
+      "step": 399
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009396624014201895,
+      "loss": 2.3011,
+      "step": 400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000939308061478413,
+      "loss": 2.1749,
+      "step": 401
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009389527513419935,
+      "loss": 2.2988,
+      "step": 402
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009385964717956195,
+      "loss": 2.2302,
+      "step": 403
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009382392236261201,
+      "loss": 2.3081,
+      "step": 404
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009378810076224644,
+      "loss": 2.2077,
+      "step": 405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009375218245757582,
+      "loss": 2.2887,
+      "step": 406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009371616752792432,
+      "loss": 2.2108,
+      "step": 407
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009368005605282949,
+      "loss": 2.1829,
+      "step": 408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009364384811204212,
+      "loss": 2.2834,
+      "step": 409
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00093607543785526,
+      "loss": 2.2851,
+      "step": 410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009357114315345787,
+      "loss": 2.2092,
+      "step": 411
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009353464629622705,
+      "loss": 2.2041,
+      "step": 412
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009349805329443544,
+      "loss": 2.2736,
+      "step": 413
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009346136422889724,
+      "loss": 2.2924,
+      "step": 414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009342457918063882,
+      "loss": 2.2213,
+      "step": 415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009338769823089853,
+      "loss": 2.3071,
+      "step": 416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009335072146112648,
+      "loss": 2.149,
+      "step": 417
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009331364895298444,
+      "loss": 2.1683,
+      "step": 418
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009327648078834559,
+      "loss": 2.2006,
+      "step": 419
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009323921704929434,
+      "loss": 2.2109,
+      "step": 420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009320185781812623,
+      "loss": 2.2834,
+      "step": 421
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009316440317734762,
+      "loss": 2.1878,
+      "step": 422
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009312685320967565,
+      "loss": 2.218,
+      "step": 423
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009308920799803793,
+      "loss": 2.2216,
+      "step": 424
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009305146762557246,
+      "loss": 2.2394,
+      "step": 425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009301363217562736,
+      "loss": 2.2487,
+      "step": 426
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009297570173176074,
+      "loss": 2.2078,
+      "step": 427
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000929376763777405,
+      "loss": 2.2533,
+      "step": 428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009289955619754413,
+      "loss": 2.2649,
+      "step": 429
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009286134127535859,
+      "loss": 2.2191,
+      "step": 430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009282303169558,
+      "loss": 2.2626,
+      "step": 431
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009278462754281359,
+      "loss": 2.2931,
+      "step": 432
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009274612890187342,
+      "loss": 2.2912,
+      "step": 433
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009270753585778222,
+      "loss": 2.2077,
+      "step": 434
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009266884849577124,
+      "loss": 2.2147,
+      "step": 435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009263006690127998,
+      "loss": 2.3138,
+      "step": 436
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009259119115995609,
+      "loss": 2.2508,
+      "step": 437
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009255222135765511,
+      "loss": 2.2129,
+      "step": 438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009251315758044032,
+      "loss": 2.2061,
+      "step": 439
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009247399991458255,
+      "loss": 2.2067,
+      "step": 440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009243474844655996,
+      "loss": 2.2334,
+      "step": 441
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009239540326305791,
+      "loss": 2.2314,
+      "step": 442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009235596445096864,
+      "loss": 2.2598,
+      "step": 443
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009231643209739127,
+      "loss": 2.2386,
+      "step": 444
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009227680628963145,
+      "loss": 2.2103,
+      "step": 445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000922370871152012,
+      "loss": 2.2092,
+      "step": 446
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009219727466181877,
+      "loss": 2.294,
+      "step": 447
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009215736901740841,
+      "loss": 2.2355,
+      "step": 448
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009211737027010016,
+      "loss": 2.2874,
+      "step": 449
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009207727850822971,
+      "loss": 2.2229,
+      "step": 450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009203709382033814,
+      "loss": 2.2109,
+      "step": 451
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009199681629517173,
+      "loss": 2.2711,
+      "step": 452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009195644602168184,
+      "loss": 2.1966,
+      "step": 453
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009191598308902464,
+      "loss": 2.2516,
+      "step": 454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009187542758656091,
+      "loss": 2.2442,
+      "step": 455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009183477960385591,
+      "loss": 2.2503,
+      "step": 456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009179403923067912,
+      "loss": 2.171,
+      "step": 457
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009175320655700406,
+      "loss": 2.2901,
+      "step": 458
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009171228167300805,
+      "loss": 2.2353,
+      "step": 459
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009167126466907215,
+      "loss": 2.1794,
+      "step": 460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009163015563578074,
+      "loss": 2.16,
+      "step": 461
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009158895466392158,
+      "loss": 2.2555,
+      "step": 462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009154766184448535,
+      "loss": 2.224,
+      "step": 463
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009150627726866568,
+      "loss": 2.2725,
+      "step": 464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000914648010278587,
+      "loss": 2.1611,
+      "step": 465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009142323321366315,
+      "loss": 2.2017,
+      "step": 466
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009138157391787986,
+      "loss": 2.2031,
+      "step": 467
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009133982323251177,
+      "loss": 2.2246,
+      "step": 468
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009129798124976365,
+      "loss": 2.1664,
+      "step": 469
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009125604806204187,
+      "loss": 2.17,
+      "step": 470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009121402376195421,
+      "loss": 2.1831,
+      "step": 471
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009117190844230972,
+      "loss": 2.2578,
+      "step": 472
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009112970219611841,
+      "loss": 2.225,
+      "step": 473
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009108740511659115,
+      "loss": 2.2217,
+      "step": 474
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009104501729713935,
+      "loss": 2.1631,
+      "step": 475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009100253883137488,
+      "loss": 2.2873,
+      "step": 476
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009095996981310974,
+      "loss": 2.1761,
+      "step": 477
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009091731033635596,
+      "loss": 2.1793,
+      "step": 478
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009087456049532529,
+      "loss": 2.2622,
+      "step": 479
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009083172038442914,
+      "loss": 2.3143,
+      "step": 480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009078879009827817,
+      "loss": 2.2381,
+      "step": 481
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009074576973168223,
+      "loss": 2.2609,
+      "step": 482
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009070265937965015,
+      "loss": 2.2649,
+      "step": 483
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009065945913738942,
+      "loss": 2.1895,
+      "step": 484
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009061616910030609,
+      "loss": 2.23,
+      "step": 485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009057278936400453,
+      "loss": 2.1688,
+      "step": 486
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009052932002428715,
+      "loss": 2.2226,
+      "step": 487
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009048576117715435,
+      "loss": 2.2262,
+      "step": 488
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009044211291880407,
+      "loss": 2.2647,
+      "step": 489
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000903983753456318,
+      "loss": 2.2519,
+      "step": 490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009035454855423026,
+      "loss": 2.1751,
+      "step": 491
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009031063264138922,
+      "loss": 2.2361,
+      "step": 492
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009026662770409522,
+      "loss": 2.2578,
+      "step": 493
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009022253383953147,
+      "loss": 2.1942,
+      "step": 494
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009017835114507753,
+      "loss": 2.2252,
+      "step": 495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009013407971830914,
+      "loss": 2.2865,
+      "step": 496
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009008971965699801,
+      "loss": 2.2129,
+      "step": 497
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009004527105911163,
+      "loss": 2.2631,
+      "step": 498
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009000073402281295,
+      "loss": 2.2027,
+      "step": 499
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008995610864646028,
+      "loss": 2.2098,
+      "step": 500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008991139502860703,
+      "loss": 2.2209,
+      "step": 501
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008986659326800146,
+      "loss": 2.2619,
+      "step": 502
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008982170346358651,
+      "loss": 2.1651,
+      "step": 503
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008977672571449956,
+      "loss": 2.256,
+      "step": 504
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008973166012007217,
+      "loss": 2.1559,
+      "step": 505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008968650677982998,
+      "loss": 2.2652,
+      "step": 506
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008964126579349236,
+      "loss": 2.2212,
+      "step": 507
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008959593726097226,
+      "loss": 2.1773,
+      "step": 508
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008955052128237596,
+      "loss": 2.2327,
+      "step": 509
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008950501795800288,
+      "loss": 2.1808,
+      "step": 510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008945942738834532,
+      "loss": 2.274,
+      "step": 511
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008941374967408826,
+      "loss": 2.2139,
+      "step": 512
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008936798491610916,
+      "loss": 2.2711,
+      "step": 513
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008932213321547768,
+      "loss": 2.1036,
+      "step": 514
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008927619467345554,
+      "loss": 2.1724,
+      "step": 515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008923016939149615,
+      "loss": 2.2723,
+      "step": 516
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008918405747124458,
+      "loss": 2.2793,
+      "step": 517
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008913785901453721,
+      "loss": 2.2271,
+      "step": 518
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000890915741234015,
+      "loss": 2.2471,
+      "step": 519
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008904520290005582,
+      "loss": 2.2215,
+      "step": 520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000889987454469092,
+      "loss": 2.2784,
+      "step": 521
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008895220186656111,
+      "loss": 2.1656,
+      "step": 522
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008890557226180122,
+      "loss": 2.2619,
+      "step": 523
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008885885673560921,
+      "loss": 2.1962,
+      "step": 524
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008881205539115444,
+      "loss": 2.1786,
+      "step": 525
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008876516833179589,
+      "loss": 2.1781,
+      "step": 526
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008871819566108177,
+      "loss": 2.144,
+      "step": 527
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000886711374827494,
+      "loss": 2.2025,
+      "step": 528
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008862399390072491,
+      "loss": 2.2369,
+      "step": 529
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008857676501912305,
+      "loss": 2.0818,
+      "step": 530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008852945094224697,
+      "loss": 2.1912,
+      "step": 531
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008848205177458795,
+      "loss": 2.2445,
+      "step": 532
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008843456762082518,
+      "loss": 2.2182,
+      "step": 533
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008838699858582557,
+      "loss": 2.1975,
+      "step": 534
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008833934477464347,
+      "loss": 2.1432,
+      "step": 535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008829160629252045,
+      "loss": 2.212,
+      "step": 536
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008824378324488509,
+      "loss": 2.2471,
+      "step": 537
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008819587573735268,
+      "loss": 2.272,
+      "step": 538
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008814788387572513,
+      "loss": 2.2311,
+      "step": 539
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008809980776599053,
+      "loss": 2.2051,
+      "step": 540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008805164751432312,
+      "loss": 2.2293,
+      "step": 541
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008800340322708292,
+      "loss": 2.265,
+      "step": 542
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008795507501081555,
+      "loss": 2.194,
+      "step": 543
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008790666297225196,
+      "loss": 2.2105,
+      "step": 544
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008785816721830829,
+      "loss": 2.2072,
+      "step": 545
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008780958785608546,
+      "loss": 2.2809,
+      "step": 546
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008776092499286912,
+      "loss": 2.1998,
+      "step": 547
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008771217873612929,
+      "loss": 2.1872,
+      "step": 548
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008766334919352017,
+      "loss": 2.2441,
+      "step": 549
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008761443647287987,
+      "loss": 2.2249,
+      "step": 550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008756544068223026,
+      "loss": 2.2217,
+      "step": 551
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008751636192977659,
+      "loss": 2.2802,
+      "step": 552
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008746720032390737,
+      "loss": 2.1561,
+      "step": 553
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008741795597319408,
+      "loss": 2.2329,
+      "step": 554
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008736862898639095,
+      "loss": 2.1594,
+      "step": 555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008731921947243468,
+      "loss": 2.2144,
+      "step": 556
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008726972754044427,
+      "loss": 2.2664,
+      "step": 557
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008722015329972069,
+      "loss": 2.1929,
+      "step": 558
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008717049685974672,
+      "loss": 2.2569,
+      "step": 559
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008712075833018665,
+      "loss": 2.1967,
+      "step": 560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008707093782088608,
+      "loss": 2.1553,
+      "step": 561
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008702103544187167,
+      "loss": 2.2275,
+      "step": 562
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008697105130335085,
+      "loss": 2.1802,
+      "step": 563
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008692098551571164,
+      "loss": 2.1868,
+      "step": 564
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008687083818952235,
+      "loss": 2.1491,
+      "step": 565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008682060943553143,
+      "loss": 2.1933,
+      "step": 566
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008677029936466707,
+      "loss": 2.2891,
+      "step": 567
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008671990808803711,
+      "loss": 2.2056,
+      "step": 568
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008666943571692871,
+      "loss": 2.2532,
+      "step": 569
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008661888236280813,
+      "loss": 2.1974,
+      "step": 570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008656824813732045,
+      "loss": 2.142,
+      "step": 571
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000865175331522894,
+      "loss": 2.3025,
+      "step": 572
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008646673751971703,
+      "loss": 2.2308,
+      "step": 573
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000864158613517835,
+      "loss": 2.1887,
+      "step": 574
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008636490476084681,
+      "loss": 2.136,
+      "step": 575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008631386785944264,
+      "loss": 2.1863,
+      "step": 576
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008626275076028397,
+      "loss": 2.2455,
+      "step": 577
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008621155357626091,
+      "loss": 2.1437,
+      "step": 578
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008616027642044042,
+      "loss": 2.1355,
+      "step": 579
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000861089194060661,
+      "loss": 2.207,
+      "step": 580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000860574826465579,
+      "loss": 2.1666,
+      "step": 581
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008600596625551191,
+      "loss": 2.1723,
+      "step": 582
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008595437034670006,
+      "loss": 2.1794,
+      "step": 583
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008590269503406985,
+      "loss": 2.2614,
+      "step": 584
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008585094043174423,
+      "loss": 2.208,
+      "step": 585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008579910665402118,
+      "loss": 2.1768,
+      "step": 586
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000857471938153736,
+      "loss": 2.1281,
+      "step": 587
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008569520203044892,
+      "loss": 2.131,
+      "step": 588
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008564313141406901,
+      "loss": 2.171,
+      "step": 589
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008559098208122973,
+      "loss": 2.1756,
+      "step": 590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008553875414710089,
+      "loss": 2.1994,
+      "step": 591
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008548644772702579,
+      "loss": 2.1804,
+      "step": 592
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008543406293652116,
+      "loss": 2.1977,
+      "step": 593
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008538159989127671,
+      "loss": 2.0861,
+      "step": 594
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008532905870715505,
+      "loss": 2.2325,
+      "step": 595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008527643950019131,
+      "loss": 2.2055,
+      "step": 596
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008522374238659296,
+      "loss": 2.2461,
+      "step": 597
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008517096748273951,
+      "loss": 2.2,
+      "step": 598
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008511811490518227,
+      "loss": 2.2306,
+      "step": 599
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008506518477064405,
+      "loss": 2.1303,
+      "step": 600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008501217719601903,
+      "loss": 2.1495,
+      "step": 601
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008495909229837233,
+      "loss": 2.2049,
+      "step": 602
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000849059301949399,
+      "loss": 2.1882,
+      "step": 603
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008485269100312812,
+      "loss": 2.2052,
+      "step": 604
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008479937484051368,
+      "loss": 2.2621,
+      "step": 605
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008474598182484323,
+      "loss": 2.2521,
+      "step": 606
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008469251207403317,
+      "loss": 2.2177,
+      "step": 607
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008463896570616934,
+      "loss": 2.158,
+      "step": 608
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008458534283950678,
+      "loss": 2.3248,
+      "step": 609
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008453164359246952,
+      "loss": 2.0971,
+      "step": 610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008447786808365022,
+      "loss": 2.162,
+      "step": 611
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008442401643181,
+      "loss": 2.1747,
+      "step": 612
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008437008875587811,
+      "loss": 2.1395,
+      "step": 613
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008431608517495171,
+      "loss": 2.1679,
+      "step": 614
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008426200580829561,
+      "loss": 2.2641,
+      "step": 615
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008420785077534195,
+      "loss": 2.1164,
+      "step": 616
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008415362019569001,
+      "loss": 2.2436,
+      "step": 617
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008409931418910591,
+      "loss": 2.1564,
+      "step": 618
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008404493287552232,
+      "loss": 2.2184,
+      "step": 619
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008399047637503825,
+      "loss": 2.1982,
+      "step": 620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008393594480791875,
+      "loss": 2.2464,
+      "step": 621
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008388133829459463,
+      "loss": 2.2026,
+      "step": 622
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008382665695566227,
+      "loss": 2.1173,
+      "step": 623
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008377190091188324,
+      "loss": 2.186,
+      "step": 624
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008371707028418413,
+      "loss": 2.1548,
+      "step": 625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008366216519365621,
+      "loss": 2.1379,
+      "step": 626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008360718576155525,
+      "loss": 2.2193,
+      "step": 627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008355213210930118,
+      "loss": 2.1953,
+      "step": 628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008349700435847778,
+      "loss": 2.2436,
+      "step": 629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008344180263083256,
+      "loss": 2.1928,
+      "step": 630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000833865270482764,
+      "loss": 2.2262,
+      "step": 631
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008333117773288324,
+      "loss": 2.2562,
+      "step": 632
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008327575480688985,
+      "loss": 2.238,
+      "step": 633
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000832202583926956,
+      "loss": 2.1439,
+      "step": 634
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008316468861286217,
+      "loss": 2.2216,
+      "step": 635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008310904559011323,
+      "loss": 2.2405,
+      "step": 636
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008305332944733419,
+      "loss": 2.1667,
+      "step": 637
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008299754030757202,
+      "loss": 2.2452,
+      "step": 638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008294167829403481,
+      "loss": 2.1824,
+      "step": 639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008288574353009164,
+      "loss": 2.1594,
+      "step": 640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008282973613927225,
+      "loss": 2.1116,
+      "step": 641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008277365624526675,
+      "loss": 2.1681,
+      "step": 642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008271750397192541,
+      "loss": 2.1704,
+      "step": 643
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008266127944325832,
+      "loss": 2.1805,
+      "step": 644
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008260498278343513,
+      "loss": 2.2577,
+      "step": 645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008254861411678485,
+      "loss": 2.2338,
+      "step": 646
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008249217356779544,
+      "loss": 2.2215,
+      "step": 647
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008243566126111363,
+      "loss": 2.1632,
+      "step": 648
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008237907732154466,
+      "loss": 2.1502,
+      "step": 649
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008232242187405194,
+      "loss": 2.2104,
+      "step": 650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000822656950437568,
+      "loss": 2.1317,
+      "step": 651
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008220889695593823,
+      "loss": 2.25,
+      "step": 652
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008215202773603259,
+      "loss": 2.1966,
+      "step": 653
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008209508750963328,
+      "loss": 2.2209,
+      "step": 654
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008203807640249062,
+      "loss": 2.2523,
+      "step": 655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008198099454051136,
+      "loss": 2.2647,
+      "step": 656
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008192384204975857,
+      "loss": 2.277,
+      "step": 657
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000818666190564513,
+      "loss": 2.1478,
+      "step": 658
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008180932568696426,
+      "loss": 2.1859,
+      "step": 659
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008175196206782764,
+      "loss": 2.1989,
+      "step": 660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008169452832572675,
+      "loss": 2.2663,
+      "step": 661
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008163702458750173,
+      "loss": 2.1861,
+      "step": 662
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008157945098014734,
+      "loss": 2.1911,
+      "step": 663
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008152180763081267,
+      "loss": 2.1608,
+      "step": 664
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008146409466680076,
+      "loss": 2.1908,
+      "step": 665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008140631221556845,
+      "loss": 2.1032,
+      "step": 666
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008134846040472599,
+      "loss": 2.2187,
+      "step": 667
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008129053936203688,
+      "loss": 2.2952,
+      "step": 668
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008123254921541745,
+      "loss": 2.1766,
+      "step": 669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008117449009293668,
+      "loss": 2.2472,
+      "step": 670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008111636212281586,
+      "loss": 2.2333,
+      "step": 671
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008105816543342833,
+      "loss": 2.1747,
+      "step": 672
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008099990015329919,
+      "loss": 2.1659,
+      "step": 673
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008094156641110504,
+      "loss": 2.1538,
+      "step": 674
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008088316433567369,
+      "loss": 2.1974,
+      "step": 675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008082469405598378,
+      "loss": 2.1957,
+      "step": 676
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008076615570116468,
+      "loss": 2.2564,
+      "step": 677
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008070754940049603,
+      "loss": 2.1155,
+      "step": 678
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008064887528340756,
+      "loss": 2.1979,
+      "step": 679
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008059013347947874,
+      "loss": 2.1373,
+      "step": 680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008053132411843857,
+      "loss": 2.1728,
+      "step": 681
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008047244733016521,
+      "loss": 2.2215,
+      "step": 682
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008041350324468573,
+      "loss": 2.1678,
+      "step": 683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008035449199217583,
+      "loss": 2.1426,
+      "step": 684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008029541370295957,
+      "loss": 2.2046,
+      "step": 685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008023626850750903,
+      "loss": 2.1192,
+      "step": 686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008017705653644406,
+      "loss": 2.1323,
+      "step": 687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008011777792053195,
+      "loss": 2.2124,
+      "step": 688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008005843279068725,
+      "loss": 2.1562,
+      "step": 689
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000799990212779713,
+      "loss": 2.1599,
+      "step": 690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007993954351359214,
+      "loss": 2.2385,
+      "step": 691
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007987999962890406,
+      "loss": 2.2231,
+      "step": 692
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007982038975540742,
+      "loss": 2.1634,
+      "step": 693
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007976071402474826,
+      "loss": 2.2676,
+      "step": 694
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007970097256871811,
+      "loss": 2.1306,
+      "step": 695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007964116551925364,
+      "loss": 2.1421,
+      "step": 696
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007958129300843637,
+      "loss": 2.0763,
+      "step": 697
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007952135516849239,
+      "loss": 2.2168,
+      "step": 698
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007946135213179207,
+      "loss": 2.1382,
+      "step": 699
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007940128403084977,
+      "loss": 2.1505,
+      "step": 700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007934115099832355,
+      "loss": 2.185,
+      "step": 701
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007928095316701483,
+      "loss": 2.1611,
+      "step": 702
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007922069066986819,
+      "loss": 2.2609,
+      "step": 703
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007916036363997097,
+      "loss": 2.1953,
+      "step": 704
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007909997221055308,
+      "loss": 2.1479,
+      "step": 705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007903951651498658,
+      "loss": 2.1264,
+      "step": 706
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007897899668678557,
+      "loss": 2.219,
+      "step": 707
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007891841285960566,
+      "loss": 2.1195,
+      "step": 708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007885776516724388,
+      "loss": 2.1848,
+      "step": 709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007879705374363831,
+      "loss": 2.2129,
+      "step": 710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000787362787228677,
+      "loss": 2.1478,
+      "step": 711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007867544023915134,
+      "loss": 2.0897,
+      "step": 712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007861453842684861,
+      "loss": 2.16,
+      "step": 713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007855357342045882,
+      "loss": 2.2531,
+      "step": 714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007849254535462074,
+      "loss": 2.0659,
+      "step": 715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007843145436411252,
+      "loss": 2.1225,
+      "step": 716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007837030058385117,
+      "loss": 2.1785,
+      "step": 717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007830908414889246,
+      "loss": 2.2098,
+      "step": 718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007824780519443046,
+      "loss": 2.1784,
+      "step": 719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007818646385579735,
+      "loss": 2.1584,
+      "step": 720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007812506026846307,
+      "loss": 2.1593,
+      "step": 721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007806359456803504,
+      "loss": 2.1489,
+      "step": 722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007800206689025785,
+      "loss": 2.1993,
+      "step": 723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007794047737101297,
+      "loss": 2.1205,
+      "step": 724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007787882614631843,
+      "loss": 2.2107,
+      "step": 725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007781711335232856,
+      "loss": 2.1552,
+      "step": 726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007775533912533363,
+      "loss": 2.209,
+      "step": 727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007769350360175962,
+      "loss": 2.1024,
+      "step": 728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007763160691816784,
+      "loss": 2.1208,
+      "step": 729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000775696492112547,
+      "loss": 2.1872,
+      "step": 730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007750763061785137,
+      "loss": 2.0808,
+      "step": 731
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000774455512749235,
+      "loss": 2.219,
+      "step": 732
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007738341131957085,
+      "loss": 2.1176,
+      "step": 733
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000773212108890271,
+      "loss": 2.1331,
+      "step": 734
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007725895012065947,
+      "loss": 2.2408,
+      "step": 735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007719662915196844,
+      "loss": 2.1412,
+      "step": 736
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007713424812058736,
+      "loss": 2.2169,
+      "step": 737
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007707180716428237,
+      "loss": 2.1561,
+      "step": 738
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007700930642095184,
+      "loss": 2.2162,
+      "step": 739
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007694674602862621,
+      "loss": 2.1857,
+      "step": 740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007688412612546769,
+      "loss": 2.1696,
+      "step": 741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007682144684976983,
+      "loss": 2.1561,
+      "step": 742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007675870833995739,
+      "loss": 2.1691,
+      "step": 743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007669591073458592,
+      "loss": 2.1969,
+      "step": 744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007663305417234146,
+      "loss": 2.2279,
+      "step": 745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007657013879204022,
+      "loss": 2.166,
+      "step": 746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007650716473262842,
+      "loss": 2.218,
+      "step": 747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007644413213318177,
+      "loss": 2.1274,
+      "step": 748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007638104113290531,
+      "loss": 2.2724,
+      "step": 749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007631789187113303,
+      "loss": 2.1528,
+      "step": 750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000762546844873276,
+      "loss": 2.0795,
+      "step": 751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007619141912108007,
+      "loss": 2.2067,
+      "step": 752
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000761280959121095,
+      "loss": 2.2099,
+      "step": 753
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007606471500026273,
+      "loss": 2.1503,
+      "step": 754
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007600127652551401,
+      "loss": 2.2444,
+      "step": 755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007593778062796472,
+      "loss": 2.1515,
+      "step": 756
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000758742274478431,
+      "loss": 2.169,
+      "step": 757
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007581061712550381,
+      "loss": 2.1916,
+      "step": 758
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007574694980142779,
+      "loss": 2.1518,
+      "step": 759
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007568322561622183,
+      "loss": 2.1252,
+      "step": 760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007561944471061826,
+      "loss": 2.1754,
+      "step": 761
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007555560722547475,
+      "loss": 2.15,
+      "step": 762
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007549171330177387,
+      "loss": 2.1921,
+      "step": 763
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007542776308062285,
+      "loss": 2.1675,
+      "step": 764
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007536375670325325,
+      "loss": 2.232,
+      "step": 765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007529969431102063,
+      "loss": 2.1594,
+      "step": 766
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000752355760454043,
+      "loss": 2.2349,
+      "step": 767
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007517140204800693,
+      "loss": 2.1656,
+      "step": 768
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007510717246055425,
+      "loss": 2.2552,
+      "step": 769
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007504288742489482,
+      "loss": 2.2704,
+      "step": 770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007497854708299963,
+      "loss": 2.1046,
+      "step": 771
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007491415157696178,
+      "loss": 2.1511,
+      "step": 772
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007484970104899623,
+      "loss": 2.1416,
+      "step": 773
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007478519564143945,
+      "loss": 2.1962,
+      "step": 774
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000747206354967491,
+      "loss": 2.1665,
+      "step": 775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007465602075750373,
+      "loss": 2.1639,
+      "step": 776
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007459135156640247,
+      "loss": 2.1764,
+      "step": 777
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007452662806626468,
+      "loss": 2.2926,
+      "step": 778
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007446185040002967,
+      "loss": 2.1464,
+      "step": 779
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007439701871075642,
+      "loss": 2.1855,
+      "step": 780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007433213314162313,
+      "loss": 2.2736,
+      "step": 781
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007426719383592705,
+      "loss": 2.1297,
+      "step": 782
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000742022009370841,
+      "loss": 2.1919,
+      "step": 783
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007413715458862855,
+      "loss": 2.1625,
+      "step": 784
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007407205493421272,
+      "loss": 2.211,
+      "step": 785
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007400690211760661,
+      "loss": 2.1379,
+      "step": 786
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007394169628269771,
+      "loss": 2.1787,
+      "step": 787
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007387643757349051,
+      "loss": 2.1285,
+      "step": 788
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007381112613410635,
+      "loss": 2.29,
+      "step": 789
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007374576210878298,
+      "loss": 2.2552,
+      "step": 790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007368034564187425,
+      "loss": 2.2654,
+      "step": 791
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007361487687784989,
+      "loss": 2.0799,
+      "step": 792
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007354935596129513,
+      "loss": 2.2196,
+      "step": 793
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000734837830369103,
+      "loss": 2.0836,
+      "step": 794
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007341815824951066,
+      "loss": 2.1537,
+      "step": 795
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007335248174402597,
+      "loss": 2.1832,
+      "step": 796
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007328675366550023,
+      "loss": 2.2419,
+      "step": 797
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007322097415909134,
+      "loss": 2.1933,
+      "step": 798
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007315514337007071,
+      "loss": 2.0782,
+      "step": 799
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007308926144382312,
+      "loss": 2.113,
+      "step": 800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007302332852584619,
+      "loss": 2.1425,
+      "step": 801
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007295734476175018,
+      "loss": 2.1862,
+      "step": 802
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007289131029725768,
+      "loss": 2.1901,
+      "step": 803
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007282522527820319,
+      "loss": 2.1201,
+      "step": 804
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000727590898505329,
+      "loss": 2.131,
+      "step": 805
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007269290416030429,
+      "loss": 2.1551,
+      "step": 806
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000726266683536859,
+      "loss": 2.153,
+      "step": 807
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007256038257695687,
+      "loss": 2.2069,
+      "step": 808
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007249404697650678,
+      "loss": 2.1093,
+      "step": 809
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007242766169883518,
+      "loss": 2.0629,
+      "step": 810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007236122689055138,
+      "loss": 2.1969,
+      "step": 811
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007229474269837401,
+      "loss": 2.2345,
+      "step": 812
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007222820926913085,
+      "loss": 2.105,
+      "step": 813
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007216162674975833,
+      "loss": 2.2131,
+      "step": 814
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007209499528730138,
+      "loss": 2.1747,
+      "step": 815
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007202831502891294,
+      "loss": 2.1874,
+      "step": 816
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007196158612185375,
+      "loss": 2.1081,
+      "step": 817
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007189480871349201,
+      "loss": 2.1374,
+      "step": 818
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007182798295130299,
+      "loss": 2.1646,
+      "step": 819
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007176110898286878,
+      "loss": 2.2222,
+      "step": 820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007169418695587791,
+      "loss": 2.1437,
+      "step": 821
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007162721701812506,
+      "loss": 2.1872,
+      "step": 822
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007156019931751072,
+      "loss": 2.1858,
+      "step": 823
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007149313400204082,
+      "loss": 2.1829,
+      "step": 824
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007142602121982653,
+      "loss": 2.1406,
+      "step": 825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007135886111908379,
+      "loss": 2.2401,
+      "step": 826
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007129165384813303,
+      "loss": 2.2078,
+      "step": 827
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007122439955539888,
+      "loss": 2.1567,
+      "step": 828
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007115709838940983,
+      "loss": 2.1342,
+      "step": 829
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007108975049879785,
+      "loss": 2.2195,
+      "step": 830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007102235603229814,
+      "loss": 2.1549,
+      "step": 831
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000709549151387487,
+      "loss": 2.121,
+      "step": 832
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007088742796709013,
+      "loss": 2.1771,
+      "step": 833
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000708198946663652,
+      "loss": 2.1839,
+      "step": 834
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007075231538571856,
+      "loss": 2.2628,
+      "step": 835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007068469027439641,
+      "loss": 2.1632,
+      "step": 836
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007061701948174613,
+      "loss": 2.1681,
+      "step": 837
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007054930315721606,
+      "loss": 2.2204,
+      "step": 838
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007048154145035501,
+      "loss": 2.2537,
+      "step": 839
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007041373451081207,
+      "loss": 2.1812,
+      "step": 840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007034588248833621,
+      "loss": 2.1853,
+      "step": 841
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007027798553277595,
+      "loss": 2.0757,
+      "step": 842
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007021004379407909,
+      "loss": 2.1556,
+      "step": 843
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007014205742229227,
+      "loss": 2.107,
+      "step": 844
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007007402656756072,
+      "loss": 2.2073,
+      "step": 845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007000595138012797,
+      "loss": 2.1769,
+      "step": 846
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006993783201033535,
+      "loss": 2.1562,
+      "step": 847
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006986966860862182,
+      "loss": 2.1008,
+      "step": 848
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000698014613255236,
+      "loss": 2.0812,
+      "step": 849
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006973321031167382,
+      "loss": 2.1189,
+      "step": 850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006966491571780216,
+      "loss": 2.1929,
+      "step": 851
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006959657769473453,
+      "loss": 2.1457,
+      "step": 852
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000695281963933928,
+      "loss": 2.1729,
+      "step": 853
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006945977196479438,
+      "loss": 2.1982,
+      "step": 854
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006939130456005196,
+      "loss": 2.1903,
+      "step": 855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006932279433037311,
+      "loss": 2.1931,
+      "step": 856
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006925424142705997,
+      "loss": 2.0886,
+      "step": 857
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006918564600150896,
+      "loss": 2.1181,
+      "step": 858
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006911700820521042,
+      "loss": 2.2031,
+      "step": 859
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006904832818974818,
+      "loss": 2.2388,
+      "step": 860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006897960610679939,
+      "loss": 2.1552,
+      "step": 861
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006891084210813407,
+      "loss": 2.1822,
+      "step": 862
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006884203634561483,
+      "loss": 2.2175,
+      "step": 863
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006877318897119651,
+      "loss": 2.1153,
+      "step": 864
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006870430013692579,
+      "loss": 2.226,
+      "step": 865
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006863536999494101,
+      "loss": 2.1628,
+      "step": 866
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006856639869747167,
+      "loss": 2.1437,
+      "step": 867
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006849738639683818,
+      "loss": 2.1477,
+      "step": 868
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000684283332454515,
+      "loss": 2.1743,
+      "step": 869
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006835923939581281,
+      "loss": 2.1171,
+      "step": 870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006829010500051318,
+      "loss": 2.2377,
+      "step": 871
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006822093021223321,
+      "loss": 2.1855,
+      "step": 872
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006815171518374268,
+      "loss": 2.1789,
+      "step": 873
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006808246006790031,
+      "loss": 2.1857,
+      "step": 874
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006801316501765329,
+      "loss": 2.2518,
+      "step": 875
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006794383018603704,
+      "loss": 2.1503,
+      "step": 876
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006787445572617481,
+      "loss": 2.1739,
+      "step": 877
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006780504179127734,
+      "loss": 2.1157,
+      "step": 878
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006773558853464265,
+      "loss": 2.1815,
+      "step": 879
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000676660961096555,
+      "loss": 2.1809,
+      "step": 880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000675965646697872,
+      "loss": 2.2477,
+      "step": 881
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006752699436859519,
+      "loss": 2.1019,
+      "step": 882
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006745738535972279,
+      "loss": 2.2319,
+      "step": 883
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006738773779689874,
+      "loss": 2.1673,
+      "step": 884
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006731805183393696,
+      "loss": 2.0892,
+      "step": 885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006724832762473618,
+      "loss": 2.1348,
+      "step": 886
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006717856532327956,
+      "loss": 2.1453,
+      "step": 887
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006710876508363444,
+      "loss": 2.1365,
+      "step": 888
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006703892705995189,
+      "loss": 2.0757,
+      "step": 889
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006696905140646647,
+      "loss": 2.1846,
+      "step": 890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006689913827749581,
+      "loss": 2.2238,
+      "step": 891
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006682918782744032,
+      "loss": 2.0971,
+      "step": 892
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006675920021078282,
+      "loss": 2.1577,
+      "step": 893
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006668917558208823,
+      "loss": 2.137,
+      "step": 894
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006661911409600321,
+      "loss": 2.1842,
+      "step": 895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006654901590725577,
+      "loss": 2.1972,
+      "step": 896
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006647888117065507,
+      "loss": 2.1022,
+      "step": 897
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006640871004109086,
+      "loss": 2.1655,
+      "step": 898
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000663385026735334,
+      "loss": 2.2102,
+      "step": 899
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006626825922303285,
+      "loss": 2.1572,
+      "step": 900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006619797984471915,
+      "loss": 2.1524,
+      "step": 901
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006612766469380158,
+      "loss": 2.0551,
+      "step": 902
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006605731392556833,
+      "loss": 2.1845,
+      "step": 903
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006598692769538637,
+      "loss": 2.1046,
+      "step": 904
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006591650615870091,
+      "loss": 2.2355,
+      "step": 905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006584604947103514,
+      "loss": 2.158,
+      "step": 906
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006577555778798993,
+      "loss": 2.1007,
+      "step": 907
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006570503126524336,
+      "loss": 2.101,
+      "step": 908
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006563447005855054,
+      "loss": 2.1198,
+      "step": 909
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000655638743237431,
+      "loss": 2.0939,
+      "step": 910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006549324421672894,
+      "loss": 2.1428,
+      "step": 911
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006542257989349194,
+      "loss": 2.1232,
+      "step": 912
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006535188151009142,
+      "loss": 2.1709,
+      "step": 913
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006528114922266204,
+      "loss": 2.1915,
+      "step": 914
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006521038318741327,
+      "loss": 2.1407,
+      "step": 915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006513958356062912,
+      "loss": 2.2391,
+      "step": 916
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006506875049866781,
+      "loss": 2.0896,
+      "step": 917
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006499788415796137,
+      "loss": 2.1563,
+      "step": 918
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006492698469501532,
+      "loss": 2.1388,
+      "step": 919
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006485605226640837,
+      "loss": 2.1807,
+      "step": 920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00064785087028792,
+      "loss": 2.1605,
+      "step": 921
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006471408913889019,
+      "loss": 2.1354,
+      "step": 922
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006464305875349892,
+      "loss": 2.2374,
+      "step": 923
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000645719960294861,
+      "loss": 2.1493,
+      "step": 924
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006450090112379092,
+      "loss": 2.2017,
+      "step": 925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006442977419342371,
+      "loss": 2.1752,
+      "step": 926
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000643586153954655,
+      "loss": 2.1979,
+      "step": 927
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006428742488706772,
+      "loss": 2.1804,
+      "step": 928
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006421620282545182,
+      "loss": 2.0971,
+      "step": 929
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006414494936790892,
+      "loss": 2.1317,
+      "step": 930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006407366467179951,
+      "loss": 2.1757,
+      "step": 931
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006400234889455301,
+      "loss": 2.0784,
+      "step": 932
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006393100219366755,
+      "loss": 2.2399,
+      "step": 933
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006385962472670953,
+      "loss": 2.2432,
+      "step": 934
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006378821665131328,
+      "loss": 2.2338,
+      "step": 935
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006371677812518072,
+      "loss": 2.1209,
+      "step": 936
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006364530930608107,
+      "loss": 2.164,
+      "step": 937
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006357381035185038,
+      "loss": 2.1351,
+      "step": 938
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006350228142039131,
+      "loss": 2.1307,
+      "step": 939
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000634307226696727,
+      "loss": 2.1183,
+      "step": 940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006335913425772926,
+      "loss": 2.1919,
+      "step": 941
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006328751634266117,
+      "loss": 2.0845,
+      "step": 942
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006321586908263382,
+      "loss": 2.2172,
+      "step": 943
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006314419263587732,
+      "loss": 2.1846,
+      "step": 944
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006307248716068637,
+      "loss": 2.1286,
+      "step": 945
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006300075281541964,
+      "loss": 2.1943,
+      "step": 946
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006292898975849966,
+      "loss": 2.0849,
+      "step": 947
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000628571981484123,
+      "loss": 2.1402,
+      "step": 948
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006278537814370654,
+      "loss": 2.164,
+      "step": 949
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006271352990299406,
+      "loss": 2.1335,
+      "step": 950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006264165358494885,
+      "loss": 2.0782,
+      "step": 951
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006256974934830694,
+      "loss": 2.1454,
+      "step": 952
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006249781735186606,
+      "loss": 2.1587,
+      "step": 953
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006242585775448518,
+      "loss": 2.1869,
+      "step": 954
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006235387071508427,
+      "loss": 2.1126,
+      "step": 955
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006228185639264384,
+      "loss": 2.0513,
+      "step": 956
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006220981494620475,
+      "loss": 2.1343,
+      "step": 957
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000621377465348677,
+      "loss": 2.2589,
+      "step": 958
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006206565131779293,
+      "loss": 2.1704,
+      "step": 959
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006199352945419994,
+      "loss": 2.1376,
+      "step": 960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00061921381103367,
+      "loss": 2.1321,
+      "step": 961
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006184920642463094,
+      "loss": 2.1077,
+      "step": 962
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006177700557738672,
+      "loss": 2.1577,
+      "step": 963
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006170477872108706,
+      "loss": 2.127,
+      "step": 964
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006163252601524216,
+      "loss": 2.1438,
+      "step": 965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006156024761941925,
+      "loss": 2.1023,
+      "step": 966
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000614879436932424,
+      "loss": 2.0877,
+      "step": 967
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006141561439639196,
+      "loss": 2.1734,
+      "step": 968
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006134325988860433,
+      "loss": 2.1936,
+      "step": 969
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006127088032967165,
+      "loss": 2.1427,
+      "step": 970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006119847587944131,
+      "loss": 2.2092,
+      "step": 971
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006112604669781572,
+      "loss": 2.1716,
+      "step": 972
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006105359294475188,
+      "loss": 2.1081,
+      "step": 973
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006098111478026107,
+      "loss": 2.1787,
+      "step": 974
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006090861236440848,
+      "loss": 2.076,
+      "step": 975
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006083608585731282,
+      "loss": 2.1543,
+      "step": 976
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006076353541914609,
+      "loss": 2.1651,
+      "step": 977
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006069096121013307,
+      "loss": 2.1391,
+      "step": 978
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006061836339055105,
+      "loss": 2.2018,
+      "step": 979
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006054574212072948,
+      "loss": 2.1089,
+      "step": 980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006047309756104958,
+      "loss": 2.1751,
+      "step": 981
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00060400429871944,
+      "loss": 2.0892,
+      "step": 982
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006032773921389654,
+      "loss": 2.1729,
+      "step": 983
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006025502574744162,
+      "loss": 2.1775,
+      "step": 984
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000601822896331641,
+      "loss": 2.1684,
+      "step": 985
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006010953103169883,
+      "loss": 2.0918,
+      "step": 986
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006003675010373034,
+      "loss": 2.2246,
+      "step": 987
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005996394700999246,
+      "loss": 2.2309,
+      "step": 988
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005989112191126794,
+      "loss": 2.1713,
+      "step": 989
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005981827496838822,
+      "loss": 2.154,
+      "step": 990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005974540634223286,
+      "loss": 2.1258,
+      "step": 991
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005967251619372939,
+      "loss": 2.1117,
+      "step": 992
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005959960468385284,
+      "loss": 2.0605,
+      "step": 993
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005952667197362542,
+      "loss": 2.0618,
+      "step": 994
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005945371822411621,
+      "loss": 2.1296,
+      "step": 995
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005938074359644063,
+      "loss": 2.1269,
+      "step": 996
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005930774825176034,
+      "loss": 2.1414,
+      "step": 997
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005923473235128268,
+      "loss": 2.0781,
+      "step": 998
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005916169605626042,
+      "loss": 2.149,
+      "step": 999
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005908863952799134,
+      "loss": 2.175,
+      "step": 1000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005901556292781793,
+      "loss": 2.0108,
+      "step": 1001
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005894246641712698,
+      "loss": 2.1913,
+      "step": 1002
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005886935015734931,
+      "loss": 2.0953,
+      "step": 1003
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005879621430995928,
+      "loss": 2.1433,
+      "step": 1004
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005872305903647455,
+      "loss": 2.1221,
+      "step": 1005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005864988449845569,
+      "loss": 2.0425,
+      "step": 1006
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005857669085750578,
+      "loss": 2.0598,
+      "step": 1007
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005850347827527013,
+      "loss": 2.2102,
+      "step": 1008
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005843024691343584,
+      "loss": 2.1275,
+      "step": 1009
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000583569969337315,
+      "loss": 2.0387,
+      "step": 1010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005828372849792686,
+      "loss": 2.2258,
+      "step": 1011
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005821044176783234,
+      "loss": 2.1185,
+      "step": 1012
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005813713690529886,
+      "loss": 2.2168,
+      "step": 1013
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005806381407221729,
+      "loss": 2.1617,
+      "step": 1014
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005799047343051826,
+      "loss": 2.1203,
+      "step": 1015
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005791711514217171,
+      "loss": 2.138,
+      "step": 1016
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005784373936918654,
+      "loss": 2.1849,
+      "step": 1017
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005777034627361025,
+      "loss": 2.0865,
+      "step": 1018
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005769693601752864,
+      "loss": 2.095,
+      "step": 1019
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005762350876306537,
+      "loss": 2.1213,
+      "step": 1020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005755006467238168,
+      "loss": 2.1252,
+      "step": 1021
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005747660390767593,
+      "loss": 2.1646,
+      "step": 1022
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005740312663118338,
+      "loss": 2.0757,
+      "step": 1023
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005732963300517568,
+      "loss": 2.1166,
+      "step": 1024
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005725612319196064,
+      "loss": 2.2827,
+      "step": 1025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000571825973538818,
+      "loss": 2.1869,
+      "step": 1026
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005710905565331811,
+      "loss": 2.1472,
+      "step": 1027
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005703549825268353,
+      "loss": 2.1481,
+      "step": 1028
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005696192531442667,
+      "loss": 2.1244,
+      "step": 1029
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000568883370010305,
+      "loss": 2.1169,
+      "step": 1030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005681473347501192,
+      "loss": 2.0681,
+      "step": 1031
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005674111489892144,
+      "loss": 2.1761,
+      "step": 1032
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005666748143534282,
+      "loss": 2.1082,
+      "step": 1033
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005659383324689266,
+      "loss": 2.2086,
+      "step": 1034
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005652017049622007,
+      "loss": 2.1915,
+      "step": 1035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005644649334600641,
+      "loss": 2.1026,
+      "step": 1036
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005637280195896474,
+      "loss": 2.1691,
+      "step": 1037
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005629909649783961,
+      "loss": 2.0506,
+      "step": 1038
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005622537712540664,
+      "loss": 2.1171,
+      "step": 1039
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005615164400447218,
+      "loss": 2.0875,
+      "step": 1040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005607789729787294,
+      "loss": 2.1403,
+      "step": 1041
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005600413716847564,
+      "loss": 2.1855,
+      "step": 1042
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000559303637791766,
+      "loss": 2.0862,
+      "step": 1043
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005585657729290151,
+      "loss": 2.1056,
+      "step": 1044
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000557827778726049,
+      "loss": 2.1922,
+      "step": 1045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005570896568126993,
+      "loss": 2.2178,
+      "step": 1046
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005563514088190788,
+      "loss": 2.1577,
+      "step": 1047
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005556130363755798,
+      "loss": 2.1506,
+      "step": 1048
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005548745411128688,
+      "loss": 2.1737,
+      "step": 1049
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005541359246618835,
+      "loss": 2.1132,
+      "step": 1050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005533971886538293,
+      "loss": 2.1867,
+      "step": 1051
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000552658334720176,
+      "loss": 2.1978,
+      "step": 1052
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005519193644926535,
+      "loss": 2.1267,
+      "step": 1053
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005511802796032485,
+      "loss": 2.0716,
+      "step": 1054
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005504410816842009,
+      "loss": 2.0925,
+      "step": 1055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005497017723680009,
+      "loss": 2.0973,
+      "step": 1056
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005489623532873836,
+      "loss": 2.1986,
+      "step": 1057
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005482228260753273,
+      "loss": 2.1715,
+      "step": 1058
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005474831923650488,
+      "loss": 2.1902,
+      "step": 1059
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00054674345379,
+      "loss": 2.0647,
+      "step": 1060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000546003611983865,
+      "loss": 2.1412,
+      "step": 1061
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005452636685805552,
+      "loss": 2.1856,
+      "step": 1062
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005445236252142066,
+      "loss": 2.1309,
+      "step": 1063
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000543783483519176,
+      "loss": 2.2178,
+      "step": 1064
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005430432451300374,
+      "loss": 2.1237,
+      "step": 1065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005423029116815781,
+      "loss": 2.0841,
+      "step": 1066
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005415624848087959,
+      "loss": 2.1046,
+      "step": 1067
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000540821966146894,
+      "loss": 2.1122,
+      "step": 1068
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005400813573312793,
+      "loss": 2.0515,
+      "step": 1069
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005393406599975572,
+      "loss": 2.1979,
+      "step": 1070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005385998757815287,
+      "loss": 2.162,
+      "step": 1071
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005378590063191867,
+      "loss": 2.0431,
+      "step": 1072
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005371180532467124,
+      "loss": 2.0687,
+      "step": 1073
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000536377018200472,
+      "loss": 2.1668,
+      "step": 1074
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005356359028170118,
+      "loss": 2.1588,
+      "step": 1075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005348947087330564,
+      "loss": 2.0893,
+      "step": 1076
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005341534375855037,
+      "loss": 2.0704,
+      "step": 1077
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005334120910114222,
+      "loss": 2.1622,
+      "step": 1078
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005326706706480467,
+      "loss": 2.0895,
+      "step": 1079
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005319291781327749,
+      "loss": 2.2136,
+      "step": 1080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005311876151031642,
+      "loss": 2.0742,
+      "step": 1081
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005304459831969274,
+      "loss": 2.1448,
+      "step": 1082
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005297042840519294,
+      "loss": 2.1602,
+      "step": 1083
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005289625193061838,
+      "loss": 2.1279,
+      "step": 1084
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005282206905978489,
+      "loss": 2.1721,
+      "step": 1085
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005274787995652246,
+      "loss": 2.0553,
+      "step": 1086
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000526736847846748,
+      "loss": 2.2213,
+      "step": 1087
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005259948370809901,
+      "loss": 2.1576,
+      "step": 1088
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005252527689066533,
+      "loss": 2.1929,
+      "step": 1089
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005245106449625654,
+      "loss": 2.2073,
+      "step": 1090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005237684668876785,
+      "loss": 2.1202,
+      "step": 1091
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005230262363210637,
+      "loss": 2.1721,
+      "step": 1092
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005222839549019079,
+      "loss": 2.1486,
+      "step": 1093
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005215416242695108,
+      "loss": 2.0978,
+      "step": 1094
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005207992460632804,
+      "loss": 2.1933,
+      "step": 1095
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005200568219227299,
+      "loss": 2.1353,
+      "step": 1096
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000519314353487474,
+      "loss": 2.1349,
+      "step": 1097
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005185718423972251,
+      "loss": 2.1692,
+      "step": 1098
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005178292902917898,
+      "loss": 2.1253,
+      "step": 1099
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005170866988110656,
+      "loss": 2.1459,
+      "step": 1100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005163440695950362,
+      "loss": 2.14,
+      "step": 1101
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005156014042837695,
+      "loss": 2.0526,
+      "step": 1102
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005148587045174128,
+      "loss": 2.1498,
+      "step": 1103
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005141159719361891,
+      "loss": 2.0228,
+      "step": 1104
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005133732081803945,
+      "loss": 2.1813,
+      "step": 1105
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005126304148903936,
+      "loss": 2.1102,
+      "step": 1106
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005118875937066161,
+      "loss": 2.1153,
+      "step": 1107
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005111447462695537,
+      "loss": 2.094,
+      "step": 1108
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005104018742197557,
+      "loss": 2.1359,
+      "step": 1109
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005096589791978261,
+      "loss": 2.1662,
+      "step": 1110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005089160628444192,
+      "loss": 2.1411,
+      "step": 1111
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005081731268002371,
+      "loss": 2.2142,
+      "step": 1112
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005074301727060243,
+      "loss": 2.122,
+      "step": 1113
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005066872022025663,
+      "loss": 2.1072,
+      "step": 1114
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005059442169306844,
+      "loss": 2.0964,
+      "step": 1115
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005052012185312321,
+      "loss": 2.1317,
+      "step": 1116
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005044582086450925,
+      "loss": 2.206,
+      "step": 1117
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005037151889131737,
+      "loss": 2.1418,
+      "step": 1118
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005029721609764059,
+      "loss": 2.2003,
+      "step": 1119
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000502229126475737,
+      "loss": 2.05,
+      "step": 1120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005014860870521293,
+      "loss": 2.0631,
+      "step": 1121
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005007430443465569,
+      "loss": 2.1435,
+      "step": 1122
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0005,
+      "loss": 2.1074,
+      "step": 1123
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004992569556534432,
+      "loss": 2.0538,
+      "step": 1124
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004985139129478707,
+      "loss": 2.0941,
+      "step": 1125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004977708735242633,
+      "loss": 2.1217,
+      "step": 1126
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004970278390235942,
+      "loss": 2.1927,
+      "step": 1127
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004962848110868262,
+      "loss": 2.2794,
+      "step": 1128
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004955417913549074,
+      "loss": 2.1733,
+      "step": 1129
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004947987814687679,
+      "loss": 2.0723,
+      "step": 1130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004940557830693157,
+      "loss": 2.0306,
+      "step": 1131
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004933127977974338,
+      "loss": 2.1107,
+      "step": 1132
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004925698272939757,
+      "loss": 2.0902,
+      "step": 1133
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004918268731997632,
+      "loss": 2.2137,
+      "step": 1134
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004910839371555809,
+      "loss": 2.2111,
+      "step": 1135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004903410208021739,
+      "loss": 2.1036,
+      "step": 1136
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004895981257802443,
+      "loss": 2.1782,
+      "step": 1137
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004888552537304463,
+      "loss": 2.0758,
+      "step": 1138
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048811240629338394,
+      "loss": 2.1614,
+      "step": 1139
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048736958510960663,
+      "loss": 2.063,
+      "step": 1140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048662679181960564,
+      "loss": 2.2176,
+      "step": 1141
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048588402806381094,
+      "loss": 2.1632,
+      "step": 1142
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004851412954825874,
+      "loss": 2.1689,
+      "step": 1143
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048439859571623034,
+      "loss": 2.0992,
+      "step": 1144
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048365593040496373,
+      "loss": 2.1009,
+      "step": 1145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048291330118893443,
+      "loss": 2.1358,
+      "step": 1146
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004821707097082102,
+      "loss": 2.0722,
+      "step": 1147
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004814281576027749,
+      "loss": 2.2294,
+      "step": 1148
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000480685646512526,
+      "loss": 2.1185,
+      "step": 1149
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047994317807727025,
+      "loss": 2.0731,
+      "step": 1150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047920075393671974,
+      "loss": 2.1485,
+      "step": 1151
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004784583757304893,
+      "loss": 2.1214,
+      "step": 1152
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047771604509809214,
+      "loss": 2.1246,
+      "step": 1153
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004769737636789364,
+      "loss": 2.1336,
+      "step": 1154
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047623153311232157,
+      "loss": 2.1942,
+      "step": 1155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004754893550374346,
+      "loss": 2.0887,
+      "step": 1156
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047474723109334685,
+      "loss": 2.2145,
+      "step": 1157
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047400516291900993,
+      "loss": 2.136,
+      "step": 1158
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004732631521532522,
+      "loss": 2.1195,
+      "step": 1159
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004725212004347755,
+      "loss": 2.177,
+      "step": 1160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047177930940215095,
+      "loss": 2.0452,
+      "step": 1161
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047103748069381624,
+      "loss": 2.1019,
+      "step": 1162
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004702957159480707,
+      "loss": 2.0436,
+      "step": 1163
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046955401680307267,
+      "loss": 2.1349,
+      "step": 1164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004688123848968359,
+      "loss": 2.087,
+      "step": 1165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046807082186722516,
+      "loss": 2.1798,
+      "step": 1166
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004673293293519535,
+      "loss": 2.1727,
+      "step": 1167
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046658790898857806,
+      "loss": 2.0816,
+      "step": 1168
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004658465624144963,
+      "loss": 2.1477,
+      "step": 1169
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004651052912669438,
+      "loss": 2.1268,
+      "step": 1170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004643640971829883,
+      "loss": 2.034,
+      "step": 1171
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004636229817995281,
+      "loss": 2.2082,
+      "step": 1172
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004628819467532876,
+      "loss": 2.0916,
+      "step": 1173
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046214099368081335,
+      "loss": 2.037,
+      "step": 1174
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004614001242184714,
+      "loss": 2.0885,
+      "step": 1175
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000460659340002443,
+      "loss": 2.1551,
+      "step": 1176
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045991864266872073,
+      "loss": 2.0791,
+      "step": 1177
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045917803385310595,
+      "loss": 2.097,
+      "step": 1178
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045843751519120417,
+      "loss": 2.0478,
+      "step": 1179
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045769708831842193,
+      "loss": 2.0477,
+      "step": 1180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045695675486996266,
+      "loss": 2.1986,
+      "step": 1181
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045621651648082405,
+      "loss": 2.2187,
+      "step": 1182
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045547637478579356,
+      "loss": 2.2056,
+      "step": 1183
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004547363314194449,
+      "loss": 2.1701,
+      "step": 1184
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000453996388016135,
+      "loss": 2.114,
+      "step": 1185
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004532565462099999,
+      "loss": 2.1639,
+      "step": 1186
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004525168076349513,
+      "loss": 2.2072,
+      "step": 1187
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004517771739246729,
+      "loss": 2.1565,
+      "step": 1188
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004510376467126165,
+      "loss": 2.0711,
+      "step": 1189
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004502982276319992,
+      "loss": 2.1523,
+      "step": 1190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004495589183157991,
+      "loss": 2.1532,
+      "step": 1191
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004488197203967517,
+      "loss": 2.121,
+      "step": 1192
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004480806355073467,
+      "loss": 2.1097,
+      "step": 1193
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000447341665279824,
+      "loss": 2.0803,
+      "step": 1194
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004466028113461708,
+      "loss": 2.1381,
+      "step": 1195
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004458640753381167,
+      "loss": 2.0545,
+      "step": 1196
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004451254588871313,
+      "loss": 2.204,
+      "step": 1197
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004443869636244203,
+      "loss": 2.117,
+      "step": 1198
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004436485911809212,
+      "loss": 2.0623,
+      "step": 1199
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044291034318730087,
+      "loss": 2.159,
+      "step": 1200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004421722212739511,
+      "loss": 2.1999,
+      "step": 1201
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004414342270709848,
+      "loss": 2.1042,
+      "step": 1202
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044069636220823397,
+      "loss": 2.0543,
+      "step": 1203
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004399586283152437,
+      "loss": 2.1158,
+      "step": 1204
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004392210270212706,
+      "loss": 2.0954,
+      "step": 1205
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043848355995527825,
+      "loss": 2.1123,
+      "step": 1206
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043774622874593374,
+      "loss": 2.1652,
+      "step": 1207
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000437009035021604,
+      "loss": 2.1222,
+      "step": 1208
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043627198041035274,
+      "loss": 2.1503,
+      "step": 1209
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043553506653993597,
+      "loss": 2.1104,
+      "step": 1210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004347982950377992,
+      "loss": 2.1109,
+      "step": 1211
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004340616675310735,
+      "loss": 2.157,
+      "step": 1212
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043332518564657193,
+      "loss": 2.1182,
+      "step": 1213
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004325888510107856,
+      "loss": 2.1004,
+      "step": 1214
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004318526652498809,
+      "loss": 2.0092,
+      "step": 1215
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043111662998969523,
+      "loss": 2.0556,
+      "step": 1216
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004303807468557335,
+      "loss": 2.1676,
+      "step": 1217
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004296450174731648,
+      "loss": 2.0775,
+      "step": 1218
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004289094434668188,
+      "loss": 2.0751,
+      "step": 1219
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042817402646118185,
+      "loss": 2.0663,
+      "step": 1220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004274387680803936,
+      "loss": 2.1038,
+      "step": 1221
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042670366994824327,
+      "loss": 2.1287,
+      "step": 1222
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004259687336881663,
+      "loss": 2.1158,
+      "step": 1223
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004252339609232408,
+      "loss": 2.1031,
+      "step": 1224
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004244993532761834,
+      "loss": 2.095,
+      "step": 1225
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042376491236934634,
+      "loss": 2.123,
+      "step": 1226
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004230306398247136,
+      "loss": 2.1351,
+      "step": 1227
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004222965372638976,
+      "loss": 2.1104,
+      "step": 1228
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004215626063081348,
+      "loss": 2.1972,
+      "step": 1229
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000420828848578283,
+      "loss": 2.0315,
+      "step": 1230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004200952656948175,
+      "loss": 2.1504,
+      "step": 1231
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004193618592778272,
+      "loss": 2.2187,
+      "step": 1232
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004186286309470116,
+      "loss": 2.1269,
+      "step": 1233
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004178955823216767,
+      "loss": 2.0759,
+      "step": 1234
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041716271502073137,
+      "loss": 2.0553,
+      "step": 1235
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000416430030662685,
+      "loss": 2.06,
+      "step": 1236
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041569753086564173,
+      "loss": 2.119,
+      "step": 1237
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004149652172472988,
+      "loss": 2.0774,
+      "step": 1238
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041423309142494234,
+      "loss": 2.2325,
+      "step": 1239
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004135011550154433,
+      "loss": 2.096,
+      "step": 1240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004127694096352546,
+      "loss": 2.1162,
+      "step": 1241
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041203785690040743,
+      "loss": 2.1107,
+      "step": 1242
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041130649842650694,
+      "loss": 2.1296,
+      "step": 1243
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041057533582873016,
+      "loss": 2.1064,
+      "step": 1244
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004098443707218208,
+      "loss": 2.1726,
+      "step": 1245
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040911360472008673,
+      "loss": 2.0792,
+      "step": 1246
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004083830394373959,
+      "loss": 2.0509,
+      "step": 1247
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040765267648717324,
+      "loss": 2.1259,
+      "step": 1248
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040692251748239677,
+      "loss": 2.1227,
+      "step": 1249
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040619256403559383,
+      "loss": 2.0888,
+      "step": 1250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000405462817758838,
+      "loss": 2.22,
+      "step": 1251
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004047332802637457,
+      "loss": 2.1303,
+      "step": 1252
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040400395316147157,
+      "loss": 2.086,
+      "step": 1253
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040327483806270627,
+      "loss": 2.1416,
+      "step": 1254
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004025459365776715,
+      "loss": 2.1551,
+      "step": 1255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040181725031611794,
+      "loss": 2.1729,
+      "step": 1256
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004010887808873206,
+      "loss": 2.0057,
+      "step": 1257
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040036052990007553,
+      "loss": 2.114,
+      "step": 1258
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003996324989626967,
+      "loss": 2.1757,
+      "step": 1259
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039890468968301166,
+      "loss": 2.1085,
+      "step": 1260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003981771036683591,
+      "loss": 2.0503,
+      "step": 1261
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039744974252558385,
+      "loss": 2.1602,
+      "step": 1262
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039672260786103463,
+      "loss": 2.0358,
+      "step": 1263
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039599570128055994,
+      "loss": 2.0299,
+      "step": 1264
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003952690243895044,
+      "loss": 2.1852,
+      "step": 1265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003945425787927054,
+      "loss": 2.1038,
+      "step": 1266
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039381636609448975,
+      "loss": 2.0354,
+      "step": 1267
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003930903878986693,
+      "loss": 2.0766,
+      "step": 1268
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039236464580853916,
+      "loss": 2.1344,
+      "step": 1269
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003916391414268718,
+      "loss": 2.1417,
+      "step": 1270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039091387635591536,
+      "loss": 2.163,
+      "step": 1271
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003901888521973894,
+      "loss": 2.0588,
+      "step": 1272
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003894640705524813,
+      "loss": 2.1371,
+      "step": 1273
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038873953302184284,
+      "loss": 2.023,
+      "step": 1274
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000388015241205587,
+      "loss": 2.0715,
+      "step": 1275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038729119670328355,
+      "loss": 2.0945,
+      "step": 1276
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038656740111395665,
+      "loss": 2.1093,
+      "step": 1277
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038584385603608053,
+      "loss": 2.0993,
+      "step": 1278
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038512056306757615,
+      "loss": 2.0952,
+      "step": 1279
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003843975238058075,
+      "loss": 2.071,
+      "step": 1280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038367473984757863,
+      "loss": 2.0965,
+      "step": 1281
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003829522127891296,
+      "loss": 2.1344,
+      "step": 1282
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003822299442261329,
+      "loss": 2.1166,
+      "step": 1283
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038150793575369063,
+      "loss": 2.0902,
+      "step": 1284
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003807861889663299,
+      "loss": 2.1238,
+      "step": 1285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003800647054580006,
+      "loss": 2.1063,
+      "step": 1286
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037934348682207064,
+      "loss": 2.1584,
+      "step": 1287
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037862253465132306,
+      "loss": 2.1388,
+      "step": 1288
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037790185053795245,
+      "loss": 2.1649,
+      "step": 1289
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003771814360735616,
+      "loss": 2.1184,
+      "step": 1290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037646129284915755,
+      "loss": 2.0568,
+      "step": 1291
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037574142245514825,
+      "loss": 2.1776,
+      "step": 1292
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003750218264813393,
+      "loss": 2.1151,
+      "step": 1293
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003743025065169305,
+      "loss": 2.1822,
+      "step": 1294
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003735834641505116,
+      "loss": 2.1257,
+      "step": 1295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037286470097005954,
+      "loss": 2.14,
+      "step": 1296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003721462185629347,
+      "loss": 2.0964,
+      "step": 1297
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037142801851587707,
+      "loss": 2.0856,
+      "step": 1298
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037071010241500357,
+      "loss": 2.1237,
+      "step": 1299
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036999247184580383,
+      "loss": 2.0365,
+      "step": 1300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036927512839313636,
+      "loss": 2.0916,
+      "step": 1301
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003685580736412268,
+      "loss": 2.1268,
+      "step": 1302
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000367841309173662,
+      "loss": 2.0698,
+      "step": 1303
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003671248365733883,
+      "loss": 2.0391,
+      "step": 1304
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003664086574227075,
+      "loss": 2.0863,
+      "step": 1305
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000365692773303273,
+      "loss": 2.0877,
+      "step": 1306
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036497718579608696,
+      "loss": 2.0832,
+      "step": 1307
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003642618964814964,
+      "loss": 2.0369,
+      "step": 1308
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036354690693918946,
+      "loss": 2.0638,
+      "step": 1309
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036283221874819284,
+      "loss": 2.1562,
+      "step": 1310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003621178334868672,
+      "loss": 2.0643,
+      "step": 1311
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036140375273290476,
+      "loss": 2.1205,
+      "step": 1312
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003606899780633245,
+      "loss": 2.1237,
+      "step": 1313
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003599765110544699,
+      "loss": 2.1207,
+      "step": 1314
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003592633532820052,
+      "loss": 2.0805,
+      "step": 1315
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003585505063209109,
+      "loss": 2.103,
+      "step": 1316
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035783797174548194,
+      "loss": 2.1047,
+      "step": 1317
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035712575112932277,
+      "loss": 2.138,
+      "step": 1318
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000356413846045345,
+      "loss": 2.0703,
+      "step": 1319
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000355702258065763,
+      "loss": 2.0604,
+      "step": 1320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003549909887620909,
+      "loss": 2.1101,
+      "step": 1321
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035428003970513914,
+      "loss": 2.0626,
+      "step": 1322
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035356941246501085,
+      "loss": 2.0407,
+      "step": 1323
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003528591086110984,
+      "loss": 2.1622,
+      "step": 1324
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035214912971208,
+      "loss": 2.1581,
+      "step": 1325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003514394773359163,
+      "loss": 2.1293,
+      "step": 1326
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003507301530498469,
+      "loss": 2.1365,
+      "step": 1327
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035002115842038646,
+      "loss": 2.136,
+      "step": 1328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034931249501332195,
+      "loss": 2.1973,
+      "step": 1329
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034860416439370885,
+      "loss": 2.2089,
+      "step": 1330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003478961681258674,
+      "loss": 2.1731,
+      "step": 1331
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003471885077733796,
+      "loss": 2.0813,
+      "step": 1332
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003464811848990859,
+      "loss": 2.0945,
+      "step": 1333
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034577420106508063,
+      "loss": 2.0959,
+      "step": 1334
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003450675578327105,
+      "loss": 2.1856,
+      "step": 1335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000344361256762569,
+      "loss": 2.1518,
+      "step": 1336
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034365529941449456,
+      "loss": 2.1893,
+      "step": 1337
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003429496873475664,
+      "loss": 2.1176,
+      "step": 1338
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003422444221201009,
+      "loss": 2.1517,
+      "step": 1339
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003415395052896487,
+      "loss": 2.1397,
+      "step": 1340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003408349384129912,
+      "loss": 2.1447,
+      "step": 1341
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00034013072304613643,
+      "loss": 2.2105,
+      "step": 1342
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033942686074431674,
+      "loss": 2.0883,
+      "step": 1343
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003387233530619843,
+      "loss": 2.159,
+      "step": 1344
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003380202015528084,
+      "loss": 2.0264,
+      "step": 1345
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003373174077696715,
+      "loss": 2.1417,
+      "step": 1346
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003366149732646661,
+      "loss": 2.1319,
+      "step": 1347
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033591289958909143,
+      "loss": 2.1539,
+      "step": 1348
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033521118829344954,
+      "loss": 2.1058,
+      "step": 1349
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003345098409274423,
+      "loss": 2.1619,
+      "step": 1350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033380885903996796,
+      "loss": 2.055,
+      "step": 1351
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033310824417911766,
+      "loss": 2.0894,
+      "step": 1352
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033240799789217184,
+      "loss": 2.1755,
+      "step": 1353
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033170812172559694,
+      "loss": 2.0987,
+      "step": 1354
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000331008617225042,
+      "loss": 2.0718,
+      "step": 1355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003303094859353355,
+      "loss": 2.0468,
+      "step": 1356
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003296107294004812,
+      "loss": 2.1009,
+      "step": 1357
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003289123491636559,
+      "loss": 1.9508,
+      "step": 1358
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032821434676720443,
+      "loss": 2.0405,
+      "step": 1359
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032751672375263836,
+      "loss": 2.09,
+      "step": 1360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003268194816606305,
+      "loss": 2.0449,
+      "step": 1361
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032612262203101267,
+      "loss": 2.1052,
+      "step": 1362
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032542614640277225,
+      "loss": 2.0609,
+      "step": 1363
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003247300563140481,
+      "loss": 2.0796,
+      "step": 1364
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032403435330212807,
+      "loss": 2.1442,
+      "step": 1365
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032333903890344515,
+      "loss": 2.1329,
+      "step": 1366
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032264411465357333,
+      "loss": 2.0433,
+      "step": 1367
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032194958208722654,
+      "loss": 2.0735,
+      "step": 1368
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032125544273825204,
+      "loss": 2.1998,
+      "step": 1369
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003205616981396297,
+      "loss": 2.1211,
+      "step": 1370
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031986834982346713,
+      "loss": 2.1942,
+      "step": 1371
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031917539932099694,
+      "loss": 2.1274,
+      "step": 1372
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031848284816257336,
+      "loss": 2.0823,
+      "step": 1373
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003177906978776682,
+      "loss": 2.0882,
+      "step": 1374
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003170989499948683,
+      "loss": 2.1303,
+      "step": 1375
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003164076060418719,
+      "loss": 2.1321,
+      "step": 1376
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000315716667545485,
+      "loss": 2.1758,
+      "step": 1377
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031502613603161836,
+      "loss": 2.1275,
+      "step": 1378
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031433601302528335,
+      "loss": 2.1393,
+      "step": 1379
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031364630005058995,
+      "loss": 2.0238,
+      "step": 1380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003129569986307422,
+      "loss": 2.1233,
+      "step": 1381
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031226811028803515,
+      "loss": 2.0761,
+      "step": 1382
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031157963654385173,
+      "loss": 2.039,
+      "step": 1383
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003108915789186592,
+      "loss": 2.1648,
+      "step": 1384
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00031020393893200604,
+      "loss": 2.0695,
+      "step": 1385
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030951671810251823,
+      "loss": 2.1882,
+      "step": 1386
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003088299179478959,
+      "loss": 2.0837,
+      "step": 1387
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003081435399849104,
+      "loss": 2.092,
+      "step": 1388
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003074575857294004,
+      "loss": 2.1608,
+      "step": 1389
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003067720566962691,
+      "loss": 2.1347,
+      "step": 1390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003060869543994806,
+      "loss": 2.0452,
+      "step": 1391
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003054022803520562,
+      "loss": 2.1247,
+      "step": 1392
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003047180360660721,
+      "loss": 2.1047,
+      "step": 1393
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030403422305265475,
+      "loss": 2.13,
+      "step": 1394
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003033508428219785,
+      "loss": 2.1118,
+      "step": 1395
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030266789688326184,
+      "loss": 2.0262,
+      "step": 1396
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030198538674476393,
+      "loss": 2.107,
+      "step": 1397
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030130331391378185,
+      "loss": 2.0822,
+      "step": 1398
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003006216798966468,
+      "loss": 2.1123,
+      "step": 1399
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029994048619872034,
+      "loss": 2.1241,
+      "step": 1400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002992597343243927,
+      "loss": 2.1842,
+      "step": 1401
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002985794257770773,
+      "loss": 2.1116,
+      "step": 1402
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002978995620592092,
+      "loss": 2.1496,
+      "step": 1403
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002972201446722405,
+      "loss": 2.1195,
+      "step": 1404
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029654117511663803,
+      "loss": 2.0942,
+      "step": 1405
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002958626548918795,
+      "loss": 2.1217,
+      "step": 1406
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029518458549645014,
+      "loss": 2.0801,
+      "step": 1407
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029450696842783954,
+      "loss": 2.0691,
+      "step": 1408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029382980518253865,
+      "loss": 2.1329,
+      "step": 1409
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029315309725603595,
+      "loss": 2.1385,
+      "step": 1410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029247684614281446,
+      "loss": 2.0821,
+      "step": 1411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002918010533363481,
+      "loss": 2.0677,
+      "step": 1412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002911257203290987,
+      "loss": 2.0731,
+      "step": 1413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029045084861251314,
+      "loss": 2.1008,
+      "step": 1414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028977643967701897,
+      "loss": 2.0557,
+      "step": 1415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028910249501202156,
+      "loss": 2.1382,
+      "step": 1416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028842901610590165,
+      "loss": 2.0943,
+      "step": 1417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028775600444601123,
+      "loss": 2.1597,
+      "step": 1418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028708346151866973,
+      "loss": 2.0327,
+      "step": 1419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002864113888091622,
+      "loss": 2.0607,
+      "step": 1420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002857397878017348,
+      "loss": 2.1301,
+      "step": 1421
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028506865997959173,
+      "loss": 2.0956,
+      "step": 1422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000284398006824893,
+      "loss": 2.0533,
+      "step": 1423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028372782981874963,
+      "loss": 2.1338,
+      "step": 1424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028305813044122096,
+      "loss": 2.1182,
+      "step": 1425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002823889101713122,
+      "loss": 2.1402,
+      "step": 1426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002817201704869701,
+      "loss": 2.1288,
+      "step": 1427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028105191286508,
+      "loss": 2.0791,
+      "step": 1428
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028038413878146245,
+      "loss": 2.0857,
+      "step": 1429
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027971684971087073,
+      "loss": 2.065,
+      "step": 1430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027905004712698643,
+      "loss": 2.0775,
+      "step": 1431
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002783837325024167,
+      "loss": 2.1449,
+      "step": 1432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027771790730869153,
+      "loss": 2.0498,
+      "step": 1433
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002770525730162599,
+      "loss": 2.084,
+      "step": 1434
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027638773109448645,
+      "loss": 2.0938,
+      "step": 1435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027572338301164824,
+      "loss": 2.1324,
+      "step": 1436
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002750595302349324,
+      "loss": 2.1674,
+      "step": 1437
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027439617423043145,
+      "loss": 2.0299,
+      "step": 1438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027373331646314114,
+      "loss": 2.1251,
+      "step": 1439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002730709583969572,
+      "loss": 2.0766,
+      "step": 1440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002724091014946711,
+      "loss": 2.0268,
+      "step": 1441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027174774721796824,
+      "loss": 2.1217,
+      "step": 1442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002710868970274232,
+      "loss": 2.0914,
+      "step": 1443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002704265523824982,
+      "loss": 2.0773,
+      "step": 1444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026976671474153826,
+      "loss": 2.0922,
+      "step": 1445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026910738556176886,
+      "loss": 2.1479,
+      "step": 1446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002684485662992929,
+      "loss": 2.0664,
+      "step": 1447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002677902584090869,
+      "loss": 2.0077,
+      "step": 1448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026713246334499774,
+      "loss": 2.139,
+      "step": 1449
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026647518255974023,
+      "loss": 2.0502,
+      "step": 1450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002658184175048934,
+      "loss": 2.0723,
+      "step": 1451
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026516216963089694,
+      "loss": 2.0666,
+      "step": 1452
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002645064403870488,
+      "loss": 2.1307,
+      "step": 1453
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002638512312215011,
+      "loss": 2.0797,
+      "step": 1454
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002631965435812575,
+      "loss": 2.0657,
+      "step": 1455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026254237891217046,
+      "loss": 2.1901,
+      "step": 1456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002618887386589367,
+      "loss": 2.0508,
+      "step": 1457
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002612356242650949,
+      "loss": 2.1007,
+      "step": 1458
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002605830371730229,
+      "loss": 2.1437,
+      "step": 1459
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002599309788239339,
+      "loss": 2.1502,
+      "step": 1460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025927945065787306,
+      "loss": 2.0772,
+      "step": 1461
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002586284541137145,
+      "loss": 2.0797,
+      "step": 1462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025797799062915905,
+      "loss": 2.0639,
+      "step": 1463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025732806164072966,
+      "loss": 2.1382,
+      "step": 1464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025667866858376874,
+      "loss": 2.0548,
+      "step": 1465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002560298128924358,
+      "loss": 2.0447,
+      "step": 1466
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002553814959997032,
+      "loss": 2.1882,
+      "step": 1467
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025473371933735334,
+      "loss": 2.0815,
+      "step": 1468
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025408648433597534,
+      "loss": 2.1107,
+      "step": 1469
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025343979242496283,
+      "loss": 2.0738,
+      "step": 1470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025279364503250925,
+      "loss": 2.1138,
+      "step": 1471
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002521480435856056,
+      "loss": 2.1081,
+      "step": 1472
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002515029895100378,
+      "loss": 2.0608,
+      "step": 1473
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002508584842303822,
+      "loss": 2.1625,
+      "step": 1474
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002502145291700038,
+      "loss": 2.0867,
+      "step": 1475
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002495711257510517,
+      "loss": 2.1061,
+      "step": 1476
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002489282753944575,
+      "loss": 1.9812,
+      "step": 1477
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024828597951993093,
+      "loss": 2.0547,
+      "step": 1478
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024764423954595706,
+      "loss": 2.0928,
+      "step": 1479
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002470030568897938,
+      "loss": 2.1233,
+      "step": 1480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024636243296746773,
+      "loss": 2.1053,
+      "step": 1481
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002457223691937716,
+      "loss": 2.1155,
+      "step": 1482
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002450828669822613,
+      "loss": 2.1187,
+      "step": 1483
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024444392774525253,
+      "loss": 2.1132,
+      "step": 1484
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024380555289381733,
+      "loss": 2.0925,
+      "step": 1485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024316774383778184,
+      "loss": 2.0866,
+      "step": 1486
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002425305019857222,
+      "loss": 2.0761,
+      "step": 1487
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024189382874496184,
+      "loss": 2.0639,
+      "step": 1488
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024125772552156916,
+      "loss": 2.1304,
+      "step": 1489
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024062219372035292,
+      "loss": 2.1748,
+      "step": 1490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023998723474486007,
+      "loss": 2.1036,
+      "step": 1491
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023935284999737272,
+      "loss": 2.0976,
+      "step": 1492
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023871904087890505,
+      "loss": 2.1609,
+      "step": 1493
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023808580878919945,
+      "loss": 2.0248,
+      "step": 1494
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023745315512672398,
+      "loss": 2.1415,
+      "step": 1495
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002368210812886698,
+      "loss": 2.0564,
+      "step": 1496
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002361895886709471,
+      "loss": 2.1143,
+      "step": 1497
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002355586786681823,
+      "loss": 2.014,
+      "step": 1498
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023492835267371575,
+      "loss": 2.1419,
+      "step": 1499
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002342986120795978,
+      "loss": 2.0445,
+      "step": 1500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002336694582765857,
+      "loss": 2.0349,
+      "step": 1501
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023304089265414085,
+      "loss": 2.0764,
+      "step": 1502
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023241291660042613,
+      "loss": 2.0617,
+      "step": 1503
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023178553150230186,
+      "loss": 2.1367,
+      "step": 1504
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023115873874532324,
+      "loss": 2.0688,
+      "step": 1505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023053253971373796,
+      "loss": 2.139,
+      "step": 1506
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022990693579048166,
+      "loss": 2.0803,
+      "step": 1507
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022928192835717644,
+      "loss": 2.0851,
+      "step": 1508
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022865751879412634,
+      "loss": 2.1487,
+      "step": 1509
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022803370848031585,
+      "loss": 2.1288,
+      "step": 1510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022741049879340542,
+      "loss": 2.0115,
+      "step": 1511
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022678789110972897,
+      "loss": 2.0537,
+      "step": 1512
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022616588680429155,
+      "loss": 2.0566,
+      "step": 1513
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022554448725076526,
+      "loss": 2.1598,
+      "step": 1514
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002249236938214863,
+      "loss": 2.0485,
+      "step": 1515
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022430350788745296,
+      "loss": 1.9762,
+      "step": 1516
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022368393081832166,
+      "loss": 2.1726,
+      "step": 1517
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022306496398240383,
+      "loss": 2.0672,
+      "step": 1518
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022244660874666373,
+      "loss": 2.0545,
+      "step": 1519
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022182886647671452,
+      "loss": 2.0943,
+      "step": 1520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002212117385368157,
+      "loss": 2.0395,
+      "step": 1521
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022059522628987038,
+      "loss": 2.187,
+      "step": 1522
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021997933109742162,
+      "loss": 2.1213,
+      "step": 1523
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021936405431964969,
+      "loss": 2.0344,
+      "step": 1524
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021874939731536926,
+      "loss": 2.1021,
+      "step": 1525
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021813536144202656,
+      "loss": 2.0673,
+      "step": 1526
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021752194805569553,
+      "loss": 2.0615,
+      "step": 1527
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002169091585110754,
+      "loss": 2.207,
+      "step": 1528
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021629699416148828,
+      "loss": 2.1089,
+      "step": 1529
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000215685456358875,
+      "loss": 2.0742,
+      "step": 1530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021507454645379258,
+      "loss": 2.0203,
+      "step": 1531
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021446426579541184,
+      "loss": 2.1318,
+      "step": 1532
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021385461573151387,
+      "loss": 2.0816,
+      "step": 1533
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021324559760848677,
+      "loss": 2.164,
+      "step": 1534
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021263721277132303,
+      "loss": 2.0285,
+      "step": 1535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002120294625636171,
+      "loss": 2.037,
+      "step": 1536
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002114223483275613,
+      "loss": 2.1123,
+      "step": 1537
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002108158714039435,
+      "loss": 2.0466,
+      "step": 1538
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021021003313214455,
+      "loss": 2.063,
+      "step": 1539
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020960483485013432,
+      "loss": 2.1197,
+      "step": 1540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002090002778944694,
+      "loss": 2.0891,
+      "step": 1541
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020839636360029025,
+      "loss": 2.08,
+      "step": 1542
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020779309330131818,
+      "loss": 2.037,
+      "step": 1543
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020719046832985184,
+      "loss": 2.1106,
+      "step": 1544
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002065884900167646,
+      "loss": 2.1008,
+      "step": 1545
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002059871596915024,
+      "loss": 2.0889,
+      "step": 1546
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002053864786820795,
+      "loss": 2.071,
+      "step": 1547
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020478644831507627,
+      "loss": 2.1288,
+      "step": 1548
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020418706991563634,
+      "loss": 2.1031,
+      "step": 1549
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020358834480746363,
+      "loss": 2.0265,
+      "step": 1550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002029902743128188,
+      "loss": 2.0225,
+      "step": 1551
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002023928597525174,
+      "loss": 2.0608,
+      "step": 1552
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020179610244592595,
+      "loss": 2.1572,
+      "step": 1553
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020120000371095937,
+      "loss": 2.0474,
+      "step": 1554
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002006045648640787,
+      "loss": 2.1332,
+      "step": 1555
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020000978722028713,
+      "loss": 2.1732,
+      "step": 1556
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019941567209312767,
+      "loss": 2.0226,
+      "step": 1557
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019882222079468036,
+      "loss": 2.0948,
+      "step": 1558
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001982294346355595,
+      "loss": 2.1375,
+      "step": 1559
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019763731492490976,
+      "loss": 2.1517,
+      "step": 1560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019704586297040422,
+      "loss": 2.1028,
+      "step": 1561
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001964550800782417,
+      "loss": 2.1255,
+      "step": 1562
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019586496755314288,
+      "loss": 2.0557,
+      "step": 1563
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019527552669834798,
+      "loss": 2.1103,
+      "step": 1564
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001946867588156142,
+      "loss": 2.1047,
+      "step": 1565
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019409866520521258,
+      "loss": 2.0343,
+      "step": 1566
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019351124716592455,
+      "loss": 2.1842,
+      "step": 1567
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001929245059950397,
+      "loss": 2.0596,
+      "step": 1568
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001923384429883533,
+      "loss": 2.1236,
+      "step": 1569
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019175305944016237,
+      "loss": 2.0418,
+      "step": 1570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019116835664326326,
+      "loss": 2.1076,
+      "step": 1571
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001905843358889497,
+      "loss": 2.0097,
+      "step": 1572
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019000099846700836,
+      "loss": 2.0513,
+      "step": 1573
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018941834566571692,
+      "loss": 2.1175,
+      "step": 1574
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018883637877184145,
+      "loss": 2.1082,
+      "step": 1575
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018825509907063325,
+      "loss": 2.0369,
+      "step": 1576
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018767450784582557,
+      "loss": 2.0999,
+      "step": 1577
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018709460637963122,
+      "loss": 2.1962,
+      "step": 1578
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018651539595274013,
+      "loss": 2.0722,
+      "step": 1579
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018593687784431578,
+      "loss": 2.0904,
+      "step": 1580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018535905333199248,
+      "loss": 2.0662,
+      "step": 1581
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001847819236918733,
+      "loss": 2.1274,
+      "step": 1582
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018420549019852655,
+      "loss": 2.117,
+      "step": 1583
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018362975412498266,
+      "loss": 2.0838,
+      "step": 1584
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018305471674273261,
+      "loss": 2.1017,
+      "step": 1585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001824803793217237,
+      "loss": 2.1705,
+      "step": 1586
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018190674313035737,
+      "loss": 2.0842,
+      "step": 1587
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018133380943548716,
+      "loss": 2.1317,
+      "step": 1588
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018076157950241452,
+      "loss": 2.0006,
+      "step": 1589
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018019005459488652,
+      "loss": 2.0939,
+      "step": 1590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017961923597509388,
+      "loss": 2.0571,
+      "step": 1591
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017904912490366722,
+      "loss": 2.0666,
+      "step": 1592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017847972263967433,
+      "loss": 2.1147,
+      "step": 1593
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001779110304406177,
+      "loss": 1.9505,
+      "step": 1594
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000177343049562432,
+      "loss": 2.1629,
+      "step": 1595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001767757812594807,
+      "loss": 2.1339,
+      "step": 1596
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001762092267845534,
+      "loss": 2.0822,
+      "step": 1597
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017564338738886365,
+      "loss": 2.0837,
+      "step": 1598
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001750782643220457,
+      "loss": 2.0826,
+      "step": 1599
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017451385883215166,
+      "loss": 2.1033,
+      "step": 1600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017395017216564863,
+      "loss": 2.0942,
+      "step": 1601
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017338720556741687,
+      "loss": 2.0635,
+      "step": 1602
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017282496028074606,
+      "loss": 2.0291,
+      "step": 1603
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017226343754733254,
+      "loss": 2.1295,
+      "step": 1604
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017170263860727769,
+      "loss": 2.21,
+      "step": 1605
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001711425646990838,
+      "loss": 2.0466,
+      "step": 1606
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017058321705965202,
+      "loss": 2.1533,
+      "step": 1607
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001700245969242798,
+      "loss": 2.0926,
+      "step": 1608
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016946670552665804,
+      "loss": 2.0887,
+      "step": 1609
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016890954409886795,
+      "loss": 2.2177,
+      "step": 1610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016835311387137836,
+      "loss": 1.9935,
+      "step": 1611
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001677974160730441,
+      "loss": 2.1974,
+      "step": 1612
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016724245193110176,
+      "loss": 2.0966,
+      "step": 1613
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016668822267116784,
+      "loss": 2.0654,
+      "step": 1614
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016613472951723597,
+      "loss": 2.1422,
+      "step": 1615
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016558197369167434,
+      "loss": 2.1319,
+      "step": 1616
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016502995641522216,
+      "loss": 2.1577,
+      "step": 1617
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016447867890698843,
+      "loss": 2.0214,
+      "step": 1618
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016392814238444753,
+      "loss": 2.099,
+      "step": 1619
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016337834806343782,
+      "loss": 2.1886,
+      "step": 1620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001628292971581588,
+      "loss": 2.1154,
+      "step": 1621
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016228099088116772,
+      "loss": 2.1359,
+      "step": 1622
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016173343044337734,
+      "loss": 2.1174,
+      "step": 1623
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016118661705405356,
+      "loss": 2.09,
+      "step": 1624
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016064055192081255,
+      "loss": 2.0364,
+      "step": 1625
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016009523624961757,
+      "loss": 2.0537,
+      "step": 1626
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015955067124477678,
+      "loss": 2.0666,
+      "step": 1627
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000159006858108941,
+      "loss": 2.1044,
+      "step": 1628
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015846379804310002,
+      "loss": 2.1147,
+      "step": 1629
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015792149224658054,
+      "loss": 2.1255,
+      "step": 1630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015737994191704385,
+      "loss": 2.1359,
+      "step": 1631
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001568391482504829,
+      "loss": 2.0515,
+      "step": 1632
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015629911244121903,
+      "loss": 2.0626,
+      "step": 1633
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001557598356819,
+      "loss": 2.1018,
+      "step": 1634
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015522131916349786,
+      "loss": 2.1277,
+      "step": 1635
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015468356407530493,
+      "loss": 2.1812,
+      "step": 1636
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015414657160493217,
+      "loss": 2.0942,
+      "step": 1637
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015361034293830673,
+      "loss": 1.99,
+      "step": 1638
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015307487925966844,
+      "loss": 2.0801,
+      "step": 1639
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015254018175156776,
+      "loss": 2.0476,
+      "step": 1640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001520062515948632,
+      "loss": 2.0522,
+      "step": 1641
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001514730899687189,
+      "loss": 2.1042,
+      "step": 1642
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015094069805060122,
+      "loss": 2.0628,
+      "step": 1643
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015040907701627666,
+      "loss": 2.1572,
+      "step": 1644
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014987822803980976,
+      "loss": 2.1597,
+      "step": 1645
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014934815229355965,
+      "loss": 2.1276,
+      "step": 1646
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014881885094817748,
+      "loss": 2.0468,
+      "step": 1647
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014829032517260488,
+      "loss": 2.097,
+      "step": 1648
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001477625761340704,
+      "loss": 2.1089,
+      "step": 1649
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001472356049980868,
+      "loss": 2.0159,
+      "step": 1650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014670941292844954,
+      "loss": 2.0471,
+      "step": 1651
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014618400108723295,
+      "loss": 2.078,
+      "step": 1652
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014565937063478862,
+      "loss": 2.099,
+      "step": 1653
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014513552272974207,
+      "loss": 2.0915,
+      "step": 1654
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001446124585289913,
+      "loss": 2.0894,
+      "step": 1655
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014409017918770266,
+      "loss": 2.1653,
+      "step": 1656
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014356868585930994,
+      "loss": 2.1269,
+      "step": 1657
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014304797969551077,
+      "loss": 2.0417,
+      "step": 1658
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014252806184626417,
+      "loss": 2.1218,
+      "step": 1659
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014200893345978817,
+      "loss": 2.0597,
+      "step": 1660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014149059568255778,
+      "loss": 2.0492,
+      "step": 1661
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014097304965930157,
+      "loss": 2.0453,
+      "step": 1662
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014045629653299953,
+      "loss": 2.0123,
+      "step": 1663
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013994033744488076,
+      "loss": 2.112,
+      "step": 1664
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013942517353442092,
+      "loss": 2.1249,
+      "step": 1665
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001389108059393391,
+      "loss": 2.0589,
+      "step": 1666
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013839723579559581,
+      "loss": 2.0376,
+      "step": 1667
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013788446423739103,
+      "loss": 2.0485,
+      "step": 1668
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013737249239716042,
+      "loss": 2.1064,
+      "step": 1669
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013686132140557355,
+      "loss": 2.0634,
+      "step": 1670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013635095239153188,
+      "loss": 1.9907,
+      "step": 1671
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013584138648216527,
+      "loss": 2.0644,
+      "step": 1672
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001353326248028298,
+      "loss": 2.1057,
+      "step": 1673
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013482466847710594,
+      "loss": 2.036,
+      "step": 1674
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013431751862679554,
+      "loss": 2.1096,
+      "step": 1675
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013381117637191887,
+      "loss": 2.0814,
+      "step": 1676
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013330564283071293,
+      "loss": 2.1145,
+      "step": 1677
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000132800919119629,
+      "loss": 2.0754,
+      "step": 1678
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013229700635332948,
+      "loss": 2.0668,
+      "step": 1679
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013179390564468585,
+      "loss": 2.0023,
+      "step": 1680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013129161810477641,
+      "loss": 2.0792,
+      "step": 1681
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001307901448428837,
+      "loss": 2.0541,
+      "step": 1682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001302894869664916,
+      "loss": 2.0897,
+      "step": 1683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012978964558128336,
+      "loss": 2.113,
+      "step": 1684
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012929062179113925,
+      "loss": 2.075,
+      "step": 1685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012879241669813368,
+      "loss": 2.0914,
+      "step": 1686
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012829503140253295,
+      "loss": 2.1633,
+      "step": 1687
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001277984670027933,
+      "loss": 2.0666,
+      "step": 1688
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012730272459555737,
+      "loss": 2.1223,
+      "step": 1689
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012680780527565312,
+      "loss": 2.06,
+      "step": 1690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001263137101360905,
+      "loss": 2.1725,
+      "step": 1691
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012582044026805922,
+      "loss": 2.1185,
+      "step": 1692
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012532799676092627,
+      "loss": 2.0098,
+      "step": 1693
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012483638070223414,
+      "loss": 2.0599,
+      "step": 1694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012434559317769752,
+      "loss": 2.0307,
+      "step": 1695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001238556352712012,
+      "loss": 2.1117,
+      "step": 1696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012336650806479827,
+      "loss": 2.1416,
+      "step": 1697
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012287821263870708,
+      "loss": 2.0924,
+      "step": 1698
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012239075007130885,
+      "loss": 2.0915,
+      "step": 1699
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012190412143914536,
+      "loss": 2.1094,
+      "step": 1700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001214183278169172,
+      "loss": 2.1761,
+      "step": 1701
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012093337027748042,
+      "loss": 2.0676,
+      "step": 1702
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012044924989184459,
+      "loss": 2.0847,
+      "step": 1703
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001199659677291709,
+      "loss": 2.0726,
+      "step": 1704
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011948352485676895,
+      "loss": 2.026,
+      "step": 1705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011900192234009477,
+      "loss": 2.1302,
+      "step": 1706
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011852116124274875,
+      "loss": 2.1108,
+      "step": 1707
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011804124262647314,
+      "loss": 2.074,
+      "step": 1708
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011756216755114929,
+      "loss": 2.0045,
+      "step": 1709
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011708393707479548,
+      "loss": 2.1042,
+      "step": 1710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011660655225356531,
+      "loss": 2.0819,
+      "step": 1711
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001161300141417444,
+      "loss": 2.1584,
+      "step": 1712
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011565432379174823,
+      "loss": 2.0473,
+      "step": 1713
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011517948225412056,
+      "loss": 2.1151,
+      "step": 1714
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011470549057753032,
+      "loss": 2.1992,
+      "step": 1715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011423234980876957,
+      "loss": 2.0184,
+      "step": 1716
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011376006099275099,
+      "loss": 2.0857,
+      "step": 1717
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011328862517250609,
+      "loss": 2.02,
+      "step": 1718
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011281804338918239,
+      "loss": 2.0443,
+      "step": 1719
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011234831668204115,
+      "loss": 2.0466,
+      "step": 1720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011187944608845569,
+      "loss": 2.0577,
+      "step": 1721
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011141143264390801,
+      "loss": 2.1138,
+      "step": 1722
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001109442773819877,
+      "loss": 2.0173,
+      "step": 1723
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001104779813343889,
+      "loss": 2.1883,
+      "step": 1724
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011001254553090812,
+      "loss": 1.9796,
+      "step": 1725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010954797099944186,
+      "loss": 2.0763,
+      "step": 1726
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001090842587659851,
+      "loss": 2.0795,
+      "step": 1727
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010862140985462804,
+      "loss": 1.9656,
+      "step": 1728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010815942528755418,
+      "loss": 2.0794,
+      "step": 1729
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010769830608503844,
+      "loss": 2.0424,
+      "step": 1730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010723805326544473,
+      "loss": 2.0307,
+      "step": 1731
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010677866784522316,
+      "loss": 2.0741,
+      "step": 1732
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010632015083890839,
+      "loss": 2.0706,
+      "step": 1733
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010586250325911745,
+      "loss": 2.0769,
+      "step": 1734
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010540572611654697,
+      "loss": 2.1257,
+      "step": 1735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010494982041997126,
+      "loss": 2.0228,
+      "step": 1736
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001044947871762405,
+      "loss": 2.1333,
+      "step": 1737
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010404062739027753,
+      "loss": 2.0523,
+      "step": 1738
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010358734206507641,
+      "loss": 2.066,
+      "step": 1739
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010313493220170017,
+      "loss": 2.034,
+      "step": 1740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010268339879927836,
+      "loss": 2.0703,
+      "step": 1741
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010223274285500466,
+      "loss": 2.0532,
+      "step": 1742
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010178296536413495,
+      "loss": 2.0873,
+      "step": 1743
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010133406731998546,
+      "loss": 2.0831,
+      "step": 1744
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010088604971392979,
+      "loss": 1.9786,
+      "step": 1745
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001004389135353972,
+      "loss": 2.0966,
+      "step": 1746
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.999265977187049e-05,
+      "loss": 2.1409,
+      "step": 1747
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.95472894088838e-05,
+      "loss": 2.0535,
+      "step": 1748
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.910280343001993e-05,
+      "loss": 2.0469,
+      "step": 1749
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.865920281690866e-05,
+      "loss": 2.1104,
+      "step": 1750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.821648854922482e-05,
+      "loss": 2.0309,
+      "step": 1751
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.77746616046854e-05,
+      "loss": 1.9851,
+      "step": 1752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.733372295904774e-05,
+      "loss": 2.0258,
+      "step": 1753
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.68936735861079e-05,
+      "loss": 2.0998,
+      "step": 1754
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.645451445769737e-05,
+      "loss": 2.118,
+      "step": 1755
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.601624654368196e-05,
+      "loss": 2.1518,
+      "step": 1756
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.557887081195938e-05,
+      "loss": 2.039,
+      "step": 1757
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.514238822845667e-05,
+      "loss": 2.1277,
+      "step": 1758
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.470679975712837e-05,
+      "loss": 2.0223,
+      "step": 1759
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.427210635995481e-05,
+      "loss": 2.0662,
+      "step": 1760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.383830899693923e-05,
+      "loss": 2.0895,
+      "step": 1761
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.340540862610591e-05,
+      "loss": 2.114,
+      "step": 1762
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.297340620349854e-05,
+      "loss": 2.1406,
+      "step": 1763
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.25423026831777e-05,
+      "loss": 2.0973,
+      "step": 1764
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.211209901721846e-05,
+      "loss": 2.1987,
+      "step": 1765
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.168279615570863e-05,
+      "loss": 1.9914,
+      "step": 1766
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.125439504674699e-05,
+      "loss": 2.0386,
+      "step": 1767
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.082689663644057e-05,
+      "loss": 2.0791,
+      "step": 1768
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.040030186890264e-05,
+      "loss": 2.065,
+      "step": 1769
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.997461168625138e-05,
+      "loss": 2.125,
+      "step": 1770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.954982702860664e-05,
+      "loss": 2.0784,
+      "step": 1771
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.912594883408865e-05,
+      "loss": 2.1218,
+      "step": 1772
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.870297803881589e-05,
+      "loss": 2.0914,
+      "step": 1773
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.828091557690287e-05,
+      "loss": 2.0735,
+      "step": 1774
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.785976238045801e-05,
+      "loss": 2.0429,
+      "step": 1775
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.743951937958144e-05,
+      "loss": 2.1134,
+      "step": 1776
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.702018750236357e-05,
+      "loss": 2.0702,
+      "step": 1777
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.660176767488237e-05,
+      "loss": 2.1788,
+      "step": 1778
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.618426082120146e-05,
+      "loss": 1.9731,
+      "step": 1779
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.576766786336854e-05,
+      "loss": 2.0054,
+      "step": 1780
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.535198972141294e-05,
+      "loss": 2.0418,
+      "step": 1781
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.493722731334347e-05,
+      "loss": 2.0749,
+      "step": 1782
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.452338155514644e-05,
+      "loss": 2.0793,
+      "step": 1783
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.411045336078426e-05,
+      "loss": 2.1661,
+      "step": 1784
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.369844364219264e-05,
+      "loss": 2.1078,
+      "step": 1785
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.328735330927873e-05,
+      "loss": 2.0298,
+      "step": 1786
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.287718326991961e-05,
+      "loss": 2.1641,
+      "step": 1787
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.246793442995954e-05,
+      "loss": 2.1215,
+      "step": 1788
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.205960769320875e-05,
+      "loss": 2.0113,
+      "step": 1789
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.165220396144085e-05,
+      "loss": 2.0632,
+      "step": 1790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.12457241343909e-05,
+      "loss": 2.0682,
+      "step": 1791
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.084016910975367e-05,
+      "loss": 2.0965,
+      "step": 1792
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.043553978318169e-05,
+      "loss": 2.0712,
+      "step": 1793
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.003183704828281e-05,
+      "loss": 2.0142,
+      "step": 1794
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.962906179661872e-05,
+      "loss": 2.0296,
+      "step": 1795
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.922721491770296e-05,
+      "loss": 2.0616,
+      "step": 1796
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.882629729899832e-05,
+      "loss": 2.0507,
+      "step": 1797
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.842630982591598e-05,
+      "loss": 2.096,
+      "step": 1798
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.802725338181232e-05,
+      "loss": 2.1142,
+      "step": 1799
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.762912884798812e-05,
+      "loss": 2.0421,
+      "step": 1800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.723193710368564e-05,
+      "loss": 2.0965,
+      "step": 1801
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.683567902608729e-05,
+      "loss": 2.0742,
+      "step": 1802
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.644035549031364e-05,
+      "loss": 2.1136,
+      "step": 1803
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.604596736942115e-05,
+      "loss": 2.054,
+      "step": 1804
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.56525155344004e-05,
+      "loss": 2.0584,
+      "step": 1805
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.52600008541745e-05,
+      "loss": 2.104,
+      "step": 1806
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.486842419559681e-05,
+      "loss": 2.1067,
+      "step": 1807
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.447778642344898e-05,
+      "loss": 2.0752,
+      "step": 1808
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.408808840043912e-05,
+      "loss": 1.9647,
+      "step": 1809
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.369933098720021e-05,
+      "loss": 1.9854,
+      "step": 1810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.331151504228767e-05,
+      "loss": 2.0391,
+      "step": 1811
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.292464142217775e-05,
+      "loss": 2.0302,
+      "step": 1812
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.25387109812658e-05,
+      "loss": 2.1139,
+      "step": 1813
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.215372457186415e-05,
+      "loss": 2.1694,
+      "step": 1814
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.176968304420007e-05,
+      "loss": 2.1448,
+      "step": 1815
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.138658724641417e-05,
+      "loss": 2.056,
+      "step": 1816
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.10044380245587e-05,
+      "loss": 2.0703,
+      "step": 1817
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.062323622259515e-05,
+      "loss": 2.0216,
+      "step": 1818
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.024298268239265e-05,
+      "loss": 2.1418,
+      "step": 1819
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.986367824372647e-05,
+      "loss": 2.0636,
+      "step": 1820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.948532374427541e-05,
+      "loss": 2.0118,
+      "step": 1821
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.910792001962063e-05,
+      "loss": 2.1433,
+      "step": 1822
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.873146790324358e-05,
+      "loss": 2.0895,
+      "step": 1823
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.83559682265239e-05,
+      "loss": 2.047,
+      "step": 1824
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.798142181873784e-05,
+      "loss": 2.0684,
+      "step": 1825
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.760782950705662e-05,
+      "loss": 2.0972,
+      "step": 1826
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.723519211654422e-05,
+      "loss": 1.9714,
+      "step": 1827
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.686351047015554e-05,
+      "loss": 2.1724,
+      "step": 1828
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.649278538873515e-05,
+      "loss": 2.0502,
+      "step": 1829
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.612301769101465e-05,
+      "loss": 2.0746,
+      "step": 1830
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.575420819361177e-05,
+      "loss": 2.1162,
+      "step": 1831
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.538635771102757e-05,
+      "loss": 2.0599,
+      "step": 1832
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.501946705564566e-05,
+      "loss": 2.0588,
+      "step": 1833
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.465353703772959e-05,
+      "loss": 2.087,
+      "step": 1834
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.428856846542136e-05,
+      "loss": 2.091,
+      "step": 1835
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.392456214473996e-05,
+      "loss": 2.0593,
+      "step": 1836
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.3561518879579e-05,
+      "loss": 2.1458,
+      "step": 1837
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.31994394717052e-05,
+      "loss": 1.9843,
+      "step": 1838
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.283832472075685e-05,
+      "loss": 2.0748,
+      "step": 1839
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.247817542424178e-05,
+      "loss": 2.0592,
+      "step": 1840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.211899237753559e-05,
+      "loss": 1.9536,
+      "step": 1841
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.176077637387984e-05,
+      "loss": 2.0588,
+      "step": 1842
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.140352820438066e-05,
+      "loss": 2.0187,
+      "step": 1843
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.104724865800665e-05,
+      "loss": 2.0577,
+      "step": 1844
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.069193852158711e-05,
+      "loss": 2.0941,
+      "step": 1845
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0337598579810584e-05,
+      "loss": 2.1577,
+      "step": 1846
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9984229615223096e-05,
+      "loss": 2.0071,
+      "step": 1847
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.963183240822606e-05,
+      "loss": 1.9676,
+      "step": 1848
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9280407737074825e-05,
+      "loss": 2.1238,
+      "step": 1849
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8929956377877125e-05,
+      "loss": 1.9759,
+      "step": 1850
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8580479104591075e-05,
+      "loss": 2.0921,
+      "step": 1851
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.823197668902341e-05,
+      "loss": 2.0206,
+      "step": 1852
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.78844499008282e-05,
+      "loss": 2.0707,
+      "step": 1853
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.753789950750454e-05,
+      "loss": 2.1262,
+      "step": 1854
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.719232627439558e-05,
+      "loss": 2.051,
+      "step": 1855
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6847730964686315e-05,
+      "loss": 2.0896,
+      "step": 1856
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.650411433940189e-05,
+      "loss": 2.0792,
+      "step": 1857
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.61614771574061e-05,
+      "loss": 2.0011,
+      "step": 1858
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.581982017539988e-05,
+      "loss": 2.0704,
+      "step": 1859
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5479144147919216e-05,
+      "loss": 2.1198,
+      "step": 1860
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.51394498273336e-05,
+      "loss": 2.1026,
+      "step": 1861
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.480073796384494e-05,
+      "loss": 2.0803,
+      "step": 1862
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.446300930548492e-05,
+      "loss": 2.0624,
+      "step": 1863
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.412626459811415e-05,
+      "loss": 2.055,
+      "step": 1864
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3790504585419954e-05,
+      "loss": 2.1526,
+      "step": 1865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.345573000891541e-05,
+      "loss": 2.0216,
+      "step": 1866
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.312194160793693e-05,
+      "loss": 2.1201,
+      "step": 1867
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.278914011964303e-05,
+      "loss": 2.1461,
+      "step": 1868
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2457326279013006e-05,
+      "loss": 2.1157,
+      "step": 1869
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2126500818844514e-05,
+      "loss": 2.0608,
+      "step": 1870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1796664469752566e-05,
+      "loss": 2.1064,
+      "step": 1871
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1467817960167975e-05,
+      "loss": 1.9304,
+      "step": 1872
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.113996201633536e-05,
+      "loss": 2.1146,
+      "step": 1873
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0813097362311765e-05,
+      "loss": 2.0619,
+      "step": 1874
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.048722471996475e-05,
+      "loss": 2.0111,
+      "step": 1875
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.016234480897158e-05,
+      "loss": 2.0959,
+      "step": 1876
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9838458346816664e-05,
+      "loss": 2.0212,
+      "step": 1877
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9515566048790485e-05,
+      "loss": 2.1361,
+      "step": 1878
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9193668627988074e-05,
+      "loss": 2.0559,
+      "step": 1879
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.887276679530744e-05,
+      "loss": 2.0899,
+      "step": 1880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.855286125944752e-05,
+      "loss": 2.1236,
+      "step": 1881
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8233952726907224e-05,
+      "loss": 2.0621,
+      "step": 1882
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7916041901983565e-05,
+      "loss": 2.0607,
+      "step": 1883
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7599129486770145e-05,
+      "loss": 2.104,
+      "step": 1884
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.728321618115555e-05,
+      "loss": 2.0792,
+      "step": 1885
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.696830268282204e-05,
+      "loss": 2.0492,
+      "step": 1886
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.665438968724361e-05,
+      "loss": 2.0482,
+      "step": 1887
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.634147788768489e-05,
+      "loss": 2.0901,
+      "step": 1888
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6029567975199414e-05,
+      "loss": 2.1038,
+      "step": 1889
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.571866063862795e-05,
+      "loss": 1.9922,
+      "step": 1890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.540875656459703e-05,
+      "loss": 2.0752,
+      "step": 1891
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.509985643751785e-05,
+      "loss": 2.0649,
+      "step": 1892
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.479196093958421e-05,
+      "loss": 2.0746,
+      "step": 1893
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4485070750771187e-05,
+      "loss": 2.0241,
+      "step": 1894
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.417918654883363e-05,
+      "loss": 2.0482,
+      "step": 1895
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3874309009305e-05,
+      "loss": 2.0527,
+      "step": 1896
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.357043880549538e-05,
+      "loss": 2.068,
+      "step": 1897
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.326757660849012e-05,
+      "loss": 2.0205,
+      "step": 1898
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2965723087148635e-05,
+      "loss": 2.0673,
+      "step": 1899
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.266487890810256e-05,
+      "loss": 2.0593,
+      "step": 1900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2365044735754365e-05,
+      "loss": 2.0954,
+      "step": 1901
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2066221232276266e-05,
+      "loss": 2.0877,
+      "step": 1902
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.176840905760815e-05,
+      "loss": 2.0428,
+      "step": 1903
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1471608869456443e-05,
+      "loss": 2.0188,
+      "step": 1904
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.117582132329284e-05,
+      "loss": 2.093,
+      "step": 1905
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.088104707235263e-05,
+      "loss": 1.9692,
+      "step": 1906
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.058728676763313e-05,
+      "loss": 2.0247,
+      "step": 1907
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0294541057892375e-05,
+      "loss": 2.0274,
+      "step": 1908
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.000281058964794e-05,
+      "loss": 2.0745,
+      "step": 1909
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.971209600717507e-05,
+      "loss": 2.0358,
+      "step": 1910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9422397952505465e-05,
+      "loss": 2.1185,
+      "step": 1911
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.913371706542596e-05,
+      "loss": 2.0425,
+      "step": 1912
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.884605398347707e-05,
+      "loss": 2.1394,
+      "step": 1913
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8559409341951456e-05,
+      "loss": 2.1011,
+      "step": 1914
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8273783773892404e-05,
+      "loss": 2.1391,
+      "step": 1915
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.798917791009293e-05,
+      "loss": 2.1209,
+      "step": 1916
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.770559237909393e-05,
+      "loss": 2.0351,
+      "step": 1917
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.742302780718288e-05,
+      "loss": 2.1256,
+      "step": 1918
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7141484818392635e-05,
+      "loss": 2.1331,
+      "step": 1919
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.686096403449973e-05,
+      "loss": 2.19,
+      "step": 1920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.658146607502344e-05,
+      "loss": 2.1408,
+      "step": 1921
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.630299155722411e-05,
+      "loss": 2.1291,
+      "step": 1922
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6025541096101676e-05,
+      "loss": 2.1449,
+      "step": 1923
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.574911530439473e-05,
+      "loss": 2.2031,
+      "step": 1924
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5473714792578606e-05,
+      "loss": 2.0459,
+      "step": 1925
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.519934016886478e-05,
+      "loss": 2.1016,
+      "step": 1926
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4925992039198776e-05,
+      "loss": 2.0611,
+      "step": 1927
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.465367100725908e-05,
+      "loss": 2.0723,
+      "step": 1928
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.438237767445618e-05,
+      "loss": 2.0074,
+      "step": 1929
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4112112639930804e-05,
+      "loss": 2.04,
+      "step": 1930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3842876500552564e-05,
+      "loss": 2.1011,
+      "step": 1931
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.357466985091906e-05,
+      "loss": 2.1036,
+      "step": 1932
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.330749328335414e-05,
+      "loss": 2.0451,
+      "step": 1933
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.304134738790659e-05,
+      "loss": 1.9643,
+      "step": 1934
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.277623275234953e-05,
+      "loss": 2.1092,
+      "step": 1935
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2512149962177994e-05,
+      "loss": 2.0986,
+      "step": 1936
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.224909960060851e-05,
+      "loss": 2.1133,
+      "step": 1937
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.198708224857755e-05,
+      "loss": 2.0998,
+      "step": 1938
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.172609848474023e-05,
+      "loss": 2.164,
+      "step": 1939
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1466148885468895e-05,
+      "loss": 2.076,
+      "step": 1940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.120723402485198e-05,
+      "loss": 2.0243,
+      "step": 1941
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.094935447469294e-05,
+      "loss": 2.0374,
+      "step": 1942
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.069251080450863e-05,
+      "loss": 2.07,
+      "step": 1943
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0436703581528113e-05,
+      "loss": 2.0521,
+      "step": 1944
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0181933370691694e-05,
+      "loss": 2.0022,
+      "step": 1945
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9928200734649523e-05,
+      "loss": 2.0179,
+      "step": 1946
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9675506233760142e-05,
+      "loss": 2.1367,
+      "step": 1947
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.942385042608925e-05,
+      "loss": 2.0642,
+      "step": 1948
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9173233867409054e-05,
+      "loss": 2.0455,
+      "step": 1949
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.892365711119638e-05,
+      "loss": 2.1295,
+      "step": 1950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8675120708631596e-05,
+      "loss": 2.149,
+      "step": 1951
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8427625208597764e-05,
+      "loss": 2.1171,
+      "step": 1952
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8181171157678874e-05,
+      "loss": 2.0782,
+      "step": 1953
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7935759100159053e-05,
+      "loss": 2.115,
+      "step": 1954
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7691389578021365e-05,
+      "loss": 2.0599,
+      "step": 1955
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7448063130946223e-05,
+      "loss": 2.012,
+      "step": 1956
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7205780296310544e-05,
+      "loss": 2.0935,
+      "step": 1957
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6964541609186378e-05,
+      "loss": 2.029,
+      "step": 1958
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6724347602340104e-05,
+      "loss": 2.1241,
+      "step": 1959
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6485198806230682e-05,
+      "loss": 2.0354,
+      "step": 1960
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6247095749008797e-05,
+      "loss": 2.1375,
+      "step": 1961
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6010038956515826e-05,
+      "loss": 1.9961,
+      "step": 1962
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5774028952282423e-05,
+      "loss": 2.1045,
+      "step": 1963
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5539066257527277e-05,
+      "loss": 2.0111,
+      "step": 1964
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.530515139115652e-05,
+      "loss": 1.9924,
+      "step": 1965
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5072284869761874e-05,
+      "loss": 2.0968,
+      "step": 1966
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4840467207619786e-05,
+      "loss": 1.9687,
+      "step": 1967
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.460969891669068e-05,
+      "loss": 2.126,
+      "step": 1968
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4379980506617272e-05,
+      "loss": 2.1104,
+      "step": 1969
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4151312484723464e-05,
+      "loss": 2.0819,
+      "step": 1970
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3923695356013798e-05,
+      "loss": 2.1042,
+      "step": 1971
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3697129623171833e-05,
+      "loss": 2.026,
+      "step": 1972
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3471615786559042e-05,
+      "loss": 2.0504,
+      "step": 1973
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3247154344213818e-05,
+      "loss": 2.021,
+      "step": 1974
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3023745791850625e-05,
+      "loss": 2.12,
+      "step": 1975
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2801390622858354e-05,
+      "loss": 1.9934,
+      "step": 1976
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2580089328299746e-05,
+      "loss": 2.056,
+      "step": 1977
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.235984239690997e-05,
+      "loss": 2.0701,
+      "step": 1978
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2140650315095934e-05,
+      "loss": 1.9618,
+      "step": 1979
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.192251356693459e-05,
+      "loss": 2.0843,
+      "step": 1980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.170543263417246e-05,
+      "loss": 2.1655,
+      "step": 1981
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1489407996224286e-05,
+      "loss": 2.041,
+      "step": 1982
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.127444013017199e-05,
+      "loss": 2.1072,
+      "step": 1983
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1060529510763648e-05,
+      "loss": 2.0757,
+      "step": 1984
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.084767661041259e-05,
+      "loss": 1.9672,
+      "step": 1985
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.063588189919596e-05,
+      "loss": 2.0148,
+      "step": 1986
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0425145844854275e-05,
+      "loss": 2.0913,
+      "step": 1987
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0215468912789693e-05,
+      "loss": 2.0937,
+      "step": 1988
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0006851566065575e-05,
+      "loss": 1.9849,
+      "step": 1989
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9799294265405166e-05,
+      "loss": 2.0497,
+      "step": 1990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9592797469190572e-05,
+      "loss": 2.1291,
+      "step": 1991
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.938736163346194e-05,
+      "loss": 2.0602,
+      "step": 1992
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9182987211916246e-05,
+      "loss": 2.0904,
+      "step": 1993
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8979674655906332e-05,
+      "loss": 2.1359,
+      "step": 1994
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8777424414440024e-05,
+      "loss": 2.1223,
+      "step": 1995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8576236934179202e-05,
+      "loss": 2.021,
+      "step": 1996
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8376112659438393e-05,
+      "loss": 2.0243,
+      "step": 1997
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8177052032184282e-05,
+      "loss": 2.0929,
+      "step": 1998
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7979055492034435e-05,
+      "loss": 2.09,
+      "step": 1999
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7782123476256407e-05,
+      "loss": 2.0799,
+      "step": 2000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7586256419766965e-05,
+      "loss": 2.0318,
+      "step": 2001
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7391454755130766e-05,
+      "loss": 2.0688,
+      "step": 2002
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7197718912559557e-05,
+      "loss": 2.0707,
+      "step": 2003
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.700504931991148e-05,
+      "loss": 2.0409,
+      "step": 2004
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.681344640268978e-05,
+      "loss": 2.0226,
+      "step": 2005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6622910584041974e-05,
+      "loss": 2.1081,
+      "step": 2006
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6433442284758903e-05,
+      "loss": 2.1642,
+      "step": 2007
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624504192327392e-05,
+      "loss": 2.1424,
+      "step": 2008
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6057709915661856e-05,
+      "loss": 2.076,
+      "step": 2009
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5871446675638057e-05,
+      "loss": 2.0996,
+      "step": 2010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5686252614557638e-05,
+      "loss": 2.0609,
+      "step": 2011
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5502128141414497e-05,
+      "loss": 2.058,
+      "step": 2012
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5319073662840188e-05,
+      "loss": 2.008,
+      "step": 2013
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5137089583103391e-05,
+      "loss": 2.0812,
+      "step": 2014
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4956176304108893e-05,
+      "loss": 2.0394,
+      "step": 2015
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4776334225396481e-05,
+      "loss": 2.025,
+      "step": 2016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4597563744140397e-05,
+      "loss": 2.0929,
+      "step": 2017
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4419865255148269e-05,
+      "loss": 2.0896,
+      "step": 2018
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4243239150860122e-05,
+      "loss": 2.0521,
+      "step": 2019
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4067685821347932e-05,
+      "loss": 2.1176,
+      "step": 2020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.389320565431429e-05,
+      "loss": 2.0096,
+      "step": 2021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3719799035091851e-05,
+      "loss": 2.0485,
+      "step": 2022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3547466346642278e-05,
+      "loss": 2.0576,
+      "step": 2023
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3376207969555577e-05,
+      "loss": 2.0306,
+      "step": 2024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.32060242820492e-05,
+      "loss": 2.0389,
+      "step": 2025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3036915659967118e-05,
+      "loss": 2.0058,
+      "step": 2026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2868882476779087e-05,
+      "loss": 2.0908,
+      "step": 2027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2701925103579815e-05,
+      "loss": 2.0841,
+      "step": 2028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2536043909088191e-05,
+      "loss": 2.026,
+      "step": 2029
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2371239259646228e-05,
+      "loss": 2.0495,
+      "step": 2030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2207511519218672e-05,
+      "loss": 2.0839,
+      "step": 2031
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2044861049391676e-05,
+      "loss": 2.1474,
+      "step": 2032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1883288209372512e-05,
+      "loss": 2.0051,
+      "step": 2033
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1722793355988471e-05,
+      "loss": 2.0592,
+      "step": 2034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1563376843686135e-05,
+      "loss": 2.1366,
+      "step": 2035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.140503902453055e-05,
+      "loss": 2.0689,
+      "step": 2036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1247780248204665e-05,
+      "loss": 1.9796,
+      "step": 2037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1091600862008333e-05,
+      "loss": 2.03,
+      "step": 2038
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0936501210857652e-05,
+      "loss": 2.0121,
+      "step": 2039
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0782481637284013e-05,
+      "loss": 2.0037,
+      "step": 2040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0629542481433663e-05,
+      "loss": 1.9718,
+      "step": 2041
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0477684081066751e-05,
+      "loss": 2.0491,
+      "step": 2042
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0326906771556566e-05,
+      "loss": 2.0001,
+      "step": 2043
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.017721088588891e-05,
+      "loss": 2.142,
+      "step": 2044
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0028596754661334e-05,
+      "loss": 1.9796,
+      "step": 2045
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.881064706082298e-06,
+      "loss": 2.0586,
+      "step": 2046
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.734615065970454e-06,
+      "loss": 2.0401,
+      "step": 2047
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.58924815775425e-06,
+      "loss": 2.0298,
+      "step": 2048
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.444964302470715e-06,
+      "loss": 2.0518,
+      "step": 2049
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.301763818765018e-06,
+      "loss": 2.1026,
+      "step": 2050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.15964702288996e-06,
+      "loss": 2.1947,
+      "step": 2051
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.018614228704925e-06,
+      "loss": 2.1025,
+      "step": 2052
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.878665747675152e-06,
+      "loss": 1.985,
+      "step": 2053
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.739801888871469e-06,
+      "loss": 2.1034,
+      "step": 2054
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.602022958969336e-06,
+      "loss": 2.0596,
+      "step": 2055
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.465329262248078e-06,
+      "loss": 2.0535,
+      "step": 2056
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.32972110059027e-06,
+      "loss": 2.0823,
+      "step": 2057
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.195198773481406e-06,
+      "loss": 2.043,
+      "step": 2058
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.061762578008613e-06,
+      "loss": 2.0695,
+      "step": 2059
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.929412808860559e-06,
+      "loss": 1.9819,
+      "step": 2060
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.79814975832649e-06,
+      "loss": 2.0607,
+      "step": 2061
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.667973716295851e-06,
+      "loss": 2.0422,
+      "step": 2062
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.5388849702571205e-06,
+      "loss": 2.0524,
+      "step": 2063
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.4108838052979185e-06,
+      "loss": 2.0307,
+      "step": 2064
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.283970504103732e-06,
+      "loss": 2.0426,
+      "step": 2065
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.1581453469575785e-06,
+      "loss": 2.0651,
+      "step": 2066
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.033408611739456e-06,
+      "loss": 2.0726,
+      "step": 2067
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.909760573925561e-06,
+      "loss": 2.1263,
+      "step": 2068
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.787201506587626e-06,
+      "loss": 2.0966,
+      "step": 2069
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.66573168039264e-06,
+      "loss": 2.0746,
+      "step": 2070
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.545351363601959e-06,
+      "loss": 2.1317,
+      "step": 2071
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.426060822070812e-06,
+      "loss": 2.1321,
+      "step": 2072
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.3078603192475716e-06,
+      "loss": 2.0335,
+      "step": 2073
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.1907501161735934e-06,
+      "loss": 2.1184,
+      "step": 2074
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.074730471482049e-06,
+      "loss": 2.034,
+      "step": 2075
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.959801641397755e-06,
+      "loss": 2.0128,
+      "step": 2076
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.845963879736627e-06,
+      "loss": 1.9919,
+      "step": 2077
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.733217437904892e-06,
+      "loss": 1.9901,
+      "step": 2078
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.621562564898597e-06,
+      "loss": 1.9913,
+      "step": 2079
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.51099950730316e-06,
+      "loss": 2.0989,
+      "step": 2080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.401528509292763e-06,
+      "loss": 2.0333,
+      "step": 2081
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.2931498126298495e-06,
+      "loss": 2.0706,
+      "step": 2082
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1858636566645135e-06,
+      "loss": 2.0499,
+      "step": 2083
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0796702783340035e-06,
+      "loss": 2.1447,
+      "step": 2084
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.97456991216233e-06,
+      "loss": 2.0227,
+      "step": 2085
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.870562790259325e-06,
+      "loss": 1.9986,
+      "step": 2086
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7676491423208625e-06,
+      "loss": 2.1057,
+      "step": 2087
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.66582919562758e-06,
+      "loss": 1.9724,
+      "step": 2088
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5651031750448825e-06,
+      "loss": 2.1093,
+      "step": 2089
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.465471303022217e-06,
+      "loss": 2.1192,
+      "step": 2090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.366933799592743e-06,
+      "loss": 2.0848,
+      "step": 2091
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.269490882372551e-06,
+      "loss": 2.0454,
+      "step": 2092
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1731427665606115e-06,
+      "loss": 1.9761,
+      "step": 2093
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.077889664937884e-06,
+      "loss": 2.0138,
+      "step": 2094
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.983731787867207e-06,
+      "loss": 2.0925,
+      "step": 2095
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.890669343292464e-06,
+      "loss": 2.0784,
+      "step": 2096
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7987025367384743e-06,
+      "loss": 2.1054,
+      "step": 2097
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.707831571310327e-06,
+      "loss": 2.1042,
+      "step": 2098
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6180566476929912e-06,
+      "loss": 2.0666,
+      "step": 2099
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.529377964150815e-06,
+      "loss": 1.9805,
+      "step": 2100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.441795716527307e-06,
+      "loss": 2.0726,
+      "step": 2101
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.355310098244302e-06,
+      "loss": 2.0751,
+      "step": 2102
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.269921300301959e-06,
+      "loss": 2.0549,
+      "step": 2103
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1856295112780988e-06,
+      "loss": 2.0723,
+      "step": 2104
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.102434917327812e-06,
+      "loss": 2.0449,
+      "step": 2105
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0203377021831292e-06,
+      "loss": 2.0705,
+      "step": 2106
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.939338047152573e-06,
+      "loss": 2.0254,
+      "step": 2107
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8594361311206073e-06,
+      "loss": 2.1122,
+      "step": 2108
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7806321305475225e-06,
+      "loss": 2.0774,
+      "step": 2109
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7029262194688818e-06,
+      "loss": 1.9792,
+      "step": 2110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.626318569495134e-06,
+      "loss": 2.0678,
+      "step": 2111
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.550809349811334e-06,
+      "loss": 2.0355,
+      "step": 2112
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.476398727176532e-06,
+      "loss": 2.0947,
+      "step": 2113
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4030868659237204e-06,
+      "loss": 2.114,
+      "step": 2114
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3308739279593317e-06,
+      "loss": 2.0437,
+      "step": 2115
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2597600727626845e-06,
+      "loss": 2.1009,
+      "step": 2116
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1897454573860387e-06,
+      "loss": 2.065,
+      "step": 2117
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1208302364538746e-06,
+      "loss": 2.0197,
+      "step": 2118
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0530145621627804e-06,
+      "loss": 2.0176,
+      "step": 2119
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9862985842810653e-06,
+      "loss": 2.0163,
+      "step": 2120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.920682450148259e-06,
+      "loss": 2.0174,
+      "step": 2121
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.856166304675111e-06,
+      "loss": 2.0082,
+      "step": 2122
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.792750290342926e-06,
+      "loss": 2.1367,
+      "step": 2123
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7304345472035632e-06,
+      "loss": 2.0183,
+      "step": 2124
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6692192128788253e-06,
+      "loss": 2.1474,
+      "step": 2125
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6091044225604035e-06,
+      "loss": 2.1252,
+      "step": 2126
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5500903090094888e-06,
+      "loss": 2.0053,
+      "step": 2127
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.492177002556383e-06,
+      "loss": 2.1673,
+      "step": 2128
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4353646311004443e-06,
+      "loss": 2.0227,
+      "step": 2129
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3796533201094752e-06,
+      "loss": 2.1036,
+      "step": 2130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3250431926197793e-06,
+      "loss": 2.044,
+      "step": 2131
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2715343692356607e-06,
+      "loss": 2.0915,
+      "step": 2132
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2191269681292582e-06,
+      "loss": 2.0659,
+      "step": 2133
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1678211050402676e-06,
+      "loss": 2.0654,
+      "step": 2134
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.117616893275719e-06,
+      "loss": 2.0421,
+      "step": 2135
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.068514443709534e-06,
+      "loss": 2.04,
+      "step": 2136
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0205138647826905e-06,
+      "loss": 2.0461,
+      "step": 2137
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.73615262502503e-07,
+      "loss": 2.1143,
+      "step": 2138
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.278187404426763e-07,
+      "loss": 2.1394,
+      "step": 2139
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.831243997431404e-07,
+      "loss": 2.1073,
+      "step": 2140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.395323391094944e-07,
+      "loss": 2.0744,
+      "step": 2141
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.970426548131183e-07,
+      "loss": 2.0364,
+      "step": 2142
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.556554406908389e-07,
+      "loss": 2.0631,
+      "step": 2143
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.153707881446536e-07,
+      "loss": 2.0452,
+      "step": 2144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.761887861417293e-07,
+      "loss": 2.042,
+      "step": 2145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.381095212139032e-07,
+      "loss": 2.0044,
+      "step": 2146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.011330774577384e-07,
+      "loss": 1.9751,
+      "step": 2147
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.652595365343016e-07,
+      "loss": 2.0012,
+      "step": 2148
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.304889776688859e-07,
+      "loss": 2.0492,
+      "step": 2149
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.968214776508994e-07,
+      "loss": 2.0123,
+      "step": 2150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6425711083375454e-07,
+      "loss": 2.1075,
+      "step": 2151
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3279594913447906e-07,
+      "loss": 2.1117,
+      "step": 2152
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.02438062033883e-07,
+      "loss": 2.1119,
+      "step": 2153
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7318351657616987e-07,
+      "loss": 2.1237,
+      "step": 2154
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4503237736882573e-07,
+      "loss": 1.9835,
+      "step": 2155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.179847065825081e-07,
+      "loss": 2.0092,
+      "step": 2156
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9204056395104594e-07,
+      "loss": 1.9905,
+      "step": 2157
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.672000067709956e-07,
+      "loss": 2.0618,
+      "step": 2158
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4346308990175204e-07,
+      "loss": 2.0225,
+      "step": 2159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.208298657653818e-07,
+      "loss": 2.035,
+      "step": 2160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9930038434645692e-07,
+      "loss": 2.0605,
+      "step": 2161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7887469319205484e-07,
+      "loss": 2.0846,
+      "step": 2162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5955283741142523e-07,
+      "loss": 2.1114,
+      "step": 2163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4133485967615655e-07,
+      "loss": 2.0099,
+      "step": 2164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2422080021995407e-07,
+      "loss": 2.0333,
+      "step": 2165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.082106968385288e-07,
+      "loss": 2.1221,
+      "step": 2166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.330458488959748e-08,
+      "loss": 2.0612,
+      "step": 2167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.950249729271608e-08,
+      "loss": 2.0647,
+      "step": 2168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.680446452922429e-08,
+      "loss": 2.0805,
+      "step": 2169
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.521051464230098e-08,
+      "loss": 2.1894,
+      "step": 2170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4720673236631206e-08,
+      "loss": 2.0118,
+      "step": 2171
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.53349634786837e-08,
+      "loss": 2.024,
+      "step": 2172
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7053406096433365e-08,
+      "loss": 2.1162,
+      "step": 2173
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.987601937930572e-08,
+      "loss": 2.0662,
+      "step": 2174
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3802819178398984e-08,
+      "loss": 2.075,
+      "step": 2175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.833818906039959e-09,
+      "loss": 2.04,
+      "step": 2176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.969029536061598e-09,
+      "loss": 2.09,
+      "step": 2177
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2084596038030037e-09,
+      "loss": 2.0409,
+      "step": 2178
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.521152057763601e-10,
+      "loss": 2.093,
+      "step": 2179
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 2.076,
+      "step": 2180
+    },
+    {
+      "epoch": 1.0,
+      "step": 2180,
+      "total_flos": 4.728618313758802e+17,
+      "train_loss": 2.2258731907660807,
+      "train_runtime": 41231.7221,
+      "train_samples_per_second": 13.536,
+      "train_steps_per_second": 0.053
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2180,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 24000,
+  "total_flos": 4.728618313758802e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}