diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,60216 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.807411566535654,
+  "global_step": 20000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999953209807224e-05,
+      "loss": 0.037,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999906419614449e-05,
+      "loss": 0.0443,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999859629421673e-05,
+      "loss": 0.0456,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999812839228898e-05,
+      "loss": 0.0554,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999766049036122e-05,
+      "loss": 0.0675,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997192588433464e-05,
+      "loss": 0.0875,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999672468650571e-05,
+      "loss": 0.0372,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9996256784577956e-05,
+      "loss": 0.0759,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9995788882650195e-05,
+      "loss": 0.0565,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999532098072244e-05,
+      "loss": 0.0625,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999485307879469e-05,
+      "loss": 0.0819,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999438517686693e-05,
+      "loss": 0.0702,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999391727493917e-05,
+      "loss": 0.0602,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999344937301142e-05,
+      "loss": 0.0636,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9992981471083664e-05,
+      "loss": 0.0527,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999251356915591e-05,
+      "loss": 0.0542,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999204566722815e-05,
+      "loss": 0.0554,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991577765300395e-05,
+      "loss": 0.097,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991109863372634e-05,
+      "loss": 0.0558,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999064196144489e-05,
+      "loss": 0.0606,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990174059517126e-05,
+      "loss": 0.0645,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998970615758937e-05,
+      "loss": 0.0463,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998923825566161e-05,
+      "loss": 0.062,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988770353733864e-05,
+      "loss": 0.0648,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99883024518061e-05,
+      "loss": 0.0652,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998783454987835e-05,
+      "loss": 0.0588,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998736664795059e-05,
+      "loss": 0.0534,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986898746022834e-05,
+      "loss": 0.0591,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998643084409508e-05,
+      "loss": 0.0734,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985962942167326e-05,
+      "loss": 0.0554,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9985495040239565e-05,
+      "loss": 0.0694,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998502713831181e-05,
+      "loss": 0.0483,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984559236384057e-05,
+      "loss": 0.06,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99840913344563e-05,
+      "loss": 0.0609,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998362343252854e-05,
+      "loss": 0.0651,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998315553060079e-05,
+      "loss": 0.0729,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998268762867303e-05,
+      "loss": 0.0757,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998221972674528e-05,
+      "loss": 0.0727,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998175182481752e-05,
+      "loss": 0.0758,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981283922889764e-05,
+      "loss": 0.0541,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998081602096201e-05,
+      "loss": 0.0508,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980348119034256e-05,
+      "loss": 0.061,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9979880217106495e-05,
+      "loss": 0.065,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997941231517874e-05,
+      "loss": 0.0799,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997894441325098e-05,
+      "loss": 0.0496,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997847651132323e-05,
+      "loss": 0.083,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997800860939547e-05,
+      "loss": 0.0735,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997754070746772e-05,
+      "loss": 0.0632,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997707280553996e-05,
+      "loss": 0.0583,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99766049036122e-05,
+      "loss": 0.0603,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997613700168445e-05,
+      "loss": 0.0608,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975669099756695e-05,
+      "loss": 0.0535,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975201197828934e-05,
+      "loss": 0.0497,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997473329590118e-05,
+      "loss": 0.0643,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9974265393973426e-05,
+      "loss": 0.0945,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997379749204567e-05,
+      "loss": 0.0629,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997332959011791e-05,
+      "loss": 0.0662,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997286168819016e-05,
+      "loss": 0.0684,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99723937862624e-05,
+      "loss": 0.0457,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997192588433465e-05,
+      "loss": 0.0592,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997145798240689e-05,
+      "loss": 0.0606,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9970990080479134e-05,
+      "loss": 0.0658,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997052217855138e-05,
+      "loss": 0.0649,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9970054276623626e-05,
+      "loss": 0.082,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9969586374695865e-05,
+      "loss": 0.064,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996911847276811e-05,
+      "loss": 0.0648,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996865057084035e-05,
+      "loss": 0.0601,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99681826689126e-05,
+      "loss": 0.0887,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996771476698484e-05,
+      "loss": 0.0643,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996724686505709e-05,
+      "loss": 0.0608,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996677896312933e-05,
+      "loss": 0.0526,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996631106120158e-05,
+      "loss": 0.0693,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996584315927382e-05,
+      "loss": 0.0642,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9965375257346064e-05,
+      "loss": 0.0506,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9964907355418303e-05,
+      "loss": 0.0759,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996443945349055e-05,
+      "loss": 0.0666,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9963971551562795e-05,
+      "loss": 0.0838,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996350364963504e-05,
+      "loss": 0.0516,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996303574770728e-05,
+      "loss": 0.0821,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9962567845779526e-05,
+      "loss": 0.0562,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996209994385177e-05,
+      "loss": 0.0587,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996163204192402e-05,
+      "loss": 0.0565,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996116413999626e-05,
+      "loss": 0.0792,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99606962380685e-05,
+      "loss": 0.0581,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996022833614075e-05,
+      "loss": 0.0609,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9959760434212995e-05,
+      "loss": 0.0733,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9959292532285234e-05,
+      "loss": 0.0535,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995882463035748e-05,
+      "loss": 0.0691,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9958356728429726e-05,
+      "loss": 0.0531,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995788882650197e-05,
+      "loss": 0.0678,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995742092457421e-05,
+      "loss": 0.071,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995695302264646e-05,
+      "loss": 0.0658,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9956485120718696e-05,
+      "loss": 0.0588,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995601721879095e-05,
+      "loss": 0.076,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995554931686319e-05,
+      "loss": 0.082,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9955081414935434e-05,
+      "loss": 0.0749,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995461351300767e-05,
+      "loss": 0.0777,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9954145611079926e-05,
+      "loss": 0.0482,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9953677709152165e-05,
+      "loss": 0.0717,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995320980722441e-05,
+      "loss": 0.0564,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995274190529665e-05,
+      "loss": 0.0572,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9952274003368896e-05,
+      "loss": 0.0558,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995180610144114e-05,
+      "loss": 0.0597,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995133819951339e-05,
+      "loss": 0.0753,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995087029758563e-05,
+      "loss": 0.0674,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995040239565787e-05,
+      "loss": 0.0579,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994993449373012e-05,
+      "loss": 0.0707,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9949466591802364e-05,
+      "loss": 0.0689,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9948998689874604e-05,
+      "loss": 0.052,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994853078794685e-05,
+      "loss": 0.0635,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9948062886019095e-05,
+      "loss": 0.0776,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994759498409134e-05,
+      "loss": 0.0745,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994712708216358e-05,
+      "loss": 0.0715,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9946659180235826e-05,
+      "loss": 0.0822,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9946191278308065e-05,
+      "loss": 0.0503,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994572337638032e-05,
+      "loss": 0.0659,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994525547445256e-05,
+      "loss": 0.0601,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99447875725248e-05,
+      "loss": 0.0695,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994431967059704e-05,
+      "loss": 0.0656,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9943851768669295e-05,
+      "loss": 0.0464,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9943383866741534e-05,
+      "loss": 0.0717,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994291596481378e-05,
+      "loss": 0.0721,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994244806288602e-05,
+      "loss": 0.0659,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9941980160958265e-05,
+      "loss": 0.0624,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994151225903051e-05,
+      "loss": 0.0821,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994104435710276e-05,
+      "loss": 0.0659,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9940576455174996e-05,
+      "loss": 0.0722,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994010855324724e-05,
+      "loss": 0.0651,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993964065131949e-05,
+      "loss": 0.063,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993917274939173e-05,
+      "loss": 0.0858,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993870484746397e-05,
+      "loss": 0.065,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993823694553621e-05,
+      "loss": 0.0508,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9937769043608465e-05,
+      "loss": 0.0733,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9937301141680704e-05,
+      "loss": 0.0532,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993683323975295e-05,
+      "loss": 0.0611,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993636533782519e-05,
+      "loss": 0.0675,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993589743589744e-05,
+      "loss": 0.0732,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993542953396968e-05,
+      "loss": 0.0837,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993496163204193e-05,
+      "loss": 0.0673,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9934493730114166e-05,
+      "loss": 0.0694,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993402582818641e-05,
+      "loss": 0.0588,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993355792625866e-05,
+      "loss": 0.0712,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9933090024330904e-05,
+      "loss": 0.0729,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993262212240314e-05,
+      "loss": 0.0628,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993215422047539e-05,
+      "loss": 0.0781,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9931686318547634e-05,
+      "loss": 0.0719,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993121841661988e-05,
+      "loss": 0.0735,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993075051469212e-05,
+      "loss": 0.0831,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9930282612764365e-05,
+      "loss": 0.0536,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992981471083661e-05,
+      "loss": 0.0818,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992934680890886e-05,
+      "loss": 0.0652,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9928878906981096e-05,
+      "loss": 0.0744,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992841100505334e-05,
+      "loss": 0.0641,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992794310312559e-05,
+      "loss": 0.0605,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9927475201197834e-05,
+      "loss": 0.0653,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992700729927007e-05,
+      "loss": 0.0743,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992653939734232e-05,
+      "loss": 0.0614,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992607149541456e-05,
+      "loss": 0.0706,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992560359348681e-05,
+      "loss": 0.0712,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.992513569155905e-05,
+      "loss": 0.0667,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9924667789631296e-05,
+      "loss": 0.0829,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9924199887703535e-05,
+      "loss": 0.0632,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992373198577579e-05,
+      "loss": 0.0668,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992326408384803e-05,
+      "loss": 0.074,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992279618192027e-05,
+      "loss": 0.0648,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992232827999251e-05,
+      "loss": 0.0703,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992186037806476e-05,
+      "loss": 0.0689,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9921392476137004e-05,
+      "loss": 0.0587,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992092457420925e-05,
+      "loss": 0.0486,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992045667228149e-05,
+      "loss": 0.0688,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9919988770353735e-05,
+      "loss": 0.0746,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991952086842598e-05,
+      "loss": 0.064,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991905296649823e-05,
+      "loss": 0.0703,
+      "step": 346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9918585064570466e-05,
+      "loss": 0.0582,
+      "step": 348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991811716264271e-05,
+      "loss": 0.0926,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991764926071496e-05,
+      "loss": 0.073,
+      "step": 352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9917181358787204e-05,
+      "loss": 0.0732,
+      "step": 354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991671345685944e-05,
+      "loss": 0.0916,
+      "step": 356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991624555493169e-05,
+      "loss": 0.055,
+      "step": 358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9915777653003934e-05,
+      "loss": 0.0583,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991530975107618e-05,
+      "loss": 0.0512,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991484184914842e-05,
+      "loss": 0.0686,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9914373947220665e-05,
+      "loss": 0.0654,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9913906045292905e-05,
+      "loss": 0.0615,
+      "step": 368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991343814336516e-05,
+      "loss": 0.0862,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9912970241437396e-05,
+      "loss": 0.0641,
+      "step": 372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991250233950964e-05,
+      "loss": 0.0601,
+      "step": 374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991203443758188e-05,
+      "loss": 0.0584,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991156653565413e-05,
+      "loss": 0.0663,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991109863372637e-05,
+      "loss": 0.0858,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991063073179862e-05,
+      "loss": 0.0638,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991016282987086e-05,
+      "loss": 0.0453,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9909694927943104e-05,
+      "loss": 0.0616,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.990922702601535e-05,
+      "loss": 0.0885,
+      "step": 388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9908759124087596e-05,
+      "loss": 0.078,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9908291222159835e-05,
+      "loss": 0.0484,
+      "step": 392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990782332023208e-05,
+      "loss": 0.0595,
+      "step": 394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990735541830433e-05,
+      "loss": 0.0726,
+      "step": 396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990688751637657e-05,
+      "loss": 0.0838,
+      "step": 398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990641961444881e-05,
+      "loss": 0.0652,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990595171252106e-05,
+      "loss": 0.0673,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9905483810593304e-05,
+      "loss": 0.055,
+      "step": 404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990501590866555e-05,
+      "loss": 0.0759,
+      "step": 406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990454800673779e-05,
+      "loss": 0.0616,
+      "step": 408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9904080104810035e-05,
+      "loss": 0.0933,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9903612202882274e-05,
+      "loss": 0.0647,
+      "step": 412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990314430095453e-05,
+      "loss": 0.0527,
+      "step": 414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9902676399026766e-05,
+      "loss": 0.0578,
+      "step": 416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990220849709901e-05,
+      "loss": 0.0808,
+      "step": 418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990174059517125e-05,
+      "loss": 0.0793,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9901272693243504e-05,
+      "loss": 0.0648,
+      "step": 422
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990080479131574e-05,
+      "loss": 0.0716,
+      "step": 424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990033688938799e-05,
+      "loss": 0.0825,
+      "step": 426
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989986898746023e-05,
+      "loss": 0.0908,
+      "step": 428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9899401085532474e-05,
+      "loss": 0.0688,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989893318360472e-05,
+      "loss": 0.0669,
+      "step": 432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9898465281676965e-05,
+      "loss": 0.0735,
+      "step": 434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9897997379749205e-05,
+      "loss": 0.0817,
+      "step": 436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989752947782145e-05,
+      "loss": 0.0792,
+      "step": 438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9897061575893696e-05,
+      "loss": 0.0509,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989659367396594e-05,
+      "loss": 0.0809,
+      "step": 442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989612577203818e-05,
+      "loss": 0.0654,
+      "step": 444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989565787011043e-05,
+      "loss": 0.0608,
+      "step": 446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989518996818267e-05,
+      "loss": 0.0928,
+      "step": 448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989472206625492e-05,
+      "loss": 0.0554,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989425416432716e-05,
+      "loss": 0.0559,
+      "step": 452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9893786262399404e-05,
+      "loss": 0.0871,
+      "step": 454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989331836047165e-05,
+      "loss": 0.0724,
+      "step": 456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9892850458543896e-05,
+      "loss": 0.0672,
+      "step": 458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9892382556616135e-05,
+      "loss": 0.0602,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989191465468838e-05,
+      "loss": 0.0616,
+      "step": 462
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.989144675276062e-05,
+      "loss": 0.0732,
+      "step": 464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.989097885083287e-05,
+      "loss": 0.0668,
+      "step": 466
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.989051094890511e-05,
+      "loss": 0.0848,
+      "step": 468
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.989004304697736e-05,
+      "loss": 0.0656,
+      "step": 470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.98895751450496e-05,
+      "loss": 0.0609,
+      "step": 472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.988910724312185e-05,
+      "loss": 0.067,
+      "step": 474
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.988863934119409e-05,
+      "loss": 0.0814,
+      "step": 476
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9888171439266335e-05,
+      "loss": 0.0736,
+      "step": 478
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9887703537338574e-05,
+      "loss": 0.0638,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.988723563541082e-05,
+      "loss": 0.0587,
+      "step": 482
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9886767733483066e-05,
+      "loss": 0.056,
+      "step": 484
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.988629983155531e-05,
+      "loss": 0.0593,
+      "step": 486
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.988583192962755e-05,
+      "loss": 0.0738,
+      "step": 488
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.98853640276998e-05,
+      "loss": 0.0849,
+      "step": 490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.988489612577204e-05,
+      "loss": 0.0673,
+      "step": 492
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.988442822384429e-05,
+      "loss": 0.0864,
+      "step": 494
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.988396032191653e-05,
+      "loss": 0.0738,
+      "step": 496
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9883492419988774e-05,
+      "loss": 0.0748,
+      "step": 498
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.988302451806102e-05,
+      "loss": 0.0529,
+      "step": 500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9882556616133265e-05,
+      "loss": 0.0544,
+      "step": 502
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9882088714205505e-05,
+      "loss": 0.077,
+      "step": 504
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.988162081227775e-05,
+      "loss": 0.0706,
+      "step": 506
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.988115291034999e-05,
+      "loss": 0.0729,
+      "step": 508
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9880685008422236e-05,
+      "loss": 0.0812,
+      "step": 510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.988021710649448e-05,
+      "loss": 0.067,
+      "step": 512
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.987974920456672e-05,
+      "loss": 0.0565,
+      "step": 514
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9879281302638967e-05,
+      "loss": 0.0597,
+      "step": 516
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.987881340071121e-05,
+      "loss": 0.095,
+      "step": 518
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.987834549878346e-05,
+      "loss": 0.0729,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.98778775968557e-05,
+      "loss": 0.0743,
+      "step": 522
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9877409694927943e-05,
+      "loss": 0.0656,
+      "step": 524
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.987694179300019e-05,
+      "loss": 0.0656,
+      "step": 526
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9876473891072435e-05,
+      "loss": 0.0761,
+      "step": 528
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9876005989144674e-05,
+      "loss": 0.0566,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.987553808721692e-05,
+      "loss": 0.0812,
+      "step": 532
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9875070185289166e-05,
+      "loss": 0.0558,
+      "step": 534
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.987460228336141e-05,
+      "loss": 0.08,
+      "step": 536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.987413438143365e-05,
+      "loss": 0.0568,
+      "step": 538
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.98736664795059e-05,
+      "loss": 0.059,
+      "step": 540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9873198577578136e-05,
+      "loss": 0.0801,
+      "step": 542
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.987273067565039e-05,
+      "loss": 0.0843,
+      "step": 544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.987226277372263e-05,
+      "loss": 0.0663,
+      "step": 546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9871794871794874e-05,
+      "loss": 0.0791,
+      "step": 548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.987132696986711e-05,
+      "loss": 0.0614,
+      "step": 550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9870859067939366e-05,
+      "loss": 0.0644,
+      "step": 552
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9870391166011605e-05,
+      "loss": 0.0819,
+      "step": 554
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.986992326408385e-05,
+      "loss": 0.0612,
+      "step": 556
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.986945536215609e-05,
+      "loss": 0.0671,
+      "step": 558
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9868987460228336e-05,
+      "loss": 0.0697,
+      "step": 560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.986851955830058e-05,
+      "loss": 0.0684,
+      "step": 562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.986805165637283e-05,
+      "loss": 0.0503,
+      "step": 564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.986758375444507e-05,
+      "loss": 0.1045,
+      "step": 566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.986711585251731e-05,
+      "loss": 0.0755,
+      "step": 568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.986664795058956e-05,
+      "loss": 0.0808,
+      "step": 570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9866180048661805e-05,
+      "loss": 0.0615,
+      "step": 572
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9865712146734044e-05,
+      "loss": 0.0703,
+      "step": 574
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.986524424480629e-05,
+      "loss": 0.0816,
+      "step": 576
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9864776342878536e-05,
+      "loss": 0.1,
+      "step": 578
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.986430844095078e-05,
+      "loss": 0.0772,
+      "step": 580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.986384053902302e-05,
+      "loss": 0.0676,
+      "step": 582
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9863372637095267e-05,
+      "loss": 0.065,
+      "step": 584
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.986290473516751e-05,
+      "loss": 0.0518,
+      "step": 586
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.986243683323976e-05,
+      "loss": 0.0837,
+      "step": 588
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9861968931312e-05,
+      "loss": 0.0935,
+      "step": 590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9861501029384243e-05,
+      "loss": 0.0638,
+      "step": 592
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.986103312745648e-05,
+      "loss": 0.0594,
+      "step": 594
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9860565225528735e-05,
+      "loss": 0.0801,
+      "step": 596
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9860097323600974e-05,
+      "loss": 0.083,
+      "step": 598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.985962942167322e-05,
+      "loss": 0.0821,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.985916151974546e-05,
+      "loss": 0.0725,
+      "step": 602
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.985869361781771e-05,
+      "loss": 0.0954,
+      "step": 604
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.985822571588995e-05,
+      "loss": 0.0728,
+      "step": 606
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.98577578139622e-05,
+      "loss": 0.0852,
+      "step": 608
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9857289912034436e-05,
+      "loss": 0.0623,
+      "step": 610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.985682201010668e-05,
+      "loss": 0.0678,
+      "step": 612
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.985635410817893e-05,
+      "loss": 0.0754,
+      "step": 614
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9855886206251174e-05,
+      "loss": 0.0643,
+      "step": 616
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.985541830432341e-05,
+      "loss": 0.0607,
+      "step": 618
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.985495040239566e-05,
+      "loss": 0.0761,
+      "step": 620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9854482500467905e-05,
+      "loss": 0.0524,
+      "step": 622
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.985401459854015e-05,
+      "loss": 0.0739,
+      "step": 624
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.985354669661239e-05,
+      "loss": 0.0671,
+      "step": 626
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9853078794684636e-05,
+      "loss": 0.1179,
+      "step": 628
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.985261089275688e-05,
+      "loss": 0.0855,
+      "step": 630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.985214299082913e-05,
+      "loss": 0.0785,
+      "step": 632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.985167508890137e-05,
+      "loss": 0.085,
+      "step": 634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.985120718697361e-05,
+      "loss": 0.0555,
+      "step": 636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.985073928504586e-05,
+      "loss": 0.0648,
+      "step": 638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9850271383118105e-05,
+      "loss": 0.0983,
+      "step": 640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9849803481190344e-05,
+      "loss": 0.065,
+      "step": 642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.984933557926259e-05,
+      "loss": 0.0807,
+      "step": 644
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.984886767733483e-05,
+      "loss": 0.081,
+      "step": 646
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.984839977540708e-05,
+      "loss": 0.067,
+      "step": 648
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.984793187347932e-05,
+      "loss": 0.0582,
+      "step": 650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9847463971551567e-05,
+      "loss": 0.061,
+      "step": 652
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9846996069623806e-05,
+      "loss": 0.0615,
+      "step": 654
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.984652816769605e-05,
+      "loss": 0.0744,
+      "step": 656
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.98460602657683e-05,
+      "loss": 0.0589,
+      "step": 658
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9845592363840543e-05,
+      "loss": 0.0687,
+      "step": 660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.984512446191278e-05,
+      "loss": 0.0741,
+      "step": 662
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.984465655998503e-05,
+      "loss": 0.0739,
+      "step": 664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9844188658057274e-05,
+      "loss": 0.0726,
+      "step": 666
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.984372075612952e-05,
+      "loss": 0.0584,
+      "step": 668
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.984325285420176e-05,
+      "loss": 0.0637,
+      "step": 670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9842784952274005e-05,
+      "loss": 0.0623,
+      "step": 672
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.984231705034625e-05,
+      "loss": 0.0817,
+      "step": 674
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.98418491484185e-05,
+      "loss": 0.0579,
+      "step": 676
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9841381246490736e-05,
+      "loss": 0.0798,
+      "step": 678
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.984091334456298e-05,
+      "loss": 0.08,
+      "step": 680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.984044544263523e-05,
+      "loss": 0.0769,
+      "step": 682
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9839977540707474e-05,
+      "loss": 0.0791,
+      "step": 684
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.983950963877971e-05,
+      "loss": 0.0704,
+      "step": 686
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.983904173685196e-05,
+      "loss": 0.0652,
+      "step": 688
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.98385738349242e-05,
+      "loss": 0.0667,
+      "step": 690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.983810593299645e-05,
+      "loss": 0.1066,
+      "step": 692
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.983763803106869e-05,
+      "loss": 0.0738,
+      "step": 694
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9837170129140936e-05,
+      "loss": 0.0779,
+      "step": 696
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9836702227213175e-05,
+      "loss": 0.0779,
+      "step": 698
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.983623432528543e-05,
+      "loss": 0.0608,
+      "step": 700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.983576642335767e-05,
+      "loss": 0.082,
+      "step": 702
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.983529852142991e-05,
+      "loss": 0.0693,
+      "step": 704
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.983483061950215e-05,
+      "loss": 0.0889,
+      "step": 706
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.98343627175744e-05,
+      "loss": 0.0858,
+      "step": 708
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9833894815646644e-05,
+      "loss": 0.0742,
+      "step": 710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.983342691371889e-05,
+      "loss": 0.0764,
+      "step": 712
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.983295901179113e-05,
+      "loss": 0.0766,
+      "step": 714
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9832491109863375e-05,
+      "loss": 0.0643,
+      "step": 716
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.983202320793562e-05,
+      "loss": 0.0594,
+      "step": 718
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9831555306007867e-05,
+      "loss": 0.0705,
+      "step": 720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9831087404080106e-05,
+      "loss": 0.0614,
+      "step": 722
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.983061950215235e-05,
+      "loss": 0.0734,
+      "step": 724
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.98301516002246e-05,
+      "loss": 0.0642,
+      "step": 726
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9829683698296843e-05,
+      "loss": 0.0855,
+      "step": 728
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.982921579636908e-05,
+      "loss": 0.0674,
+      "step": 730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.982874789444133e-05,
+      "loss": 0.0818,
+      "step": 732
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9828279992513574e-05,
+      "loss": 0.0786,
+      "step": 734
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.982781209058582e-05,
+      "loss": 0.0543,
+      "step": 736
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.982734418865806e-05,
+      "loss": 0.0741,
+      "step": 738
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9826876286730305e-05,
+      "loss": 0.0657,
+      "step": 740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9826408384802545e-05,
+      "loss": 0.0634,
+      "step": 742
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.98259404828748e-05,
+      "loss": 0.0747,
+      "step": 744
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9825472580947036e-05,
+      "loss": 0.0765,
+      "step": 746
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.982500467901928e-05,
+      "loss": 0.0785,
+      "step": 748
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.982453677709152e-05,
+      "loss": 0.0719,
+      "step": 750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9824068875163774e-05,
+      "loss": 0.0752,
+      "step": 752
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.982360097323601e-05,
+      "loss": 0.0667,
+      "step": 754
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.982313307130826e-05,
+      "loss": 0.0795,
+      "step": 756
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.98226651693805e-05,
+      "loss": 0.0718,
+      "step": 758
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9822197267452744e-05,
+      "loss": 0.0618,
+      "step": 760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.982172936552499e-05,
+      "loss": 0.0672,
+      "step": 762
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.982126146359723e-05,
+      "loss": 0.0586,
+      "step": 764
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9820793561669475e-05,
+      "loss": 0.0782,
+      "step": 766
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.982032565974172e-05,
+      "loss": 0.086,
+      "step": 768
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.981985775781397e-05,
+      "loss": 0.0726,
+      "step": 770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9819389855886206e-05,
+      "loss": 0.0705,
+      "step": 772
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.981892195395845e-05,
+      "loss": 0.0829,
+      "step": 774
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.981845405203069e-05,
+      "loss": 0.0539,
+      "step": 776
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9817986150102944e-05,
+      "loss": 0.0653,
+      "step": 778
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.981751824817518e-05,
+      "loss": 0.0795,
+      "step": 780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.981705034624743e-05,
+      "loss": 0.0688,
+      "step": 782
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.981658244431967e-05,
+      "loss": 0.0733,
+      "step": 784
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.981611454239192e-05,
+      "loss": 0.079,
+      "step": 786
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.981564664046416e-05,
+      "loss": 0.0719,
+      "step": 788
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9815178738536406e-05,
+      "loss": 0.0719,
+      "step": 790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9814710836608645e-05,
+      "loss": 0.0795,
+      "step": 792
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.981424293468089e-05,
+      "loss": 0.0839,
+      "step": 794
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.981377503275314e-05,
+      "loss": 0.0543,
+      "step": 796
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.981330713082538e-05,
+      "loss": 0.0901,
+      "step": 798
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.981283922889762e-05,
+      "loss": 0.085,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.981237132696987e-05,
+      "loss": 0.0841,
+      "step": 802
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9811903425042114e-05,
+      "loss": 0.0873,
+      "step": 804
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.981143552311436e-05,
+      "loss": 0.0738,
+      "step": 806
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.98109676211866e-05,
+      "loss": 0.0911,
+      "step": 808
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9810499719258845e-05,
+      "loss": 0.0635,
+      "step": 810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.981003181733109e-05,
+      "loss": 0.0814,
+      "step": 812
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9809563915403336e-05,
+      "loss": 0.0875,
+      "step": 814
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9809096013475575e-05,
+      "loss": 0.0586,
+      "step": 816
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.980862811154782e-05,
+      "loss": 0.076,
+      "step": 818
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.980816020962006e-05,
+      "loss": 0.0733,
+      "step": 820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.980769230769231e-05,
+      "loss": 0.0748,
+      "step": 822
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.980722440576455e-05,
+      "loss": 0.0799,
+      "step": 824
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.98067565038368e-05,
+      "loss": 0.0687,
+      "step": 826
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.980628860190904e-05,
+      "loss": 0.0638,
+      "step": 828
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.980582069998129e-05,
+      "loss": 0.0611,
+      "step": 830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.980535279805353e-05,
+      "loss": 0.0895,
+      "step": 832
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9804884896125775e-05,
+      "loss": 0.0886,
+      "step": 834
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9804416994198014e-05,
+      "loss": 0.0741,
+      "step": 836
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.980394909227026e-05,
+      "loss": 0.0634,
+      "step": 838
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9803481190342506e-05,
+      "loss": 0.0722,
+      "step": 840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.980301328841475e-05,
+      "loss": 0.079,
+      "step": 842
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.980254538648699e-05,
+      "loss": 0.0603,
+      "step": 844
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.980207748455924e-05,
+      "loss": 0.0672,
+      "step": 846
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.980160958263148e-05,
+      "loss": 0.0855,
+      "step": 848
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.980114168070373e-05,
+      "loss": 0.0704,
+      "step": 850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.980067377877597e-05,
+      "loss": 0.0627,
+      "step": 852
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9800205876848214e-05,
+      "loss": 0.0737,
+      "step": 854
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.979973797492046e-05,
+      "loss": 0.064,
+      "step": 856
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9799270072992706e-05,
+      "loss": 0.0789,
+      "step": 858
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9798802171064945e-05,
+      "loss": 0.0821,
+      "step": 860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.979833426913719e-05,
+      "loss": 0.0795,
+      "step": 862
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.979786636720944e-05,
+      "loss": 0.0738,
+      "step": 864
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.979739846528168e-05,
+      "loss": 0.0792,
+      "step": 866
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.979693056335392e-05,
+      "loss": 0.0799,
+      "step": 868
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.979646266142617e-05,
+      "loss": 0.0891,
+      "step": 870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.979599475949841e-05,
+      "loss": 0.076,
+      "step": 872
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.979552685757066e-05,
+      "loss": 0.084,
+      "step": 874
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.97950589556429e-05,
+      "loss": 0.1022,
+      "step": 876
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9794591053715145e-05,
+      "loss": 0.0674,
+      "step": 878
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9794123151787384e-05,
+      "loss": 0.0637,
+      "step": 880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9793655249859636e-05,
+      "loss": 0.0675,
+      "step": 882
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9793187347931876e-05,
+      "loss": 0.0714,
+      "step": 884
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.979271944600412e-05,
+      "loss": 0.0624,
+      "step": 886
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.979225154407636e-05,
+      "loss": 0.0751,
+      "step": 888
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9791783642148606e-05,
+      "loss": 0.0722,
+      "step": 890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.979131574022085e-05,
+      "loss": 0.0787,
+      "step": 892
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.97908478382931e-05,
+      "loss": 0.0885,
+      "step": 894
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.979037993636534e-05,
+      "loss": 0.084,
+      "step": 896
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.978991203443758e-05,
+      "loss": 0.0701,
+      "step": 898
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.978944413250983e-05,
+      "loss": 0.0832,
+      "step": 900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9788976230582075e-05,
+      "loss": 0.0629,
+      "step": 902
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9788508328654314e-05,
+      "loss": 0.0877,
+      "step": 904
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.978804042672656e-05,
+      "loss": 0.0544,
+      "step": 906
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9787572524798806e-05,
+      "loss": 0.0992,
+      "step": 908
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.978710462287105e-05,
+      "loss": 0.0708,
+      "step": 910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.978663672094329e-05,
+      "loss": 0.1194,
+      "step": 912
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.978616881901554e-05,
+      "loss": 0.0667,
+      "step": 914
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.978570091708778e-05,
+      "loss": 0.0704,
+      "step": 916
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.978523301516003e-05,
+      "loss": 0.0848,
+      "step": 918
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.978476511323227e-05,
+      "loss": 0.0714,
+      "step": 920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9784297211304514e-05,
+      "loss": 0.0827,
+      "step": 922
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.978382930937675e-05,
+      "loss": 0.0753,
+      "step": 924
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9783361407449006e-05,
+      "loss": 0.0912,
+      "step": 926
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9782893505521245e-05,
+      "loss": 0.0688,
+      "step": 928
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.978242560359349e-05,
+      "loss": 0.0669,
+      "step": 930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.978195770166573e-05,
+      "loss": 0.0779,
+      "step": 932
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9781489799737976e-05,
+      "loss": 0.0745,
+      "step": 934
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.978102189781022e-05,
+      "loss": 0.0952,
+      "step": 936
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.978055399588247e-05,
+      "loss": 0.0925,
+      "step": 938
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.978008609395471e-05,
+      "loss": 0.0822,
+      "step": 940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.977961819202695e-05,
+      "loss": 0.0715,
+      "step": 942
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.97791502900992e-05,
+      "loss": 0.0817,
+      "step": 944
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9778682388171445e-05,
+      "loss": 0.0675,
+      "step": 946
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9778214486243684e-05,
+      "loss": 0.0612,
+      "step": 948
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.977774658431593e-05,
+      "loss": 0.0768,
+      "step": 950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9777278682388176e-05,
+      "loss": 0.1107,
+      "step": 952
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.977681078046042e-05,
+      "loss": 0.0712,
+      "step": 954
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.977634287853266e-05,
+      "loss": 0.0712,
+      "step": 956
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9775874976604906e-05,
+      "loss": 0.0705,
+      "step": 958
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.977540707467715e-05,
+      "loss": 0.0801,
+      "step": 960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.97749391727494e-05,
+      "loss": 0.0883,
+      "step": 962
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.977447127082164e-05,
+      "loss": 0.0704,
+      "step": 964
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.977400336889388e-05,
+      "loss": 0.0827,
+      "step": 966
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.977353546696612e-05,
+      "loss": 0.0675,
+      "step": 968
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9773067565038375e-05,
+      "loss": 0.0628,
+      "step": 970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9772599663110614e-05,
+      "loss": 0.0976,
+      "step": 972
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.977213176118286e-05,
+      "loss": 0.0806,
+      "step": 974
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.97716638592551e-05,
+      "loss": 0.0684,
+      "step": 976
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.977119595732735e-05,
+      "loss": 0.069,
+      "step": 978
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.977072805539959e-05,
+      "loss": 0.1089,
+      "step": 980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.977026015347184e-05,
+      "loss": 0.0738,
+      "step": 982
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9769792251544076e-05,
+      "loss": 0.0895,
+      "step": 984
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.976932434961632e-05,
+      "loss": 0.0637,
+      "step": 986
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.976885644768857e-05,
+      "loss": 0.0724,
+      "step": 988
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9768388545760814e-05,
+      "loss": 0.0606,
+      "step": 990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.976792064383305e-05,
+      "loss": 0.0679,
+      "step": 992
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.97674527419053e-05,
+      "loss": 0.071,
+      "step": 994
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9766984839977545e-05,
+      "loss": 0.0872,
+      "step": 996
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.976651693804979e-05,
+      "loss": 0.0745,
+      "step": 998
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.976604903612203e-05,
+      "loss": 0.0872,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "eval_gen_len": 30.8621,
+      "eval_loss": 1.0442978143692017,
+      "eval_meteor": 0.0398,
+      "eval_runtime": 16.8946,
+      "eval_samples_per_second": 3.433,
+      "eval_steps_per_second": 0.474,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9765581134194276e-05,
+      "loss": 0.0728,
+      "step": 1002
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.976511323226652e-05,
+      "loss": 0.0724,
+      "step": 1004
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.976464533033877e-05,
+      "loss": 0.0668,
+      "step": 1006
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.976417742841101e-05,
+      "loss": 0.0669,
+      "step": 1008
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.976370952648325e-05,
+      "loss": 0.0745,
+      "step": 1010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.97632416245555e-05,
+      "loss": 0.0742,
+      "step": 1012
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.976277372262774e-05,
+      "loss": 0.0789,
+      "step": 1014
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9762305820699984e-05,
+      "loss": 0.0911,
+      "step": 1016
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.976183791877222e-05,
+      "loss": 0.0868,
+      "step": 1018
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.976137001684447e-05,
+      "loss": 0.0685,
+      "step": 1020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9760902114916715e-05,
+      "loss": 0.0819,
+      "step": 1022
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.976043421298896e-05,
+      "loss": 0.0797,
+      "step": 1024
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.97599663110612e-05,
+      "loss": 0.0794,
+      "step": 1026
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9759498409133446e-05,
+      "loss": 0.0893,
+      "step": 1028
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.975903050720569e-05,
+      "loss": 0.1106,
+      "step": 1030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.975856260527794e-05,
+      "loss": 0.0689,
+      "step": 1032
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9758094703350177e-05,
+      "loss": 0.0797,
+      "step": 1034
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.975762680142242e-05,
+      "loss": 0.0948,
+      "step": 1036
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.975715889949467e-05,
+      "loss": 0.0694,
+      "step": 1038
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9756690997566914e-05,
+      "loss": 0.0811,
+      "step": 1040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9756223095639153e-05,
+      "loss": 0.0649,
+      "step": 1042
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.97557551937114e-05,
+      "loss": 0.0638,
+      "step": 1044
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9755287291783645e-05,
+      "loss": 0.0785,
+      "step": 1046
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.975481938985589e-05,
+      "loss": 0.0799,
+      "step": 1048
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.975435148792813e-05,
+      "loss": 0.0764,
+      "step": 1050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9753883586000376e-05,
+      "loss": 0.0614,
+      "step": 1052
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9753415684072615e-05,
+      "loss": 0.096,
+      "step": 1054
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.975294778214487e-05,
+      "loss": 0.0754,
+      "step": 1056
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.975247988021711e-05,
+      "loss": 0.0629,
+      "step": 1058
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.975201197828935e-05,
+      "loss": 0.0691,
+      "step": 1060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.975154407636159e-05,
+      "loss": 0.0723,
+      "step": 1062
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9751076174433845e-05,
+      "loss": 0.0716,
+      "step": 1064
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9750608272506084e-05,
+      "loss": 0.103,
+      "step": 1066
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.975014037057833e-05,
+      "loss": 0.0746,
+      "step": 1068
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.974967246865057e-05,
+      "loss": 0.0882,
+      "step": 1070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9749204566722815e-05,
+      "loss": 0.0734,
+      "step": 1072
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.974873666479506e-05,
+      "loss": 0.083,
+      "step": 1074
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.974826876286731e-05,
+      "loss": 0.0744,
+      "step": 1076
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9747800860939546e-05,
+      "loss": 0.0658,
+      "step": 1078
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.974733295901179e-05,
+      "loss": 0.0993,
+      "step": 1080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.974686505708404e-05,
+      "loss": 0.092,
+      "step": 1082
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9746397155156284e-05,
+      "loss": 0.0776,
+      "step": 1084
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.974592925322852e-05,
+      "loss": 0.091,
+      "step": 1086
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.974546135130077e-05,
+      "loss": 0.0613,
+      "step": 1088
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9744993449373015e-05,
+      "loss": 0.0721,
+      "step": 1090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.974452554744526e-05,
+      "loss": 0.0725,
+      "step": 1092
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.97440576455175e-05,
+      "loss": 0.0686,
+      "step": 1094
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9743589743589746e-05,
+      "loss": 0.0711,
+      "step": 1096
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.974312184166199e-05,
+      "loss": 0.0771,
+      "step": 1098
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.974265393973424e-05,
+      "loss": 0.0777,
+      "step": 1100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.974218603780648e-05,
+      "loss": 0.0696,
+      "step": 1102
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.974171813587872e-05,
+      "loss": 0.0804,
+      "step": 1104
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.974125023395096e-05,
+      "loss": 0.0923,
+      "step": 1106
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9740782332023214e-05,
+      "loss": 0.0609,
+      "step": 1108
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9740314430095453e-05,
+      "loss": 0.095,
+      "step": 1110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.97398465281677e-05,
+      "loss": 0.0916,
+      "step": 1112
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.973937862623994e-05,
+      "loss": 0.0593,
+      "step": 1114
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9738910724312184e-05,
+      "loss": 0.0868,
+      "step": 1116
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.973844282238443e-05,
+      "loss": 0.0645,
+      "step": 1118
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9737974920456676e-05,
+      "loss": 0.0774,
+      "step": 1120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9737507018528915e-05,
+      "loss": 0.0776,
+      "step": 1122
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.973703911660116e-05,
+      "loss": 0.0825,
+      "step": 1124
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.973657121467341e-05,
+      "loss": 0.0764,
+      "step": 1126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.973610331274565e-05,
+      "loss": 0.0721,
+      "step": 1128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.973563541081789e-05,
+      "loss": 0.0665,
+      "step": 1130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.973516750889014e-05,
+      "loss": 0.0799,
+      "step": 1132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9734699606962384e-05,
+      "loss": 0.0929,
+      "step": 1134
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.973423170503463e-05,
+      "loss": 0.0776,
+      "step": 1136
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.973376380310687e-05,
+      "loss": 0.0908,
+      "step": 1138
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9733295901179115e-05,
+      "loss": 0.0844,
+      "step": 1140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.973282799925136e-05,
+      "loss": 0.0699,
+      "step": 1142
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.973236009732361e-05,
+      "loss": 0.0734,
+      "step": 1144
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9731892195395846e-05,
+      "loss": 0.0761,
+      "step": 1146
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.973142429346809e-05,
+      "loss": 0.0918,
+      "step": 1148
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.973095639154033e-05,
+      "loss": 0.0646,
+      "step": 1150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9730488489612584e-05,
+      "loss": 0.0588,
+      "step": 1152
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.973002058768482e-05,
+      "loss": 0.113,
+      "step": 1154
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.972955268575707e-05,
+      "loss": 0.0782,
+      "step": 1156
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.972908478382931e-05,
+      "loss": 0.0707,
+      "step": 1158
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.972861688190156e-05,
+      "loss": 0.0718,
+      "step": 1160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.97281489799738e-05,
+      "loss": 0.0861,
+      "step": 1162
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9727681078046046e-05,
+      "loss": 0.0731,
+      "step": 1164
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9727213176118285e-05,
+      "loss": 0.0554,
+      "step": 1166
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.972674527419053e-05,
+      "loss": 0.0887,
+      "step": 1168
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.972627737226278e-05,
+      "loss": 0.07,
+      "step": 1170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.972580947033502e-05,
+      "loss": 0.0858,
+      "step": 1172
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.972534156840726e-05,
+      "loss": 0.0911,
+      "step": 1174
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.972487366647951e-05,
+      "loss": 0.1087,
+      "step": 1176
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9724405764551754e-05,
+      "loss": 0.0649,
+      "step": 1178
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9723937862624e-05,
+      "loss": 0.074,
+      "step": 1180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.972346996069624e-05,
+      "loss": 0.0718,
+      "step": 1182
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9723002058768484e-05,
+      "loss": 0.0853,
+      "step": 1184
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.972253415684073e-05,
+      "loss": 0.0804,
+      "step": 1186
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9722066254912976e-05,
+      "loss": 0.0734,
+      "step": 1188
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9721598352985215e-05,
+      "loss": 0.0886,
+      "step": 1190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.972113045105746e-05,
+      "loss": 0.0633,
+      "step": 1192
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.972066254912971e-05,
+      "loss": 0.0832,
+      "step": 1194
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.972019464720195e-05,
+      "loss": 0.064,
+      "step": 1196
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.971972674527419e-05,
+      "loss": 0.093,
+      "step": 1198
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.971925884334644e-05,
+      "loss": 0.0718,
+      "step": 1200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.971879094141868e-05,
+      "loss": 0.0691,
+      "step": 1202
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.971832303949093e-05,
+      "loss": 0.071,
+      "step": 1204
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.971785513756317e-05,
+      "loss": 0.0604,
+      "step": 1206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9717387235635415e-05,
+      "loss": 0.0821,
+      "step": 1208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9716919333707654e-05,
+      "loss": 0.0756,
+      "step": 1210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.97164514317799e-05,
+      "loss": 0.0643,
+      "step": 1212
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9715983529852146e-05,
+      "loss": 0.1184,
+      "step": 1214
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.971551562792439e-05,
+      "loss": 0.0718,
+      "step": 1216
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.971504772599663e-05,
+      "loss": 0.0878,
+      "step": 1218
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.971457982406888e-05,
+      "loss": 0.0799,
+      "step": 1220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.971411192214112e-05,
+      "loss": 0.0843,
+      "step": 1222
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.971364402021337e-05,
+      "loss": 0.085,
+      "step": 1224
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.971317611828561e-05,
+      "loss": 0.0735,
+      "step": 1226
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9712708216357854e-05,
+      "loss": 0.0596,
+      "step": 1228
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.97122403144301e-05,
+      "loss": 0.0913,
+      "step": 1230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9711772412502346e-05,
+      "loss": 0.0649,
+      "step": 1232
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9711304510574585e-05,
+      "loss": 0.0815,
+      "step": 1234
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.971083660864683e-05,
+      "loss": 0.0671,
+      "step": 1236
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.971036870671908e-05,
+      "loss": 0.0909,
+      "step": 1238
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.970990080479132e-05,
+      "loss": 0.0842,
+      "step": 1240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.970943290286356e-05,
+      "loss": 0.0687,
+      "step": 1242
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.970896500093581e-05,
+      "loss": 0.0708,
+      "step": 1244
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.970849709900805e-05,
+      "loss": 0.0764,
+      "step": 1246
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.97080291970803e-05,
+      "loss": 0.097,
+      "step": 1248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.970756129515254e-05,
+      "loss": 0.0811,
+      "step": 1250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9707093393224784e-05,
+      "loss": 0.0683,
+      "step": 1252
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9706625491297024e-05,
+      "loss": 0.0878,
+      "step": 1254
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9706157589369276e-05,
+      "loss": 0.0864,
+      "step": 1256
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9705689687441515e-05,
+      "loss": 0.0921,
+      "step": 1258
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.970522178551376e-05,
+      "loss": 0.0966,
+      "step": 1260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9704753883586e-05,
+      "loss": 0.081,
+      "step": 1262
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9704285981658246e-05,
+      "loss": 0.1004,
+      "step": 1264
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.970381807973049e-05,
+      "loss": 0.0775,
+      "step": 1266
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.970335017780273e-05,
+      "loss": 0.0964,
+      "step": 1268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.970288227587498e-05,
+      "loss": 0.0648,
+      "step": 1270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.970241437394722e-05,
+      "loss": 0.0906,
+      "step": 1272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.970194647201947e-05,
+      "loss": 0.0682,
+      "step": 1274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.970147857009171e-05,
+      "loss": 0.0713,
+      "step": 1276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9701010668163954e-05,
+      "loss": 0.0702,
+      "step": 1278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.970054276623619e-05,
+      "loss": 0.0809,
+      "step": 1280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9700074864308446e-05,
+      "loss": 0.0708,
+      "step": 1282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9699606962380685e-05,
+      "loss": 0.0661,
+      "step": 1284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.969913906045293e-05,
+      "loss": 0.0763,
+      "step": 1286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.969867115852517e-05,
+      "loss": 0.0946,
+      "step": 1288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.969820325659742e-05,
+      "loss": 0.0696,
+      "step": 1290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.969773535466966e-05,
+      "loss": 0.1217,
+      "step": 1292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.969726745274191e-05,
+      "loss": 0.0738,
+      "step": 1294
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.969679955081415e-05,
+      "loss": 0.0721,
+      "step": 1296
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.969633164888639e-05,
+      "loss": 0.0921,
+      "step": 1298
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.969586374695864e-05,
+      "loss": 0.0764,
+      "step": 1300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9695395845030885e-05,
+      "loss": 0.0703,
+      "step": 1302
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9694927943103124e-05,
+      "loss": 0.0706,
+      "step": 1304
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.969446004117537e-05,
+      "loss": 0.0656,
+      "step": 1306
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9693992139247616e-05,
+      "loss": 0.0598,
+      "step": 1308
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.969352423731986e-05,
+      "loss": 0.1019,
+      "step": 1310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.96930563353921e-05,
+      "loss": 0.0666,
+      "step": 1312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.969258843346435e-05,
+      "loss": 0.0836,
+      "step": 1314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.969212053153659e-05,
+      "loss": 0.1002,
+      "step": 1316
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.969165262960884e-05,
+      "loss": 0.0828,
+      "step": 1318
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.969118472768108e-05,
+      "loss": 0.0883,
+      "step": 1320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9690716825753324e-05,
+      "loss": 0.0744,
+      "step": 1322
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.969024892382557e-05,
+      "loss": 0.0969,
+      "step": 1324
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9689781021897815e-05,
+      "loss": 0.0907,
+      "step": 1326
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9689313119970055e-05,
+      "loss": 0.0732,
+      "step": 1328
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.96888452180423e-05,
+      "loss": 0.0717,
+      "step": 1330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.968837731611454e-05,
+      "loss": 0.073,
+      "step": 1332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.968790941418679e-05,
+      "loss": 0.0989,
+      "step": 1334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.968744151225903e-05,
+      "loss": 0.0718,
+      "step": 1336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.968697361033128e-05,
+      "loss": 0.0672,
+      "step": 1338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9686505708403517e-05,
+      "loss": 0.0547,
+      "step": 1340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.968603780647577e-05,
+      "loss": 0.0972,
+      "step": 1342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.968556990454801e-05,
+      "loss": 0.0707,
+      "step": 1344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9685102002620254e-05,
+      "loss": 0.0644,
+      "step": 1346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.968463410069249e-05,
+      "loss": 0.0886,
+      "step": 1348
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.968416619876474e-05,
+      "loss": 0.1001,
+      "step": 1350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9683698296836985e-05,
+      "loss": 0.0992,
+      "step": 1352
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.968323039490923e-05,
+      "loss": 0.0732,
+      "step": 1354
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.968276249298147e-05,
+      "loss": 0.086,
+      "step": 1356
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9682294591053716e-05,
+      "loss": 0.08,
+      "step": 1358
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.968182668912596e-05,
+      "loss": 0.1046,
+      "step": 1360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.968135878719821e-05,
+      "loss": 0.0528,
+      "step": 1362
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.968089088527045e-05,
+      "loss": 0.0786,
+      "step": 1364
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.968042298334269e-05,
+      "loss": 0.0762,
+      "step": 1366
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.967995508141494e-05,
+      "loss": 0.098,
+      "step": 1368
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9679487179487185e-05,
+      "loss": 0.0655,
+      "step": 1370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9679019277559424e-05,
+      "loss": 0.079,
+      "step": 1372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.967855137563167e-05,
+      "loss": 0.0746,
+      "step": 1374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9678083473703916e-05,
+      "loss": 0.0864,
+      "step": 1376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.967761557177616e-05,
+      "loss": 0.0659,
+      "step": 1378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.96771476698484e-05,
+      "loss": 0.0851,
+      "step": 1380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.967667976792065e-05,
+      "loss": 0.0959,
+      "step": 1382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9676211865992886e-05,
+      "loss": 0.103,
+      "step": 1384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.967574396406514e-05,
+      "loss": 0.0967,
+      "step": 1386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.967527606213738e-05,
+      "loss": 0.0623,
+      "step": 1388
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9674808160209624e-05,
+      "loss": 0.0926,
+      "step": 1390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.967434025828186e-05,
+      "loss": 0.09,
+      "step": 1392
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.967387235635411e-05,
+      "loss": 0.0794,
+      "step": 1394
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9673404454426355e-05,
+      "loss": 0.096,
+      "step": 1396
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.96729365524986e-05,
+      "loss": 0.0793,
+      "step": 1398
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.967246865057084e-05,
+      "loss": 0.0692,
+      "step": 1400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9672000748643086e-05,
+      "loss": 0.0781,
+      "step": 1402
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.967153284671533e-05,
+      "loss": 0.0773,
+      "step": 1404
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.967106494478758e-05,
+      "loss": 0.0641,
+      "step": 1406
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9670597042859817e-05,
+      "loss": 0.0957,
+      "step": 1408
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.967012914093206e-05,
+      "loss": 0.0772,
+      "step": 1410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.966966123900431e-05,
+      "loss": 0.0795,
+      "step": 1412
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9669193337076554e-05,
+      "loss": 0.0783,
+      "step": 1414
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9668725435148793e-05,
+      "loss": 0.0703,
+      "step": 1416
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.966825753322104e-05,
+      "loss": 0.0983,
+      "step": 1418
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9667789631293285e-05,
+      "loss": 0.1161,
+      "step": 1420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.966732172936553e-05,
+      "loss": 0.0694,
+      "step": 1422
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.966685382743777e-05,
+      "loss": 0.0845,
+      "step": 1424
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9666385925510016e-05,
+      "loss": 0.0769,
+      "step": 1426
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9665918023582255e-05,
+      "loss": 0.0982,
+      "step": 1428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.966545012165451e-05,
+      "loss": 0.08,
+      "step": 1430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.966498221972675e-05,
+      "loss": 0.0748,
+      "step": 1432
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.966451431779899e-05,
+      "loss": 0.1047,
+      "step": 1434
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.966404641587123e-05,
+      "loss": 0.0887,
+      "step": 1436
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9663578513943485e-05,
+      "loss": 0.0544,
+      "step": 1438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9663110612015724e-05,
+      "loss": 0.068,
+      "step": 1440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.966264271008797e-05,
+      "loss": 0.0825,
+      "step": 1442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.966217480816021e-05,
+      "loss": 0.0847,
+      "step": 1444
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9661706906232455e-05,
+      "loss": 0.0638,
+      "step": 1446
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.96612390043047e-05,
+      "loss": 0.0694,
+      "step": 1448
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.966077110237695e-05,
+      "loss": 0.1114,
+      "step": 1450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9660303200449186e-05,
+      "loss": 0.084,
+      "step": 1452
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.965983529852143e-05,
+      "loss": 0.0949,
+      "step": 1454
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.965936739659368e-05,
+      "loss": 0.0833,
+      "step": 1456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9658899494665924e-05,
+      "loss": 0.0765,
+      "step": 1458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.965843159273816e-05,
+      "loss": 0.1126,
+      "step": 1460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.965796369081041e-05,
+      "loss": 0.0751,
+      "step": 1462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9657495788882655e-05,
+      "loss": 0.0913,
+      "step": 1464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.96570278869549e-05,
+      "loss": 0.0793,
+      "step": 1466
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.965655998502714e-05,
+      "loss": 0.0663,
+      "step": 1468
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9656092083099386e-05,
+      "loss": 0.0732,
+      "step": 1470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.965562418117163e-05,
+      "loss": 0.0885,
+      "step": 1472
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.965515627924388e-05,
+      "loss": 0.0798,
+      "step": 1474
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9654688377316117e-05,
+      "loss": 0.083,
+      "step": 1476
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.965422047538836e-05,
+      "loss": 0.0648,
+      "step": 1478
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.96537525734606e-05,
+      "loss": 0.0861,
+      "step": 1480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9653284671532854e-05,
+      "loss": 0.0934,
+      "step": 1482
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9652816769605093e-05,
+      "loss": 0.0993,
+      "step": 1484
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.965234886767734e-05,
+      "loss": 0.0917,
+      "step": 1486
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.965188096574958e-05,
+      "loss": 0.0759,
+      "step": 1488
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.965141306382183e-05,
+      "loss": 0.0847,
+      "step": 1490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.965094516189407e-05,
+      "loss": 0.0854,
+      "step": 1492
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9650477259966316e-05,
+      "loss": 0.0781,
+      "step": 1494
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9650009358038555e-05,
+      "loss": 0.089,
+      "step": 1496
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.96495414561108e-05,
+      "loss": 0.0684,
+      "step": 1498
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.964907355418305e-05,
+      "loss": 0.0786,
+      "step": 1500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.964860565225529e-05,
+      "loss": 0.089,
+      "step": 1502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.964813775032753e-05,
+      "loss": 0.0815,
+      "step": 1504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.964766984839978e-05,
+      "loss": 0.067,
+      "step": 1506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9647201946472024e-05,
+      "loss": 0.0887,
+      "step": 1508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.964673404454427e-05,
+      "loss": 0.0788,
+      "step": 1510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.964626614261651e-05,
+      "loss": 0.062,
+      "step": 1512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9645798240688755e-05,
+      "loss": 0.0913,
+      "step": 1514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9645330338761e-05,
+      "loss": 0.0805,
+      "step": 1516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.964486243683325e-05,
+      "loss": 0.0733,
+      "step": 1518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9644394534905486e-05,
+      "loss": 0.0832,
+      "step": 1520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9643926632977725e-05,
+      "loss": 0.0691,
+      "step": 1522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.964345873104997e-05,
+      "loss": 0.0676,
+      "step": 1524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.964299082912222e-05,
+      "loss": 0.069,
+      "step": 1526
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.964252292719446e-05,
+      "loss": 0.0831,
+      "step": 1528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.96420550252667e-05,
+      "loss": 0.0967,
+      "step": 1530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.964158712333895e-05,
+      "loss": 0.0859,
+      "step": 1532
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9641119221411194e-05,
+      "loss": 0.0698,
+      "step": 1534
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.964065131948344e-05,
+      "loss": 0.0774,
+      "step": 1536
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.964018341755568e-05,
+      "loss": 0.0931,
+      "step": 1538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9639715515627925e-05,
+      "loss": 0.0637,
+      "step": 1540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.963924761370017e-05,
+      "loss": 0.0944,
+      "step": 1542
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9638779711772417e-05,
+      "loss": 0.0829,
+      "step": 1544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9638311809844656e-05,
+      "loss": 0.0705,
+      "step": 1546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.96378439079169e-05,
+      "loss": 0.0874,
+      "step": 1548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.963737600598915e-05,
+      "loss": 0.0955,
+      "step": 1550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9636908104061393e-05,
+      "loss": 0.0767,
+      "step": 1552
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.963644020213363e-05,
+      "loss": 0.0824,
+      "step": 1554
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.963597230020588e-05,
+      "loss": 0.0621,
+      "step": 1556
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.963550439827812e-05,
+      "loss": 0.0668,
+      "step": 1558
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.963503649635037e-05,
+      "loss": 0.0944,
+      "step": 1560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.963456859442261e-05,
+      "loss": 0.0821,
+      "step": 1562
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9634100692494855e-05,
+      "loss": 0.0802,
+      "step": 1564
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9633632790567094e-05,
+      "loss": 0.0609,
+      "step": 1566
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.963316488863935e-05,
+      "loss": 0.0801,
+      "step": 1568
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9632696986711586e-05,
+      "loss": 0.0837,
+      "step": 1570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.963222908478383e-05,
+      "loss": 0.0898,
+      "step": 1572
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.963176118285607e-05,
+      "loss": 0.0818,
+      "step": 1574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.963129328092832e-05,
+      "loss": 0.0602,
+      "step": 1576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.963082537900056e-05,
+      "loss": 0.0868,
+      "step": 1578
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.963035747707281e-05,
+      "loss": 0.077,
+      "step": 1580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.962988957514505e-05,
+      "loss": 0.1011,
+      "step": 1582
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9629421673217294e-05,
+      "loss": 0.0658,
+      "step": 1584
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.962895377128954e-05,
+      "loss": 0.0791,
+      "step": 1586
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9628485869361786e-05,
+      "loss": 0.0889,
+      "step": 1588
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9628017967434025e-05,
+      "loss": 0.0783,
+      "step": 1590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.962755006550627e-05,
+      "loss": 0.0944,
+      "step": 1592
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.962708216357852e-05,
+      "loss": 0.0814,
+      "step": 1594
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.962661426165076e-05,
+      "loss": 0.1023,
+      "step": 1596
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9626146359723e-05,
+      "loss": 0.0762,
+      "step": 1598
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.962567845779525e-05,
+      "loss": 0.0939,
+      "step": 1600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9625210555867494e-05,
+      "loss": 0.0706,
+      "step": 1602
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.962474265393974e-05,
+      "loss": 0.0997,
+      "step": 1604
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.962427475201198e-05,
+      "loss": 0.0837,
+      "step": 1606
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9623806850084225e-05,
+      "loss": 0.0888,
+      "step": 1608
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9623338948156464e-05,
+      "loss": 0.1059,
+      "step": 1610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9622871046228717e-05,
+      "loss": 0.0897,
+      "step": 1612
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9622403144300956e-05,
+      "loss": 0.0891,
+      "step": 1614
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.96219352423732e-05,
+      "loss": 0.0714,
+      "step": 1616
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.962146734044544e-05,
+      "loss": 0.1006,
+      "step": 1618
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9620999438517693e-05,
+      "loss": 0.0806,
+      "step": 1620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.962053153658993e-05,
+      "loss": 0.0737,
+      "step": 1622
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.962006363466218e-05,
+      "loss": 0.0902,
+      "step": 1624
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.961959573273442e-05,
+      "loss": 0.0863,
+      "step": 1626
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9619127830806664e-05,
+      "loss": 0.0983,
+      "step": 1628
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.961865992887891e-05,
+      "loss": 0.083,
+      "step": 1630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9618192026951155e-05,
+      "loss": 0.0634,
+      "step": 1632
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9617724125023395e-05,
+      "loss": 0.0901,
+      "step": 1634
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.961725622309564e-05,
+      "loss": 0.0803,
+      "step": 1636
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9616788321167886e-05,
+      "loss": 0.083,
+      "step": 1638
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.961632041924013e-05,
+      "loss": 0.0967,
+      "step": 1640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.961585251731237e-05,
+      "loss": 0.0762,
+      "step": 1642
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.961538461538462e-05,
+      "loss": 0.0673,
+      "step": 1644
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.961491671345686e-05,
+      "loss": 0.0974,
+      "step": 1646
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.961444881152911e-05,
+      "loss": 0.0882,
+      "step": 1648
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.961398090960135e-05,
+      "loss": 0.0851,
+      "step": 1650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9613513007673594e-05,
+      "loss": 0.0705,
+      "step": 1652
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.961304510574584e-05,
+      "loss": 0.0681,
+      "step": 1654
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9612577203818086e-05,
+      "loss": 0.0806,
+      "step": 1656
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9612109301890325e-05,
+      "loss": 0.085,
+      "step": 1658
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.961164139996257e-05,
+      "loss": 0.0866,
+      "step": 1660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.961117349803481e-05,
+      "loss": 0.07,
+      "step": 1662
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.961070559610706e-05,
+      "loss": 0.0758,
+      "step": 1664
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.96102376941793e-05,
+      "loss": 0.0642,
+      "step": 1666
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.960976979225155e-05,
+      "loss": 0.0705,
+      "step": 1668
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.960930189032379e-05,
+      "loss": 0.0739,
+      "step": 1670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.960883398839603e-05,
+      "loss": 0.1064,
+      "step": 1672
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.960836608646828e-05,
+      "loss": 0.0583,
+      "step": 1674
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9607898184540525e-05,
+      "loss": 0.0852,
+      "step": 1676
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9607430282612764e-05,
+      "loss": 0.0754,
+      "step": 1678
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.960696238068501e-05,
+      "loss": 0.0618,
+      "step": 1680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9606494478757256e-05,
+      "loss": 0.0784,
+      "step": 1682
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.96060265768295e-05,
+      "loss": 0.0766,
+      "step": 1684
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.960555867490174e-05,
+      "loss": 0.0703,
+      "step": 1686
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.960509077297399e-05,
+      "loss": 0.0819,
+      "step": 1688
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.960462287104623e-05,
+      "loss": 0.0892,
+      "step": 1690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.960415496911848e-05,
+      "loss": 0.0824,
+      "step": 1692
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.960368706719072e-05,
+      "loss": 0.0842,
+      "step": 1694
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9603219165262964e-05,
+      "loss": 0.0952,
+      "step": 1696
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.960275126333521e-05,
+      "loss": 0.0705,
+      "step": 1698
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9602283361407455e-05,
+      "loss": 0.0768,
+      "step": 1700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9601815459479695e-05,
+      "loss": 0.0775,
+      "step": 1702
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.960134755755194e-05,
+      "loss": 0.0928,
+      "step": 1704
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.960087965562418e-05,
+      "loss": 0.0663,
+      "step": 1706
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.960041175369643e-05,
+      "loss": 0.075,
+      "step": 1708
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.959994385176867e-05,
+      "loss": 0.1009,
+      "step": 1710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.959947594984092e-05,
+      "loss": 0.0737,
+      "step": 1712
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9599008047913156e-05,
+      "loss": 0.0872,
+      "step": 1714
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.959854014598541e-05,
+      "loss": 0.0707,
+      "step": 1716
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.959807224405765e-05,
+      "loss": 0.0861,
+      "step": 1718
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9597604342129894e-05,
+      "loss": 0.0675,
+      "step": 1720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.959713644020213e-05,
+      "loss": 0.0742,
+      "step": 1722
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.959666853827438e-05,
+      "loss": 0.0651,
+      "step": 1724
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9596200636346625e-05,
+      "loss": 0.0899,
+      "step": 1726
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.959573273441887e-05,
+      "loss": 0.0718,
+      "step": 1728
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.959526483249111e-05,
+      "loss": 0.0892,
+      "step": 1730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9594796930563356e-05,
+      "loss": 0.072,
+      "step": 1732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.95943290286356e-05,
+      "loss": 0.0897,
+      "step": 1734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.959386112670785e-05,
+      "loss": 0.0705,
+      "step": 1736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.959339322478009e-05,
+      "loss": 0.0808,
+      "step": 1738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.959292532285233e-05,
+      "loss": 0.076,
+      "step": 1740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.959245742092458e-05,
+      "loss": 0.0787,
+      "step": 1742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9591989518996825e-05,
+      "loss": 0.1035,
+      "step": 1744
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9591521617069064e-05,
+      "loss": 0.0927,
+      "step": 1746
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.959105371514131e-05,
+      "loss": 0.0718,
+      "step": 1748
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9590585813213556e-05,
+      "loss": 0.0716,
+      "step": 1750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.95901179112858e-05,
+      "loss": 0.0739,
+      "step": 1752
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.958965000935804e-05,
+      "loss": 0.0586,
+      "step": 1754
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.958918210743029e-05,
+      "loss": 0.0888,
+      "step": 1756
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9588714205502526e-05,
+      "loss": 0.0948,
+      "step": 1758
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.958824630357478e-05,
+      "loss": 0.0726,
+      "step": 1760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.958777840164702e-05,
+      "loss": 0.0675,
+      "step": 1762
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9587310499719264e-05,
+      "loss": 0.0904,
+      "step": 1764
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.95868425977915e-05,
+      "loss": 0.0768,
+      "step": 1766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9586374695863755e-05,
+      "loss": 0.09,
+      "step": 1768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9585906793935995e-05,
+      "loss": 0.1021,
+      "step": 1770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.958543889200824e-05,
+      "loss": 0.0889,
+      "step": 1772
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.958497099008048e-05,
+      "loss": 0.0765,
+      "step": 1774
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9584503088152726e-05,
+      "loss": 0.0895,
+      "step": 1776
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.958403518622497e-05,
+      "loss": 0.0798,
+      "step": 1778
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.958356728429721e-05,
+      "loss": 0.1208,
+      "step": 1780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9583099382369456e-05,
+      "loss": 0.0793,
+      "step": 1782
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.95826314804417e-05,
+      "loss": 0.0725,
+      "step": 1784
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.958216357851395e-05,
+      "loss": 0.0829,
+      "step": 1786
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.958169567658619e-05,
+      "loss": 0.069,
+      "step": 1788
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.958122777465843e-05,
+      "loss": 0.069,
+      "step": 1790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.958075987273067e-05,
+      "loss": 0.0839,
+      "step": 1792
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9580291970802925e-05,
+      "loss": 0.0852,
+      "step": 1794
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9579824068875164e-05,
+      "loss": 0.1122,
+      "step": 1796
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.957935616694741e-05,
+      "loss": 0.1048,
+      "step": 1798
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.957888826501965e-05,
+      "loss": 0.0711,
+      "step": 1800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9578420363091895e-05,
+      "loss": 0.073,
+      "step": 1802
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.957795246116414e-05,
+      "loss": 0.1139,
+      "step": 1804
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.957748455923639e-05,
+      "loss": 0.0678,
+      "step": 1806
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9577016657308626e-05,
+      "loss": 0.0936,
+      "step": 1808
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.957654875538087e-05,
+      "loss": 0.0722,
+      "step": 1810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.957608085345312e-05,
+      "loss": 0.0779,
+      "step": 1812
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9575612951525364e-05,
+      "loss": 0.0842,
+      "step": 1814
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.95751450495976e-05,
+      "loss": 0.1146,
+      "step": 1816
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.957467714766985e-05,
+      "loss": 0.0801,
+      "step": 1818
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9574209245742095e-05,
+      "loss": 0.0851,
+      "step": 1820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.957374134381434e-05,
+      "loss": 0.0865,
+      "step": 1822
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.957327344188658e-05,
+      "loss": 0.0789,
+      "step": 1824
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9572805539958826e-05,
+      "loss": 0.0866,
+      "step": 1826
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.957233763803107e-05,
+      "loss": 0.0828,
+      "step": 1828
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.957186973610332e-05,
+      "loss": 0.1413,
+      "step": 1830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.957140183417556e-05,
+      "loss": 0.0734,
+      "step": 1832
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.95709339322478e-05,
+      "loss": 0.0669,
+      "step": 1834
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.957046603032004e-05,
+      "loss": 0.0824,
+      "step": 1836
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9569998128392295e-05,
+      "loss": 0.0812,
+      "step": 1838
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9569530226464534e-05,
+      "loss": 0.0805,
+      "step": 1840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.956906232453678e-05,
+      "loss": 0.1054,
+      "step": 1842
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.956859442260902e-05,
+      "loss": 0.0721,
+      "step": 1844
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.956812652068127e-05,
+      "loss": 0.0824,
+      "step": 1846
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.956765861875351e-05,
+      "loss": 0.0905,
+      "step": 1848
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9567190716825756e-05,
+      "loss": 0.0787,
+      "step": 1850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9566722814897996e-05,
+      "loss": 0.0864,
+      "step": 1852
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.956625491297024e-05,
+      "loss": 0.0656,
+      "step": 1854
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.956578701104249e-05,
+      "loss": 0.1018,
+      "step": 1856
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.956531910911473e-05,
+      "loss": 0.0841,
+      "step": 1858
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.956485120718697e-05,
+      "loss": 0.0772,
+      "step": 1860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.956438330525922e-05,
+      "loss": 0.0838,
+      "step": 1862
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9563915403331464e-05,
+      "loss": 0.0848,
+      "step": 1864
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.956344750140371e-05,
+      "loss": 0.0752,
+      "step": 1866
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.956297959947595e-05,
+      "loss": 0.0683,
+      "step": 1868
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9562511697548195e-05,
+      "loss": 0.0707,
+      "step": 1870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.956204379562044e-05,
+      "loss": 0.0966,
+      "step": 1872
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.956157589369269e-05,
+      "loss": 0.1103,
+      "step": 1874
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9561107991764926e-05,
+      "loss": 0.0819,
+      "step": 1876
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.956064008983717e-05,
+      "loss": 0.0815,
+      "step": 1878
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.956017218790942e-05,
+      "loss": 0.087,
+      "step": 1880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9559704285981664e-05,
+      "loss": 0.0682,
+      "step": 1882
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.95592363840539e-05,
+      "loss": 0.0894,
+      "step": 1884
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.955876848212615e-05,
+      "loss": 0.0693,
+      "step": 1886
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.955830058019839e-05,
+      "loss": 0.0699,
+      "step": 1888
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.955783267827064e-05,
+      "loss": 0.0955,
+      "step": 1890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.955736477634288e-05,
+      "loss": 0.0724,
+      "step": 1892
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9556896874415126e-05,
+      "loss": 0.0868,
+      "step": 1894
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9556428972487365e-05,
+      "loss": 0.079,
+      "step": 1896
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.955596107055962e-05,
+      "loss": 0.0919,
+      "step": 1898
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.955549316863186e-05,
+      "loss": 0.0741,
+      "step": 1900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.95550252667041e-05,
+      "loss": 0.0702,
+      "step": 1902
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.955455736477634e-05,
+      "loss": 0.0649,
+      "step": 1904
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.955408946284859e-05,
+      "loss": 0.1234,
+      "step": 1906
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9553621560920834e-05,
+      "loss": 0.0759,
+      "step": 1908
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.955315365899308e-05,
+      "loss": 0.0899,
+      "step": 1910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.955268575706532e-05,
+      "loss": 0.0933,
+      "step": 1912
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9552217855137565e-05,
+      "loss": 0.1276,
+      "step": 1914
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.955174995320981e-05,
+      "loss": 0.0794,
+      "step": 1916
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9551282051282056e-05,
+      "loss": 0.1122,
+      "step": 1918
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9550814149354296e-05,
+      "loss": 0.1009,
+      "step": 1920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.955034624742654e-05,
+      "loss": 0.0805,
+      "step": 1922
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.954987834549879e-05,
+      "loss": 0.0824,
+      "step": 1924
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.954941044357103e-05,
+      "loss": 0.0767,
+      "step": 1926
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.954894254164327e-05,
+      "loss": 0.0992,
+      "step": 1928
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.954847463971552e-05,
+      "loss": 0.0582,
+      "step": 1930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9548006737787764e-05,
+      "loss": 0.0696,
+      "step": 1932
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.954753883586001e-05,
+      "loss": 0.1047,
+      "step": 1934
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.954707093393225e-05,
+      "loss": 0.0732,
+      "step": 1936
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9546603032004495e-05,
+      "loss": 0.0788,
+      "step": 1938
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9546135130076734e-05,
+      "loss": 0.0817,
+      "step": 1940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.954566722814899e-05,
+      "loss": 0.0883,
+      "step": 1942
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9545199326221226e-05,
+      "loss": 0.0814,
+      "step": 1944
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.954473142429347e-05,
+      "loss": 0.0878,
+      "step": 1946
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.954426352236571e-05,
+      "loss": 0.0919,
+      "step": 1948
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.954379562043796e-05,
+      "loss": 0.0728,
+      "step": 1950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.95433277185102e-05,
+      "loss": 0.0961,
+      "step": 1952
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.954285981658245e-05,
+      "loss": 0.0727,
+      "step": 1954
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.954239191465469e-05,
+      "loss": 0.0969,
+      "step": 1956
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9541924012726934e-05,
+      "loss": 0.0743,
+      "step": 1958
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.954145611079918e-05,
+      "loss": 0.0761,
+      "step": 1960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9540988208871426e-05,
+      "loss": 0.0691,
+      "step": 1962
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9540520306943665e-05,
+      "loss": 0.0725,
+      "step": 1964
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.954005240501591e-05,
+      "loss": 0.1018,
+      "step": 1966
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.953958450308816e-05,
+      "loss": 0.0669,
+      "step": 1968
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.95391166011604e-05,
+      "loss": 0.0777,
+      "step": 1970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.953864869923264e-05,
+      "loss": 0.1015,
+      "step": 1972
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.953818079730489e-05,
+      "loss": 0.099,
+      "step": 1974
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9537712895377134e-05,
+      "loss": 0.0754,
+      "step": 1976
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.953724499344938e-05,
+      "loss": 0.0703,
+      "step": 1978
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.953677709152162e-05,
+      "loss": 0.0935,
+      "step": 1980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9536309189593865e-05,
+      "loss": 0.0701,
+      "step": 1982
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9535841287666104e-05,
+      "loss": 0.1014,
+      "step": 1984
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9535373385738357e-05,
+      "loss": 0.0649,
+      "step": 1986
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9534905483810596e-05,
+      "loss": 0.0947,
+      "step": 1988
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.953443758188284e-05,
+      "loss": 0.0857,
+      "step": 1990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.953396967995508e-05,
+      "loss": 0.0781,
+      "step": 1992
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.953350177802733e-05,
+      "loss": 0.0849,
+      "step": 1994
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.953303387609957e-05,
+      "loss": 0.0767,
+      "step": 1996
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.953256597417182e-05,
+      "loss": 0.1001,
+      "step": 1998
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.953209807224406e-05,
+      "loss": 0.09,
+      "step": 2000
+    },
+    {
+      "epoch": 0.28,
+      "eval_gen_len": 27.0862,
+      "eval_loss": 1.035199761390686,
+      "eval_meteor": 0.0382,
+      "eval_runtime": 13.218,
+      "eval_samples_per_second": 4.388,
+      "eval_steps_per_second": 0.605,
+      "step": 2000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9531630170316303e-05,
+      "loss": 0.1195,
+      "step": 2002
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.953116226838855e-05,
+      "loss": 0.0717,
+      "step": 2004
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9530694366460795e-05,
+      "loss": 0.101,
+      "step": 2006
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9530226464533034e-05,
+      "loss": 0.0867,
+      "step": 2008
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.952975856260528e-05,
+      "loss": 0.0734,
+      "step": 2010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9529290660677526e-05,
+      "loss": 0.0957,
+      "step": 2012
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.952882275874977e-05,
+      "loss": 0.1171,
+      "step": 2014
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.952835485682201e-05,
+      "loss": 0.103,
+      "step": 2016
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.952788695489426e-05,
+      "loss": 0.0913,
+      "step": 2018
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.95274190529665e-05,
+      "loss": 0.0911,
+      "step": 2020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.952695115103875e-05,
+      "loss": 0.0803,
+      "step": 2022
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.952648324911099e-05,
+      "loss": 0.0831,
+      "step": 2024
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.952601534718323e-05,
+      "loss": 0.0769,
+      "step": 2026
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.952554744525548e-05,
+      "loss": 0.072,
+      "step": 2028
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.952507954332772e-05,
+      "loss": 0.0863,
+      "step": 2030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9524611641399965e-05,
+      "loss": 0.0999,
+      "step": 2032
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9524143739472204e-05,
+      "loss": 0.097,
+      "step": 2034
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.952367583754445e-05,
+      "loss": 0.0917,
+      "step": 2036
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9523207935616696e-05,
+      "loss": 0.0717,
+      "step": 2038
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.952274003368894e-05,
+      "loss": 0.0909,
+      "step": 2040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.952227213176118e-05,
+      "loss": 0.0911,
+      "step": 2042
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.952180422983343e-05,
+      "loss": 0.0811,
+      "step": 2044
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.952133632790567e-05,
+      "loss": 0.0892,
+      "step": 2046
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.952086842597792e-05,
+      "loss": 0.0879,
+      "step": 2048
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.952040052405016e-05,
+      "loss": 0.084,
+      "step": 2050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9519932622122404e-05,
+      "loss": 0.075,
+      "step": 2052
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.951946472019465e-05,
+      "loss": 0.0923,
+      "step": 2054
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9518996818266896e-05,
+      "loss": 0.0782,
+      "step": 2056
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9518528916339135e-05,
+      "loss": 0.1017,
+      "step": 2058
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.951806101441138e-05,
+      "loss": 0.0976,
+      "step": 2060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.951759311248363e-05,
+      "loss": 0.0843,
+      "step": 2062
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.951712521055587e-05,
+      "loss": 0.1091,
+      "step": 2064
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.951665730862811e-05,
+      "loss": 0.0778,
+      "step": 2066
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.951618940670036e-05,
+      "loss": 0.08,
+      "step": 2068
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.95157215047726e-05,
+      "loss": 0.0813,
+      "step": 2070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.951525360284485e-05,
+      "loss": 0.1102,
+      "step": 2072
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.951478570091709e-05,
+      "loss": 0.0795,
+      "step": 2074
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9514317798989334e-05,
+      "loss": 0.0959,
+      "step": 2076
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9513849897061574e-05,
+      "loss": 0.0886,
+      "step": 2078
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9513381995133826e-05,
+      "loss": 0.0764,
+      "step": 2080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9512914093206065e-05,
+      "loss": 0.0692,
+      "step": 2082
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.951244619127831e-05,
+      "loss": 0.0869,
+      "step": 2084
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.951197828935055e-05,
+      "loss": 0.0853,
+      "step": 2086
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9511510387422796e-05,
+      "loss": 0.1056,
+      "step": 2088
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.951104248549504e-05,
+      "loss": 0.0794,
+      "step": 2090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.951057458356729e-05,
+      "loss": 0.0834,
+      "step": 2092
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.951010668163953e-05,
+      "loss": 0.1245,
+      "step": 2094
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.950963877971177e-05,
+      "loss": 0.0811,
+      "step": 2096
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.950917087778402e-05,
+      "loss": 0.0852,
+      "step": 2098
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9508702975856265e-05,
+      "loss": 0.0776,
+      "step": 2100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9508235073928504e-05,
+      "loss": 0.0913,
+      "step": 2102
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.950776717200075e-05,
+      "loss": 0.074,
+      "step": 2104
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9507299270072996e-05,
+      "loss": 0.0734,
+      "step": 2106
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.950683136814524e-05,
+      "loss": 0.0771,
+      "step": 2108
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.950636346621748e-05,
+      "loss": 0.0767,
+      "step": 2110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.950589556428973e-05,
+      "loss": 0.0856,
+      "step": 2112
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9505427662361966e-05,
+      "loss": 0.0851,
+      "step": 2114
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.950495976043422e-05,
+      "loss": 0.0734,
+      "step": 2116
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.950449185850646e-05,
+      "loss": 0.0829,
+      "step": 2118
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9504023956578704e-05,
+      "loss": 0.0889,
+      "step": 2120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.950355605465094e-05,
+      "loss": 0.0918,
+      "step": 2122
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9503088152723196e-05,
+      "loss": 0.1152,
+      "step": 2124
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9502620250795435e-05,
+      "loss": 0.1038,
+      "step": 2126
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.950215234886768e-05,
+      "loss": 0.0785,
+      "step": 2128
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.950168444693992e-05,
+      "loss": 0.0699,
+      "step": 2130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9501216545012166e-05,
+      "loss": 0.0966,
+      "step": 2132
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.950074864308441e-05,
+      "loss": 0.1138,
+      "step": 2134
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.950028074115666e-05,
+      "loss": 0.086,
+      "step": 2136
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.94998128392289e-05,
+      "loss": 0.0641,
+      "step": 2138
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.949934493730114e-05,
+      "loss": 0.0762,
+      "step": 2140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.949887703537339e-05,
+      "loss": 0.1115,
+      "step": 2142
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9498409133445634e-05,
+      "loss": 0.1117,
+      "step": 2144
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9497941231517874e-05,
+      "loss": 0.089,
+      "step": 2146
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.949747332959012e-05,
+      "loss": 0.0802,
+      "step": 2148
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9497005427662365e-05,
+      "loss": 0.0747,
+      "step": 2150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.949653752573461e-05,
+      "loss": 0.085,
+      "step": 2152
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.949606962380685e-05,
+      "loss": 0.0879,
+      "step": 2154
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9495601721879096e-05,
+      "loss": 0.0959,
+      "step": 2156
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.949513381995134e-05,
+      "loss": 0.0959,
+      "step": 2158
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.949466591802359e-05,
+      "loss": 0.0837,
+      "step": 2160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.949419801609583e-05,
+      "loss": 0.0731,
+      "step": 2162
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.949373011416807e-05,
+      "loss": 0.0748,
+      "step": 2164
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.949326221224031e-05,
+      "loss": 0.0935,
+      "step": 2166
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9492794310312565e-05,
+      "loss": 0.0672,
+      "step": 2168
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9492326408384804e-05,
+      "loss": 0.1234,
+      "step": 2170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.949185850645705e-05,
+      "loss": 0.0684,
+      "step": 2172
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.949139060452929e-05,
+      "loss": 0.0832,
+      "step": 2174
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.949092270260154e-05,
+      "loss": 0.0881,
+      "step": 2176
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.949045480067378e-05,
+      "loss": 0.0823,
+      "step": 2178
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.948998689874603e-05,
+      "loss": 0.1178,
+      "step": 2180
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9489518996818266e-05,
+      "loss": 0.091,
+      "step": 2182
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.948905109489051e-05,
+      "loss": 0.1102,
+      "step": 2184
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.948858319296276e-05,
+      "loss": 0.0945,
+      "step": 2186
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9488115291035004e-05,
+      "loss": 0.097,
+      "step": 2188
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.948764738910724e-05,
+      "loss": 0.0815,
+      "step": 2190
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.948717948717949e-05,
+      "loss": 0.0947,
+      "step": 2192
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9486711585251735e-05,
+      "loss": 0.0601,
+      "step": 2194
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.948624368332398e-05,
+      "loss": 0.079,
+      "step": 2196
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.948577578139622e-05,
+      "loss": 0.1026,
+      "step": 2198
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9485307879468466e-05,
+      "loss": 0.099,
+      "step": 2200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.948483997754071e-05,
+      "loss": 0.0662,
+      "step": 2202
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.948437207561296e-05,
+      "loss": 0.0896,
+      "step": 2204
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.94839041736852e-05,
+      "loss": 0.0807,
+      "step": 2206
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.948343627175744e-05,
+      "loss": 0.1046,
+      "step": 2208
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.948296836982969e-05,
+      "loss": 0.0735,
+      "step": 2210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9482500467901934e-05,
+      "loss": 0.0695,
+      "step": 2212
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9482032565974174e-05,
+      "loss": 0.0924,
+      "step": 2214
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.948156466404642e-05,
+      "loss": 0.1034,
+      "step": 2216
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.948109676211866e-05,
+      "loss": 0.0763,
+      "step": 2218
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.948062886019091e-05,
+      "loss": 0.0851,
+      "step": 2220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.948016095826315e-05,
+      "loss": 0.0719,
+      "step": 2222
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9479693056335396e-05,
+      "loss": 0.0899,
+      "step": 2224
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9479225154407636e-05,
+      "loss": 0.0821,
+      "step": 2226
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.947875725247988e-05,
+      "loss": 0.0697,
+      "step": 2228
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.947828935055213e-05,
+      "loss": 0.0901,
+      "step": 2230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.947782144862437e-05,
+      "loss": 0.0631,
+      "step": 2232
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.947735354669661e-05,
+      "loss": 0.078,
+      "step": 2234
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.947688564476886e-05,
+      "loss": 0.0959,
+      "step": 2236
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9476417742841104e-05,
+      "loss": 0.0922,
+      "step": 2238
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.947594984091335e-05,
+      "loss": 0.1077,
+      "step": 2240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.947548193898559e-05,
+      "loss": 0.0851,
+      "step": 2242
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9475014037057835e-05,
+      "loss": 0.0806,
+      "step": 2244
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.947454613513008e-05,
+      "loss": 0.0849,
+      "step": 2246
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.947407823320233e-05,
+      "loss": 0.093,
+      "step": 2248
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9473610331274566e-05,
+      "loss": 0.0759,
+      "step": 2250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.947314242934681e-05,
+      "loss": 0.0918,
+      "step": 2252
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.947267452741906e-05,
+      "loss": 0.0587,
+      "step": 2254
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9472206625491304e-05,
+      "loss": 0.0857,
+      "step": 2256
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.947173872356354e-05,
+      "loss": 0.0774,
+      "step": 2258
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.947127082163579e-05,
+      "loss": 0.0937,
+      "step": 2260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.947080291970803e-05,
+      "loss": 0.076,
+      "step": 2262
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.947033501778028e-05,
+      "loss": 0.0994,
+      "step": 2264
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.946986711585252e-05,
+      "loss": 0.0815,
+      "step": 2266
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9469399213924766e-05,
+      "loss": 0.108,
+      "step": 2268
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9468931311997005e-05,
+      "loss": 0.1041,
+      "step": 2270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.946846341006926e-05,
+      "loss": 0.0705,
+      "step": 2272
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.94679955081415e-05,
+      "loss": 0.1079,
+      "step": 2274
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.946752760621374e-05,
+      "loss": 0.0984,
+      "step": 2276
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.946705970428598e-05,
+      "loss": 0.0885,
+      "step": 2278
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.946659180235823e-05,
+      "loss": 0.0901,
+      "step": 2280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9466123900430474e-05,
+      "loss": 0.0859,
+      "step": 2282
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.946565599850271e-05,
+      "loss": 0.0949,
+      "step": 2284
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.946518809657496e-05,
+      "loss": 0.0816,
+      "step": 2286
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9464720194647205e-05,
+      "loss": 0.1089,
+      "step": 2288
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.946425229271945e-05,
+      "loss": 0.1052,
+      "step": 2290
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.946378439079169e-05,
+      "loss": 0.0762,
+      "step": 2292
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9463316488863936e-05,
+      "loss": 0.0896,
+      "step": 2294
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9462848586936175e-05,
+      "loss": 0.0887,
+      "step": 2296
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.946238068500843e-05,
+      "loss": 0.0937,
+      "step": 2298
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9461912783080667e-05,
+      "loss": 0.0998,
+      "step": 2300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.946144488115291e-05,
+      "loss": 0.0834,
+      "step": 2302
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.946097697922515e-05,
+      "loss": 0.0535,
+      "step": 2304
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9460509077297404e-05,
+      "loss": 0.0749,
+      "step": 2306
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9460041175369643e-05,
+      "loss": 0.072,
+      "step": 2308
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.945957327344189e-05,
+      "loss": 0.0742,
+      "step": 2310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.945910537151413e-05,
+      "loss": 0.092,
+      "step": 2312
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9458637469586374e-05,
+      "loss": 0.0796,
+      "step": 2314
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.945816956765862e-05,
+      "loss": 0.0694,
+      "step": 2316
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9457701665730866e-05,
+      "loss": 0.0766,
+      "step": 2318
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9457233763803105e-05,
+      "loss": 0.0923,
+      "step": 2320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.945676586187535e-05,
+      "loss": 0.1028,
+      "step": 2322
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.94562979599476e-05,
+      "loss": 0.0795,
+      "step": 2324
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.945583005801984e-05,
+      "loss": 0.078,
+      "step": 2326
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.945536215609208e-05,
+      "loss": 0.0799,
+      "step": 2328
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.945489425416433e-05,
+      "loss": 0.1079,
+      "step": 2330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9454426352236574e-05,
+      "loss": 0.0796,
+      "step": 2332
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.945395845030882e-05,
+      "loss": 0.098,
+      "step": 2334
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.945349054838106e-05,
+      "loss": 0.0919,
+      "step": 2336
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9453022646453305e-05,
+      "loss": 0.0801,
+      "step": 2338
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.945255474452555e-05,
+      "loss": 0.0834,
+      "step": 2340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.94520868425978e-05,
+      "loss": 0.0893,
+      "step": 2342
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9451618940670036e-05,
+      "loss": 0.1059,
+      "step": 2344
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.945115103874228e-05,
+      "loss": 0.0724,
+      "step": 2346
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.945068313681452e-05,
+      "loss": 0.0923,
+      "step": 2348
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9450215234886774e-05,
+      "loss": 0.075,
+      "step": 2350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.944974733295901e-05,
+      "loss": 0.0787,
+      "step": 2352
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.944927943103126e-05,
+      "loss": 0.0868,
+      "step": 2354
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.94488115291035e-05,
+      "loss": 0.0791,
+      "step": 2356
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.944834362717575e-05,
+      "loss": 0.0778,
+      "step": 2358
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.944787572524799e-05,
+      "loss": 0.0834,
+      "step": 2360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9447407823320236e-05,
+      "loss": 0.077,
+      "step": 2362
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9446939921392475e-05,
+      "loss": 0.0619,
+      "step": 2364
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.944647201946472e-05,
+      "loss": 0.0895,
+      "step": 2366
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9446004117536967e-05,
+      "loss": 0.0713,
+      "step": 2368
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.944553621560921e-05,
+      "loss": 0.0743,
+      "step": 2370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.944506831368145e-05,
+      "loss": 0.075,
+      "step": 2372
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.94446004117537e-05,
+      "loss": 0.0811,
+      "step": 2374
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9444132509825943e-05,
+      "loss": 0.0738,
+      "step": 2376
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.944366460789819e-05,
+      "loss": 0.0831,
+      "step": 2378
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.944319670597043e-05,
+      "loss": 0.0905,
+      "step": 2380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9442728804042674e-05,
+      "loss": 0.1058,
+      "step": 2382
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.944226090211492e-05,
+      "loss": 0.0913,
+      "step": 2384
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9441793000187166e-05,
+      "loss": 0.0977,
+      "step": 2386
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9441325098259405e-05,
+      "loss": 0.0655,
+      "step": 2388
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.944085719633165e-05,
+      "loss": 0.0907,
+      "step": 2390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.94403892944039e-05,
+      "loss": 0.1052,
+      "step": 2392
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.943992139247614e-05,
+      "loss": 0.0933,
+      "step": 2394
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.943945349054838e-05,
+      "loss": 0.075,
+      "step": 2396
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.943898558862063e-05,
+      "loss": 0.0644,
+      "step": 2398
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.943851768669287e-05,
+      "loss": 0.0824,
+      "step": 2400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.943804978476512e-05,
+      "loss": 0.09,
+      "step": 2402
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.943758188283736e-05,
+      "loss": 0.0828,
+      "step": 2404
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9437113980909605e-05,
+      "loss": 0.0981,
+      "step": 2406
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9436646078981844e-05,
+      "loss": 0.0962,
+      "step": 2408
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.943617817705409e-05,
+      "loss": 0.0749,
+      "step": 2410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9435710275126336e-05,
+      "loss": 0.1149,
+      "step": 2412
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.943524237319858e-05,
+      "loss": 0.1069,
+      "step": 2414
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.943477447127082e-05,
+      "loss": 0.0873,
+      "step": 2416
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.943430656934307e-05,
+      "loss": 0.0975,
+      "step": 2418
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.943383866741531e-05,
+      "loss": 0.0846,
+      "step": 2420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.943337076548756e-05,
+      "loss": 0.0992,
+      "step": 2422
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.94329028635598e-05,
+      "loss": 0.0794,
+      "step": 2424
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9432434961632044e-05,
+      "loss": 0.0782,
+      "step": 2426
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.943196705970429e-05,
+      "loss": 0.0959,
+      "step": 2428
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9431499157776536e-05,
+      "loss": 0.0882,
+      "step": 2430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9431031255848775e-05,
+      "loss": 0.0715,
+      "step": 2432
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.943056335392102e-05,
+      "loss": 0.074,
+      "step": 2434
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9430095451993267e-05,
+      "loss": 0.089,
+      "step": 2436
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.942962755006551e-05,
+      "loss": 0.0881,
+      "step": 2438
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.942915964813775e-05,
+      "loss": 0.1152,
+      "step": 2440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.942869174621e-05,
+      "loss": 0.0627,
+      "step": 2442
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.942822384428224e-05,
+      "loss": 0.0979,
+      "step": 2444
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.942775594235449e-05,
+      "loss": 0.0961,
+      "step": 2446
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.942728804042673e-05,
+      "loss": 0.1038,
+      "step": 2448
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9426820138498974e-05,
+      "loss": 0.0805,
+      "step": 2450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9426352236571214e-05,
+      "loss": 0.0731,
+      "step": 2452
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9425884334643466e-05,
+      "loss": 0.0756,
+      "step": 2454
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9425416432715705e-05,
+      "loss": 0.0756,
+      "step": 2456
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.942494853078795e-05,
+      "loss": 0.109,
+      "step": 2458
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.942448062886019e-05,
+      "loss": 0.0913,
+      "step": 2460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9424012726932436e-05,
+      "loss": 0.0695,
+      "step": 2462
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.942354482500468e-05,
+      "loss": 0.0919,
+      "step": 2464
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.942307692307693e-05,
+      "loss": 0.0844,
+      "step": 2466
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.942260902114917e-05,
+      "loss": 0.0902,
+      "step": 2468
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.942214111922141e-05,
+      "loss": 0.0867,
+      "step": 2470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.942167321729366e-05,
+      "loss": 0.1038,
+      "step": 2472
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9421205315365905e-05,
+      "loss": 0.0879,
+      "step": 2474
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9420737413438144e-05,
+      "loss": 0.0834,
+      "step": 2476
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.942026951151039e-05,
+      "loss": 0.0751,
+      "step": 2478
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9419801609582636e-05,
+      "loss": 0.0794,
+      "step": 2480
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.941933370765488e-05,
+      "loss": 0.0741,
+      "step": 2482
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.941886580572712e-05,
+      "loss": 0.0955,
+      "step": 2484
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.941839790379937e-05,
+      "loss": 0.0644,
+      "step": 2486
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.941793000187161e-05,
+      "loss": 0.0998,
+      "step": 2488
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.941746209994386e-05,
+      "loss": 0.0783,
+      "step": 2490
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.94169941980161e-05,
+      "loss": 0.0843,
+      "step": 2492
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9416526296088344e-05,
+      "loss": 0.0812,
+      "step": 2494
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.941605839416058e-05,
+      "loss": 0.0885,
+      "step": 2496
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9415590492232836e-05,
+      "loss": 0.0798,
+      "step": 2498
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9415122590305075e-05,
+      "loss": 0.0867,
+      "step": 2500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.941465468837732e-05,
+      "loss": 0.0652,
+      "step": 2502
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.941418678644956e-05,
+      "loss": 0.074,
+      "step": 2504
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9413718884521806e-05,
+      "loss": 0.0926,
+      "step": 2506
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.941325098259405e-05,
+      "loss": 0.0821,
+      "step": 2508
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.94127830806663e-05,
+      "loss": 0.0757,
+      "step": 2510
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.941231517873854e-05,
+      "loss": 0.0862,
+      "step": 2512
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.941184727681078e-05,
+      "loss": 0.0833,
+      "step": 2514
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.941137937488303e-05,
+      "loss": 0.086,
+      "step": 2516
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9410911472955274e-05,
+      "loss": 0.082,
+      "step": 2518
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9410443571027514e-05,
+      "loss": 0.0901,
+      "step": 2520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.940997566909976e-05,
+      "loss": 0.0925,
+      "step": 2522
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9409507767172005e-05,
+      "loss": 0.0891,
+      "step": 2524
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.940903986524425e-05,
+      "loss": 0.0839,
+      "step": 2526
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.940857196331649e-05,
+      "loss": 0.0825,
+      "step": 2528
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9408104061388736e-05,
+      "loss": 0.1103,
+      "step": 2530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.940763615946098e-05,
+      "loss": 0.0753,
+      "step": 2532
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.940716825753322e-05,
+      "loss": 0.1045,
+      "step": 2534
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.940670035560547e-05,
+      "loss": 0.0909,
+      "step": 2536
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9406232453677706e-05,
+      "loss": 0.0801,
+      "step": 2538
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.940576455174995e-05,
+      "loss": 0.0956,
+      "step": 2540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.94052966498222e-05,
+      "loss": 0.0888,
+      "step": 2542
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9404828747894444e-05,
+      "loss": 0.0784,
+      "step": 2544
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.940436084596668e-05,
+      "loss": 0.0807,
+      "step": 2546
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.940389294403893e-05,
+      "loss": 0.0854,
+      "step": 2548
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9403425042111175e-05,
+      "loss": 0.0882,
+      "step": 2550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.940295714018342e-05,
+      "loss": 0.0825,
+      "step": 2552
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.940248923825566e-05,
+      "loss": 0.0677,
+      "step": 2554
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9402021336327906e-05,
+      "loss": 0.0805,
+      "step": 2556
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.940155343440015e-05,
+      "loss": 0.063,
+      "step": 2558
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.94010855324724e-05,
+      "loss": 0.0733,
+      "step": 2560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.940061763054464e-05,
+      "loss": 0.0732,
+      "step": 2562
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.940014972861688e-05,
+      "loss": 0.1077,
+      "step": 2564
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.939968182668913e-05,
+      "loss": 0.0888,
+      "step": 2566
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9399213924761375e-05,
+      "loss": 0.0857,
+      "step": 2568
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9398746022833614e-05,
+      "loss": 0.0794,
+      "step": 2570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.939827812090586e-05,
+      "loss": 0.0967,
+      "step": 2572
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.93978102189781e-05,
+      "loss": 0.1005,
+      "step": 2574
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.939734231705035e-05,
+      "loss": 0.0941,
+      "step": 2576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.939687441512259e-05,
+      "loss": 0.1091,
+      "step": 2578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.939640651319484e-05,
+      "loss": 0.0826,
+      "step": 2580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9395938611267076e-05,
+      "loss": 0.0791,
+      "step": 2582
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.939547070933933e-05,
+      "loss": 0.0869,
+      "step": 2584
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.939500280741157e-05,
+      "loss": 0.0631,
+      "step": 2586
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9394534905483814e-05,
+      "loss": 0.0909,
+      "step": 2588
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.939406700355605e-05,
+      "loss": 0.0927,
+      "step": 2590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.93935991016283e-05,
+      "loss": 0.0659,
+      "step": 2592
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9393131199700545e-05,
+      "loss": 0.0723,
+      "step": 2594
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.939266329777279e-05,
+      "loss": 0.1038,
+      "step": 2596
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.939219539584503e-05,
+      "loss": 0.0827,
+      "step": 2598
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9391727493917275e-05,
+      "loss": 0.0997,
+      "step": 2600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.939125959198952e-05,
+      "loss": 0.0814,
+      "step": 2602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.939079169006177e-05,
+      "loss": 0.0988,
+      "step": 2604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9390323788134006e-05,
+      "loss": 0.094,
+      "step": 2606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.938985588620625e-05,
+      "loss": 0.0926,
+      "step": 2608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.93893879842785e-05,
+      "loss": 0.0775,
+      "step": 2610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9388920082350744e-05,
+      "loss": 0.0849,
+      "step": 2612
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.938845218042298e-05,
+      "loss": 0.0768,
+      "step": 2614
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.938798427849523e-05,
+      "loss": 0.1012,
+      "step": 2616
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9387516376567475e-05,
+      "loss": 0.1016,
+      "step": 2618
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.938704847463972e-05,
+      "loss": 0.0953,
+      "step": 2620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.938658057271196e-05,
+      "loss": 0.1024,
+      "step": 2622
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9386112670784206e-05,
+      "loss": 0.1112,
+      "step": 2624
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9385644768856445e-05,
+      "loss": 0.0912,
+      "step": 2626
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.93851768669287e-05,
+      "loss": 0.0913,
+      "step": 2628
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.938470896500094e-05,
+      "loss": 0.081,
+      "step": 2630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.938424106307318e-05,
+      "loss": 0.0804,
+      "step": 2632
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.938377316114542e-05,
+      "loss": 0.0689,
+      "step": 2634
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9383305259217675e-05,
+      "loss": 0.0885,
+      "step": 2636
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9382837357289914e-05,
+      "loss": 0.0695,
+      "step": 2638
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.938236945536216e-05,
+      "loss": 0.0922,
+      "step": 2640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.93819015534344e-05,
+      "loss": 0.0983,
+      "step": 2642
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9381433651506645e-05,
+      "loss": 0.0997,
+      "step": 2644
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.938096574957889e-05,
+      "loss": 0.0841,
+      "step": 2646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.938049784765114e-05,
+      "loss": 0.0863,
+      "step": 2648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9380029945723376e-05,
+      "loss": 0.0944,
+      "step": 2650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.937956204379562e-05,
+      "loss": 0.1024,
+      "step": 2652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.937909414186787e-05,
+      "loss": 0.0689,
+      "step": 2654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9378626239940114e-05,
+      "loss": 0.0946,
+      "step": 2656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.937815833801235e-05,
+      "loss": 0.0825,
+      "step": 2658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.93776904360846e-05,
+      "loss": 0.0975,
+      "step": 2660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9377222534156845e-05,
+      "loss": 0.0915,
+      "step": 2662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.937675463222909e-05,
+      "loss": 0.0676,
+      "step": 2664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.937628673030133e-05,
+      "loss": 0.102,
+      "step": 2666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9375818828373575e-05,
+      "loss": 0.1083,
+      "step": 2668
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.937535092644582e-05,
+      "loss": 0.0844,
+      "step": 2670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.937488302451807e-05,
+      "loss": 0.0943,
+      "step": 2672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9374415122590306e-05,
+      "loss": 0.0872,
+      "step": 2674
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.937394722066255e-05,
+      "loss": 0.0766,
+      "step": 2676
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.937347931873479e-05,
+      "loss": 0.069,
+      "step": 2678
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9373011416807044e-05,
+      "loss": 0.0887,
+      "step": 2680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.937254351487928e-05,
+      "loss": 0.0843,
+      "step": 2682
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.937207561295153e-05,
+      "loss": 0.062,
+      "step": 2684
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.937160771102377e-05,
+      "loss": 0.1122,
+      "step": 2686
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9371139809096014e-05,
+      "loss": 0.0852,
+      "step": 2688
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.937067190716826e-05,
+      "loss": 0.0827,
+      "step": 2690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9370204005240506e-05,
+      "loss": 0.0627,
+      "step": 2692
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9369736103312745e-05,
+      "loss": 0.1137,
+      "step": 2694
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.936926820138499e-05,
+      "loss": 0.1036,
+      "step": 2696
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.936880029945724e-05,
+      "loss": 0.0767,
+      "step": 2698
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.936833239752948e-05,
+      "loss": 0.0915,
+      "step": 2700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.936786449560172e-05,
+      "loss": 0.1017,
+      "step": 2702
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.936739659367397e-05,
+      "loss": 0.0586,
+      "step": 2704
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9366928691746214e-05,
+      "loss": 0.1029,
+      "step": 2706
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.936646078981846e-05,
+      "loss": 0.0846,
+      "step": 2708
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.93659928878907e-05,
+      "loss": 0.0791,
+      "step": 2710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9365524985962945e-05,
+      "loss": 0.0909,
+      "step": 2712
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.936505708403519e-05,
+      "loss": 0.102,
+      "step": 2714
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.936458918210744e-05,
+      "loss": 0.0842,
+      "step": 2716
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9364121280179676e-05,
+      "loss": 0.0893,
+      "step": 2718
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.936365337825192e-05,
+      "loss": 0.1152,
+      "step": 2720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.936318547632416e-05,
+      "loss": 0.0785,
+      "step": 2722
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9362717574396414e-05,
+      "loss": 0.0631,
+      "step": 2724
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.936224967246865e-05,
+      "loss": 0.0979,
+      "step": 2726
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.93617817705409e-05,
+      "loss": 0.0591,
+      "step": 2728
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.936131386861314e-05,
+      "loss": 0.1029,
+      "step": 2730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.936084596668539e-05,
+      "loss": 0.1041,
+      "step": 2732
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.936037806475763e-05,
+      "loss": 0.0838,
+      "step": 2734
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9359910162829876e-05,
+      "loss": 0.1018,
+      "step": 2736
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9359442260902115e-05,
+      "loss": 0.1033,
+      "step": 2738
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.935897435897436e-05,
+      "loss": 0.0843,
+      "step": 2740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9358506457046606e-05,
+      "loss": 0.0835,
+      "step": 2742
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.935803855511885e-05,
+      "loss": 0.1003,
+      "step": 2744
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.935757065319109e-05,
+      "loss": 0.0987,
+      "step": 2746
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.935710275126334e-05,
+      "loss": 0.0769,
+      "step": 2748
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.935663484933558e-05,
+      "loss": 0.0842,
+      "step": 2750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.935616694740783e-05,
+      "loss": 0.0742,
+      "step": 2752
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.935569904548007e-05,
+      "loss": 0.0921,
+      "step": 2754
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9355231143552314e-05,
+      "loss": 0.0929,
+      "step": 2756
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.935476324162456e-05,
+      "loss": 0.0645,
+      "step": 2758
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9354295339696806e-05,
+      "loss": 0.0891,
+      "step": 2760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9353827437769045e-05,
+      "loss": 0.0806,
+      "step": 2762
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.935335953584129e-05,
+      "loss": 0.1043,
+      "step": 2764
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.935289163391354e-05,
+      "loss": 0.1009,
+      "step": 2766
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.935242373198578e-05,
+      "loss": 0.0715,
+      "step": 2768
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.935195583005802e-05,
+      "loss": 0.0918,
+      "step": 2770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.935148792813027e-05,
+      "loss": 0.0957,
+      "step": 2772
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.935102002620251e-05,
+      "loss": 0.0927,
+      "step": 2774
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.935055212427476e-05,
+      "loss": 0.1075,
+      "step": 2776
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9350084222347e-05,
+      "loss": 0.0845,
+      "step": 2778
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9349616320419245e-05,
+      "loss": 0.0982,
+      "step": 2780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9349148418491484e-05,
+      "loss": 0.1105,
+      "step": 2782
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.934868051656374e-05,
+      "loss": 0.1015,
+      "step": 2784
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9348212614635976e-05,
+      "loss": 0.0815,
+      "step": 2786
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9347744712708215e-05,
+      "loss": 0.1058,
+      "step": 2788
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.934727681078046e-05,
+      "loss": 0.0769,
+      "step": 2790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.934680890885271e-05,
+      "loss": 0.0965,
+      "step": 2792
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.934634100692495e-05,
+      "loss": 0.0706,
+      "step": 2794
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.934587310499719e-05,
+      "loss": 0.077,
+      "step": 2796
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.934540520306944e-05,
+      "loss": 0.0854,
+      "step": 2798
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9344937301141684e-05,
+      "loss": 0.0888,
+      "step": 2800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.934446939921393e-05,
+      "loss": 0.07,
+      "step": 2802
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.934400149728617e-05,
+      "loss": 0.0731,
+      "step": 2804
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9343533595358415e-05,
+      "loss": 0.0812,
+      "step": 2806
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9343065693430654e-05,
+      "loss": 0.0853,
+      "step": 2808
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9342597791502906e-05,
+      "loss": 0.084,
+      "step": 2810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9342129889575146e-05,
+      "loss": 0.093,
+      "step": 2812
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.934166198764739e-05,
+      "loss": 0.0735,
+      "step": 2814
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.934119408571963e-05,
+      "loss": 0.07,
+      "step": 2816
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9340726183791877e-05,
+      "loss": 0.1021,
+      "step": 2818
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.934025828186412e-05,
+      "loss": 0.0915,
+      "step": 2820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.933979037993637e-05,
+      "loss": 0.0718,
+      "step": 2822
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.933932247800861e-05,
+      "loss": 0.0619,
+      "step": 2824
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9338854576080853e-05,
+      "loss": 0.1078,
+      "step": 2826
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.93383866741531e-05,
+      "loss": 0.1037,
+      "step": 2828
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9337918772225345e-05,
+      "loss": 0.104,
+      "step": 2830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9337450870297584e-05,
+      "loss": 0.0801,
+      "step": 2832
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.933698296836983e-05,
+      "loss": 0.079,
+      "step": 2834
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9336515066442076e-05,
+      "loss": 0.0832,
+      "step": 2836
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.933604716451432e-05,
+      "loss": 0.1078,
+      "step": 2838
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.933557926258656e-05,
+      "loss": 0.1121,
+      "step": 2840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.933511136065881e-05,
+      "loss": 0.0727,
+      "step": 2842
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.933464345873105e-05,
+      "loss": 0.0748,
+      "step": 2844
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.93341755568033e-05,
+      "loss": 0.1016,
+      "step": 2846
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.933370765487554e-05,
+      "loss": 0.0796,
+      "step": 2848
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9333239752947784e-05,
+      "loss": 0.0688,
+      "step": 2850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.933277185102002e-05,
+      "loss": 0.0842,
+      "step": 2852
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9332303949092276e-05,
+      "loss": 0.079,
+      "step": 2854
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9331836047164515e-05,
+      "loss": 0.0742,
+      "step": 2856
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.933136814523676e-05,
+      "loss": 0.1007,
+      "step": 2858
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9330900243309e-05,
+      "loss": 0.1008,
+      "step": 2860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.933043234138125e-05,
+      "loss": 0.1142,
+      "step": 2862
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.932996443945349e-05,
+      "loss": 0.0813,
+      "step": 2864
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.932949653752574e-05,
+      "loss": 0.1144,
+      "step": 2866
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.932902863559798e-05,
+      "loss": 0.0992,
+      "step": 2868
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.932856073367022e-05,
+      "loss": 0.0968,
+      "step": 2870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.932809283174247e-05,
+      "loss": 0.1014,
+      "step": 2872
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9327624929814715e-05,
+      "loss": 0.1101,
+      "step": 2874
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9327157027886954e-05,
+      "loss": 0.1064,
+      "step": 2876
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.93266891259592e-05,
+      "loss": 0.0975,
+      "step": 2878
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9326221224031446e-05,
+      "loss": 0.0771,
+      "step": 2880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.932575332210369e-05,
+      "loss": 0.0935,
+      "step": 2882
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.932528542017593e-05,
+      "loss": 0.0904,
+      "step": 2884
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9324817518248177e-05,
+      "loss": 0.089,
+      "step": 2886
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.932434961632042e-05,
+      "loss": 0.0849,
+      "step": 2888
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.932388171439267e-05,
+      "loss": 0.0978,
+      "step": 2890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.932341381246491e-05,
+      "loss": 0.0862,
+      "step": 2892
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9322945910537153e-05,
+      "loss": 0.1218,
+      "step": 2894
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.93224780086094e-05,
+      "loss": 0.0879,
+      "step": 2896
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9322010106681645e-05,
+      "loss": 0.0632,
+      "step": 2898
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9321542204753884e-05,
+      "loss": 0.0839,
+      "step": 2900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.932107430282613e-05,
+      "loss": 0.0785,
+      "step": 2902
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.932060640089837e-05,
+      "loss": 0.081,
+      "step": 2904
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.932013849897062e-05,
+      "loss": 0.0749,
+      "step": 2906
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.931967059704286e-05,
+      "loss": 0.0961,
+      "step": 2908
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.931920269511511e-05,
+      "loss": 0.0717,
+      "step": 2910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9318734793187346e-05,
+      "loss": 0.0949,
+      "step": 2912
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.93182668912596e-05,
+      "loss": 0.1055,
+      "step": 2914
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.931779898933184e-05,
+      "loss": 0.1064,
+      "step": 2916
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9317331087404084e-05,
+      "loss": 0.0878,
+      "step": 2918
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.931686318547632e-05,
+      "loss": 0.1077,
+      "step": 2920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.931639528354857e-05,
+      "loss": 0.097,
+      "step": 2922
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9315927381620815e-05,
+      "loss": 0.0931,
+      "step": 2924
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.931545947969306e-05,
+      "loss": 0.0806,
+      "step": 2926
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.93149915777653e-05,
+      "loss": 0.1012,
+      "step": 2928
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9314523675837546e-05,
+      "loss": 0.0795,
+      "step": 2930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.931405577390979e-05,
+      "loss": 0.0863,
+      "step": 2932
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.931358787198204e-05,
+      "loss": 0.1046,
+      "step": 2934
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.931311997005428e-05,
+      "loss": 0.0997,
+      "step": 2936
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.931265206812652e-05,
+      "loss": 0.0833,
+      "step": 2938
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.931218416619877e-05,
+      "loss": 0.0915,
+      "step": 2940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9311716264271015e-05,
+      "loss": 0.0701,
+      "step": 2942
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9311248362343254e-05,
+      "loss": 0.0793,
+      "step": 2944
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.93107804604155e-05,
+      "loss": 0.0771,
+      "step": 2946
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9310312558487746e-05,
+      "loss": 0.0764,
+      "step": 2948
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.930984465655999e-05,
+      "loss": 0.1202,
+      "step": 2950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.930937675463223e-05,
+      "loss": 0.0874,
+      "step": 2952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.930890885270448e-05,
+      "loss": 0.0884,
+      "step": 2954
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9308440950776716e-05,
+      "loss": 0.0876,
+      "step": 2956
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.930797304884897e-05,
+      "loss": 0.0696,
+      "step": 2958
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.930750514692121e-05,
+      "loss": 0.0949,
+      "step": 2960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9307037244993453e-05,
+      "loss": 0.0941,
+      "step": 2962
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.930656934306569e-05,
+      "loss": 0.0809,
+      "step": 2964
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.930610144113794e-05,
+      "loss": 0.0756,
+      "step": 2966
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9305633539210184e-05,
+      "loss": 0.0949,
+      "step": 2968
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.930516563728243e-05,
+      "loss": 0.0901,
+      "step": 2970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.930469773535467e-05,
+      "loss": 0.0894,
+      "step": 2972
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9304229833426915e-05,
+      "loss": 0.0945,
+      "step": 2974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.930376193149916e-05,
+      "loss": 0.089,
+      "step": 2976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.930329402957141e-05,
+      "loss": 0.1,
+      "step": 2978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9302826127643646e-05,
+      "loss": 0.0658,
+      "step": 2980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.930235822571589e-05,
+      "loss": 0.1,
+      "step": 2982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.930189032378814e-05,
+      "loss": 0.0922,
+      "step": 2984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9301422421860384e-05,
+      "loss": 0.0654,
+      "step": 2986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.930095451993262e-05,
+      "loss": 0.1063,
+      "step": 2988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.930048661800487e-05,
+      "loss": 0.0785,
+      "step": 2990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9300018716077115e-05,
+      "loss": 0.0839,
+      "step": 2992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.929955081414936e-05,
+      "loss": 0.0981,
+      "step": 2994
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.92990829122216e-05,
+      "loss": 0.0758,
+      "step": 2996
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9298615010293846e-05,
+      "loss": 0.08,
+      "step": 2998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9298147108366085e-05,
+      "loss": 0.104,
+      "step": 3000
+    },
+    {
+      "epoch": 0.42,
+      "eval_gen_len": 31.5172,
+      "eval_loss": 1.0198298692703247,
+      "eval_meteor": 0.0411,
+      "eval_runtime": 15.656,
+      "eval_samples_per_second": 3.705,
+      "eval_steps_per_second": 0.511,
+      "step": 3000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.929767920643834e-05,
+      "loss": 0.1136,
+      "step": 3002
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.929721130451058e-05,
+      "loss": 0.0673,
+      "step": 3004
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.929674340258282e-05,
+      "loss": 0.1076,
+      "step": 3006
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.929627550065506e-05,
+      "loss": 0.0948,
+      "step": 3008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9295807598727315e-05,
+      "loss": 0.0816,
+      "step": 3010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9295339696799554e-05,
+      "loss": 0.0813,
+      "step": 3012
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.92948717948718e-05,
+      "loss": 0.087,
+      "step": 3014
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.929440389294404e-05,
+      "loss": 0.0923,
+      "step": 3016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9293935991016285e-05,
+      "loss": 0.0832,
+      "step": 3018
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.929346808908853e-05,
+      "loss": 0.121,
+      "step": 3020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.929300018716078e-05,
+      "loss": 0.0928,
+      "step": 3022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9292532285233016e-05,
+      "loss": 0.0719,
+      "step": 3024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.929206438330526e-05,
+      "loss": 0.0995,
+      "step": 3026
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.929159648137751e-05,
+      "loss": 0.0781,
+      "step": 3028
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9291128579449754e-05,
+      "loss": 0.1279,
+      "step": 3030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.929066067752199e-05,
+      "loss": 0.0931,
+      "step": 3032
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.929019277559424e-05,
+      "loss": 0.0818,
+      "step": 3034
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9289724873666484e-05,
+      "loss": 0.0922,
+      "step": 3036
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.928925697173873e-05,
+      "loss": 0.0798,
+      "step": 3038
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.928878906981097e-05,
+      "loss": 0.0919,
+      "step": 3040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.928832116788321e-05,
+      "loss": 0.0774,
+      "step": 3042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.928785326595546e-05,
+      "loss": 0.0971,
+      "step": 3044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.92873853640277e-05,
+      "loss": 0.0831,
+      "step": 3046
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9286917462099946e-05,
+      "loss": 0.1036,
+      "step": 3048
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9286449560172186e-05,
+      "loss": 0.0944,
+      "step": 3050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.928598165824443e-05,
+      "loss": 0.0763,
+      "step": 3052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.928551375631668e-05,
+      "loss": 0.0883,
+      "step": 3054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.928504585438892e-05,
+      "loss": 0.0872,
+      "step": 3056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.928457795246116e-05,
+      "loss": 0.1013,
+      "step": 3058
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.928411005053341e-05,
+      "loss": 0.0789,
+      "step": 3060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9283642148605654e-05,
+      "loss": 0.0916,
+      "step": 3062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.92831742466779e-05,
+      "loss": 0.0894,
+      "step": 3064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.928270634475014e-05,
+      "loss": 0.0807,
+      "step": 3066
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9282238442822385e-05,
+      "loss": 0.0952,
+      "step": 3068
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.928177054089463e-05,
+      "loss": 0.1163,
+      "step": 3070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.928130263896688e-05,
+      "loss": 0.0867,
+      "step": 3072
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9280834737039116e-05,
+      "loss": 0.0805,
+      "step": 3074
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.928036683511136e-05,
+      "loss": 0.0857,
+      "step": 3076
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.927989893318361e-05,
+      "loss": 0.103,
+      "step": 3078
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9279431031255854e-05,
+      "loss": 0.0869,
+      "step": 3080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.927896312932809e-05,
+      "loss": 0.0897,
+      "step": 3082
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.927849522740034e-05,
+      "loss": 0.0912,
+      "step": 3084
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.927802732547258e-05,
+      "loss": 0.1038,
+      "step": 3086
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.927755942354483e-05,
+      "loss": 0.0832,
+      "step": 3088
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.927709152161707e-05,
+      "loss": 0.1027,
+      "step": 3090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9276623619689316e-05,
+      "loss": 0.0765,
+      "step": 3092
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9276155717761555e-05,
+      "loss": 0.0644,
+      "step": 3094
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.927568781583381e-05,
+      "loss": 0.0852,
+      "step": 3096
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.927521991390605e-05,
+      "loss": 0.1084,
+      "step": 3098
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.927475201197829e-05,
+      "loss": 0.1356,
+      "step": 3100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.927428411005053e-05,
+      "loss": 0.1009,
+      "step": 3102
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.927381620812278e-05,
+      "loss": 0.0932,
+      "step": 3104
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9273348306195024e-05,
+      "loss": 0.0939,
+      "step": 3106
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.927288040426727e-05,
+      "loss": 0.1043,
+      "step": 3108
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.927241250233951e-05,
+      "loss": 0.0928,
+      "step": 3110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9271944600411755e-05,
+      "loss": 0.0829,
+      "step": 3112
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9271476698484e-05,
+      "loss": 0.0711,
+      "step": 3114
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9271008796556246e-05,
+      "loss": 0.0959,
+      "step": 3116
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9270540894628486e-05,
+      "loss": 0.0769,
+      "step": 3118
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.927007299270073e-05,
+      "loss": 0.087,
+      "step": 3120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.926960509077298e-05,
+      "loss": 0.0955,
+      "step": 3122
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.926913718884522e-05,
+      "loss": 0.0665,
+      "step": 3124
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.926866928691746e-05,
+      "loss": 0.108,
+      "step": 3126
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.926820138498971e-05,
+      "loss": 0.0868,
+      "step": 3128
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.926773348306195e-05,
+      "loss": 0.0698,
+      "step": 3130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.92672655811342e-05,
+      "loss": 0.1027,
+      "step": 3132
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.926679767920644e-05,
+      "loss": 0.0982,
+      "step": 3134
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9266329777278685e-05,
+      "loss": 0.1106,
+      "step": 3136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9265861875350924e-05,
+      "loss": 0.106,
+      "step": 3138
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.926539397342318e-05,
+      "loss": 0.0866,
+      "step": 3140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9264926071495416e-05,
+      "loss": 0.0842,
+      "step": 3142
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.926445816956766e-05,
+      "loss": 0.0922,
+      "step": 3144
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.92639902676399e-05,
+      "loss": 0.0924,
+      "step": 3146
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.926352236571215e-05,
+      "loss": 0.0974,
+      "step": 3148
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.926305446378439e-05,
+      "loss": 0.0888,
+      "step": 3150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.926258656185664e-05,
+      "loss": 0.0867,
+      "step": 3152
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.926211865992888e-05,
+      "loss": 0.084,
+      "step": 3154
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9261650758001124e-05,
+      "loss": 0.0833,
+      "step": 3156
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.926118285607337e-05,
+      "loss": 0.0991,
+      "step": 3158
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9260714954145616e-05,
+      "loss": 0.0926,
+      "step": 3160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9260247052217855e-05,
+      "loss": 0.0762,
+      "step": 3162
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.92597791502901e-05,
+      "loss": 0.0903,
+      "step": 3164
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.925931124836235e-05,
+      "loss": 0.0944,
+      "step": 3166
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.925884334643459e-05,
+      "loss": 0.0783,
+      "step": 3168
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.925837544450683e-05,
+      "loss": 0.0828,
+      "step": 3170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.925790754257908e-05,
+      "loss": 0.0821,
+      "step": 3172
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9257439640651324e-05,
+      "loss": 0.0774,
+      "step": 3174
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.925697173872357e-05,
+      "loss": 0.0988,
+      "step": 3176
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.925650383679581e-05,
+      "loss": 0.0883,
+      "step": 3178
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9256035934868055e-05,
+      "loss": 0.0986,
+      "step": 3180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9255568032940294e-05,
+      "loss": 0.0735,
+      "step": 3182
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9255100131012546e-05,
+      "loss": 0.085,
+      "step": 3184
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9254632229084786e-05,
+      "loss": 0.0614,
+      "step": 3186
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.925416432715703e-05,
+      "loss": 0.1033,
+      "step": 3188
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.925369642522927e-05,
+      "loss": 0.0694,
+      "step": 3190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.925322852330152e-05,
+      "loss": 0.1439,
+      "step": 3192
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.925276062137376e-05,
+      "loss": 0.0791,
+      "step": 3194
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.925229271944601e-05,
+      "loss": 0.0969,
+      "step": 3196
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.925182481751825e-05,
+      "loss": 0.1305,
+      "step": 3198
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.925135691559049e-05,
+      "loss": 0.1025,
+      "step": 3200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.925088901366274e-05,
+      "loss": 0.0891,
+      "step": 3202
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9250421111734985e-05,
+      "loss": 0.0853,
+      "step": 3204
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9249953209807224e-05,
+      "loss": 0.0804,
+      "step": 3206
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.924948530787947e-05,
+      "loss": 0.1006,
+      "step": 3208
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9249017405951716e-05,
+      "loss": 0.0594,
+      "step": 3210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.924854950402396e-05,
+      "loss": 0.0936,
+      "step": 3212
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.92480816020962e-05,
+      "loss": 0.0892,
+      "step": 3214
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.924761370016845e-05,
+      "loss": 0.0965,
+      "step": 3216
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.924714579824069e-05,
+      "loss": 0.0896,
+      "step": 3218
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.924667789631294e-05,
+      "loss": 0.0858,
+      "step": 3220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.924620999438518e-05,
+      "loss": 0.1017,
+      "step": 3222
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9245742092457424e-05,
+      "loss": 0.1014,
+      "step": 3224
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.924527419052967e-05,
+      "loss": 0.1362,
+      "step": 3226
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9244806288601916e-05,
+      "loss": 0.0861,
+      "step": 3228
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9244338386674155e-05,
+      "loss": 0.0786,
+      "step": 3230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.92438704847464e-05,
+      "loss": 0.0862,
+      "step": 3232
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.924340258281864e-05,
+      "loss": 0.0964,
+      "step": 3234
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.924293468089089e-05,
+      "loss": 0.0864,
+      "step": 3236
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.924246677896313e-05,
+      "loss": 0.0824,
+      "step": 3238
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.924199887703538e-05,
+      "loss": 0.0787,
+      "step": 3240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.924153097510762e-05,
+      "loss": 0.0682,
+      "step": 3242
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.924106307317986e-05,
+      "loss": 0.0717,
+      "step": 3244
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.924059517125211e-05,
+      "loss": 0.0737,
+      "step": 3246
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9240127269324355e-05,
+      "loss": 0.0897,
+      "step": 3248
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9239659367396594e-05,
+      "loss": 0.0791,
+      "step": 3250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.923919146546884e-05,
+      "loss": 0.0948,
+      "step": 3252
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9238723563541086e-05,
+      "loss": 0.0983,
+      "step": 3254
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.923825566161333e-05,
+      "loss": 0.09,
+      "step": 3256
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.923778775968557e-05,
+      "loss": 0.0888,
+      "step": 3258
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9237319857757817e-05,
+      "loss": 0.0903,
+      "step": 3260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.923685195583006e-05,
+      "loss": 0.0757,
+      "step": 3262
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.923638405390231e-05,
+      "loss": 0.1054,
+      "step": 3264
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.923591615197455e-05,
+      "loss": 0.0773,
+      "step": 3266
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9235448250046793e-05,
+      "loss": 0.1131,
+      "step": 3268
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.923498034811904e-05,
+      "loss": 0.0962,
+      "step": 3270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9234512446191285e-05,
+      "loss": 0.1024,
+      "step": 3272
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9234044544263524e-05,
+      "loss": 0.0872,
+      "step": 3274
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.923357664233577e-05,
+      "loss": 0.0985,
+      "step": 3276
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.923310874040801e-05,
+      "loss": 0.0879,
+      "step": 3278
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.923264083848026e-05,
+      "loss": 0.1048,
+      "step": 3280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.92321729365525e-05,
+      "loss": 0.0863,
+      "step": 3282
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.923170503462475e-05,
+      "loss": 0.0783,
+      "step": 3284
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9231237132696986e-05,
+      "loss": 0.0929,
+      "step": 3286
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.923076923076924e-05,
+      "loss": 0.0823,
+      "step": 3288
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.923030132884148e-05,
+      "loss": 0.0793,
+      "step": 3290
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.922983342691372e-05,
+      "loss": 0.1179,
+      "step": 3292
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.922936552498596e-05,
+      "loss": 0.0903,
+      "step": 3294
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.922889762305821e-05,
+      "loss": 0.0917,
+      "step": 3296
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9228429721130455e-05,
+      "loss": 0.0788,
+      "step": 3298
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9227961819202694e-05,
+      "loss": 0.076,
+      "step": 3300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.922749391727494e-05,
+      "loss": 0.0905,
+      "step": 3302
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9227026015347186e-05,
+      "loss": 0.0986,
+      "step": 3304
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.922655811341943e-05,
+      "loss": 0.0629,
+      "step": 3306
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.922609021149167e-05,
+      "loss": 0.0851,
+      "step": 3308
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.922562230956392e-05,
+      "loss": 0.0812,
+      "step": 3310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9225154407636156e-05,
+      "loss": 0.0563,
+      "step": 3312
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.922468650570841e-05,
+      "loss": 0.1,
+      "step": 3314
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.922421860378065e-05,
+      "loss": 0.0857,
+      "step": 3316
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9223750701852894e-05,
+      "loss": 0.079,
+      "step": 3318
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.922328279992513e-05,
+      "loss": 0.0901,
+      "step": 3320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9222814897997386e-05,
+      "loss": 0.082,
+      "step": 3322
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9222346996069625e-05,
+      "loss": 0.0875,
+      "step": 3324
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.922187909414187e-05,
+      "loss": 0.0915,
+      "step": 3326
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.922141119221411e-05,
+      "loss": 0.0988,
+      "step": 3328
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9220943290286356e-05,
+      "loss": 0.0775,
+      "step": 3330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.92204753883586e-05,
+      "loss": 0.077,
+      "step": 3332
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.922000748643085e-05,
+      "loss": 0.0846,
+      "step": 3334
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.921953958450309e-05,
+      "loss": 0.0951,
+      "step": 3336
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.921907168257533e-05,
+      "loss": 0.0938,
+      "step": 3338
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.921860378064758e-05,
+      "loss": 0.0741,
+      "step": 3340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9218135878719824e-05,
+      "loss": 0.092,
+      "step": 3342
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9217667976792064e-05,
+      "loss": 0.0932,
+      "step": 3344
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.921720007486431e-05,
+      "loss": 0.0955,
+      "step": 3346
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9216732172936555e-05,
+      "loss": 0.1082,
+      "step": 3348
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.92162642710088e-05,
+      "loss": 0.0878,
+      "step": 3350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.921579636908104e-05,
+      "loss": 0.0945,
+      "step": 3352
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9215328467153286e-05,
+      "loss": 0.0896,
+      "step": 3354
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.921486056522553e-05,
+      "loss": 0.0855,
+      "step": 3356
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.921439266329778e-05,
+      "loss": 0.1167,
+      "step": 3358
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.921392476137002e-05,
+      "loss": 0.104,
+      "step": 3360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.921345685944226e-05,
+      "loss": 0.0772,
+      "step": 3362
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.92129889575145e-05,
+      "loss": 0.0921,
+      "step": 3364
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9212521055586755e-05,
+      "loss": 0.0733,
+      "step": 3366
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9212053153658994e-05,
+      "loss": 0.105,
+      "step": 3368
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.921158525173124e-05,
+      "loss": 0.0738,
+      "step": 3370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.921111734980348e-05,
+      "loss": 0.1045,
+      "step": 3372
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.921064944787573e-05,
+      "loss": 0.0756,
+      "step": 3374
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.921018154594797e-05,
+      "loss": 0.0824,
+      "step": 3376
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.920971364402022e-05,
+      "loss": 0.0694,
+      "step": 3378
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9209245742092456e-05,
+      "loss": 0.0884,
+      "step": 3380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.92087778401647e-05,
+      "loss": 0.0801,
+      "step": 3382
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.920830993823695e-05,
+      "loss": 0.0818,
+      "step": 3384
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9207842036309194e-05,
+      "loss": 0.102,
+      "step": 3386
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.920737413438143e-05,
+      "loss": 0.0931,
+      "step": 3388
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.920690623245368e-05,
+      "loss": 0.0782,
+      "step": 3390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9206438330525925e-05,
+      "loss": 0.0823,
+      "step": 3392
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.920597042859817e-05,
+      "loss": 0.0899,
+      "step": 3394
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.920550252667041e-05,
+      "loss": 0.0896,
+      "step": 3396
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9205034624742656e-05,
+      "loss": 0.0832,
+      "step": 3398
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.92045667228149e-05,
+      "loss": 0.0772,
+      "step": 3400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.920409882088715e-05,
+      "loss": 0.0912,
+      "step": 3402
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.920363091895939e-05,
+      "loss": 0.0745,
+      "step": 3404
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.920316301703163e-05,
+      "loss": 0.0826,
+      "step": 3406
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.920269511510387e-05,
+      "loss": 0.0785,
+      "step": 3408
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9202227213176124e-05,
+      "loss": 0.0764,
+      "step": 3410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9201759311248364e-05,
+      "loss": 0.0993,
+      "step": 3412
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.920129140932061e-05,
+      "loss": 0.0614,
+      "step": 3414
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.920082350739285e-05,
+      "loss": 0.0791,
+      "step": 3416
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.92003556054651e-05,
+      "loss": 0.0739,
+      "step": 3418
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.919988770353734e-05,
+      "loss": 0.0815,
+      "step": 3420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9199419801609586e-05,
+      "loss": 0.0853,
+      "step": 3422
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9198951899681825e-05,
+      "loss": 0.0914,
+      "step": 3424
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.919848399775407e-05,
+      "loss": 0.099,
+      "step": 3426
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.919801609582632e-05,
+      "loss": 0.0893,
+      "step": 3428
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.919754819389856e-05,
+      "loss": 0.0865,
+      "step": 3430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.91970802919708e-05,
+      "loss": 0.0925,
+      "step": 3432
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.919661239004305e-05,
+      "loss": 0.088,
+      "step": 3434
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9196144488115294e-05,
+      "loss": 0.1066,
+      "step": 3436
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.919567658618754e-05,
+      "loss": 0.0967,
+      "step": 3438
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.919520868425978e-05,
+      "loss": 0.1086,
+      "step": 3440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9194740782332025e-05,
+      "loss": 0.078,
+      "step": 3442
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.919427288040427e-05,
+      "loss": 0.0949,
+      "step": 3444
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.919380497847652e-05,
+      "loss": 0.0922,
+      "step": 3446
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9193337076548756e-05,
+      "loss": 0.0815,
+      "step": 3448
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9192869174621e-05,
+      "loss": 0.0727,
+      "step": 3450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.919240127269325e-05,
+      "loss": 0.0812,
+      "step": 3452
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9191933370765494e-05,
+      "loss": 0.0907,
+      "step": 3454
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.919146546883773e-05,
+      "loss": 0.0951,
+      "step": 3456
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.919099756690998e-05,
+      "loss": 0.0956,
+      "step": 3458
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.919052966498222e-05,
+      "loss": 0.0864,
+      "step": 3460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.919006176305447e-05,
+      "loss": 0.0829,
+      "step": 3462
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.918959386112671e-05,
+      "loss": 0.0738,
+      "step": 3464
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9189125959198956e-05,
+      "loss": 0.0959,
+      "step": 3466
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9188658057271195e-05,
+      "loss": 0.0988,
+      "step": 3468
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.918819015534345e-05,
+      "loss": 0.087,
+      "step": 3470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.918772225341569e-05,
+      "loss": 0.0897,
+      "step": 3472
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.918725435148793e-05,
+      "loss": 0.1108,
+      "step": 3474
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.918678644956017e-05,
+      "loss": 0.0892,
+      "step": 3476
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.918631854763242e-05,
+      "loss": 0.0837,
+      "step": 3478
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9185850645704664e-05,
+      "loss": 0.0924,
+      "step": 3480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.918538274377691e-05,
+      "loss": 0.0895,
+      "step": 3482
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.918491484184915e-05,
+      "loss": 0.0828,
+      "step": 3484
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9184446939921395e-05,
+      "loss": 0.1,
+      "step": 3486
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.918397903799364e-05,
+      "loss": 0.0853,
+      "step": 3488
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9183511136065886e-05,
+      "loss": 0.1041,
+      "step": 3490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9183043234138125e-05,
+      "loss": 0.0838,
+      "step": 3492
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.918257533221037e-05,
+      "loss": 0.084,
+      "step": 3494
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.918210743028262e-05,
+      "loss": 0.0974,
+      "step": 3496
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.918163952835486e-05,
+      "loss": 0.0789,
+      "step": 3498
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.91811716264271e-05,
+      "loss": 0.088,
+      "step": 3500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.918070372449935e-05,
+      "loss": 0.0601,
+      "step": 3502
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9180235822571594e-05,
+      "loss": 0.1086,
+      "step": 3504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.917976792064384e-05,
+      "loss": 0.0887,
+      "step": 3506
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.917930001871608e-05,
+      "loss": 0.0916,
+      "step": 3508
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9178832116788325e-05,
+      "loss": 0.0867,
+      "step": 3510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9178364214860564e-05,
+      "loss": 0.0938,
+      "step": 3512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.917789631293282e-05,
+      "loss": 0.0803,
+      "step": 3514
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9177428411005056e-05,
+      "loss": 0.0915,
+      "step": 3516
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.91769605090773e-05,
+      "loss": 0.0574,
+      "step": 3518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.917649260714954e-05,
+      "loss": 0.0889,
+      "step": 3520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.917602470522179e-05,
+      "loss": 0.0987,
+      "step": 3522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.917555680329403e-05,
+      "loss": 0.0592,
+      "step": 3524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.917508890136628e-05,
+      "loss": 0.1026,
+      "step": 3526
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.917462099943852e-05,
+      "loss": 0.1007,
+      "step": 3528
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9174153097510764e-05,
+      "loss": 0.0932,
+      "step": 3530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.917368519558301e-05,
+      "loss": 0.1059,
+      "step": 3532
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9173217293655256e-05,
+      "loss": 0.0822,
+      "step": 3534
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9172749391727495e-05,
+      "loss": 0.0979,
+      "step": 3536
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.917228148979974e-05,
+      "loss": 0.1007,
+      "step": 3538
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.917181358787199e-05,
+      "loss": 0.0944,
+      "step": 3540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.917134568594423e-05,
+      "loss": 0.102,
+      "step": 3542
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.917087778401647e-05,
+      "loss": 0.0743,
+      "step": 3544
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.917040988208871e-05,
+      "loss": 0.0897,
+      "step": 3546
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9169941980160964e-05,
+      "loss": 0.1136,
+      "step": 3548
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.91694740782332e-05,
+      "loss": 0.0874,
+      "step": 3550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.916900617630545e-05,
+      "loss": 0.079,
+      "step": 3552
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.916853827437769e-05,
+      "loss": 0.0844,
+      "step": 3554
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9168070372449934e-05,
+      "loss": 0.1033,
+      "step": 3556
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.916760247052218e-05,
+      "loss": 0.1093,
+      "step": 3558
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9167134568594425e-05,
+      "loss": 0.0963,
+      "step": 3560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9166666666666665e-05,
+      "loss": 0.101,
+      "step": 3562
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.916619876473891e-05,
+      "loss": 0.0802,
+      "step": 3564
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9165730862811156e-05,
+      "loss": 0.0968,
+      "step": 3566
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.91652629608834e-05,
+      "loss": 0.0982,
+      "step": 3568
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.916479505895564e-05,
+      "loss": 0.0885,
+      "step": 3570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.916432715702789e-05,
+      "loss": 0.0921,
+      "step": 3572
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.916385925510013e-05,
+      "loss": 0.0744,
+      "step": 3574
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.916339135317238e-05,
+      "loss": 0.0824,
+      "step": 3576
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.916292345124462e-05,
+      "loss": 0.1129,
+      "step": 3578
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9162455549316864e-05,
+      "loss": 0.0937,
+      "step": 3580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.916198764738911e-05,
+      "loss": 0.0924,
+      "step": 3582
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9161519745461356e-05,
+      "loss": 0.0828,
+      "step": 3584
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9161051843533595e-05,
+      "loss": 0.0917,
+      "step": 3586
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.916058394160584e-05,
+      "loss": 0.0766,
+      "step": 3588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.916011603967808e-05,
+      "loss": 0.0963,
+      "step": 3590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.915964813775033e-05,
+      "loss": 0.0912,
+      "step": 3592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.915918023582257e-05,
+      "loss": 0.0903,
+      "step": 3594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.915871233389482e-05,
+      "loss": 0.0843,
+      "step": 3596
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.915824443196706e-05,
+      "loss": 0.0897,
+      "step": 3598
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.915777653003931e-05,
+      "loss": 0.0859,
+      "step": 3600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.915730862811155e-05,
+      "loss": 0.0991,
+      "step": 3602
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9156840726183795e-05,
+      "loss": 0.1032,
+      "step": 3604
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9156372824256034e-05,
+      "loss": 0.0866,
+      "step": 3606
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.915590492232828e-05,
+      "loss": 0.0994,
+      "step": 3608
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9155437020400526e-05,
+      "loss": 0.0831,
+      "step": 3610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.915496911847277e-05,
+      "loss": 0.1023,
+      "step": 3612
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.915450121654501e-05,
+      "loss": 0.1046,
+      "step": 3614
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.915403331461726e-05,
+      "loss": 0.0863,
+      "step": 3616
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.91535654126895e-05,
+      "loss": 0.0761,
+      "step": 3618
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.915309751076175e-05,
+      "loss": 0.1039,
+      "step": 3620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.915262960883399e-05,
+      "loss": 0.0741,
+      "step": 3622
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9152161706906234e-05,
+      "loss": 0.0682,
+      "step": 3624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.915169380497848e-05,
+      "loss": 0.0898,
+      "step": 3626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9151225903050725e-05,
+      "loss": 0.1086,
+      "step": 3628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9150758001122965e-05,
+      "loss": 0.1067,
+      "step": 3630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.915029009919521e-05,
+      "loss": 0.0946,
+      "step": 3632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9149822197267456e-05,
+      "loss": 0.0644,
+      "step": 3634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.91493542953397e-05,
+      "loss": 0.0866,
+      "step": 3636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.914888639341194e-05,
+      "loss": 0.1014,
+      "step": 3638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.914841849148419e-05,
+      "loss": 0.0865,
+      "step": 3640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9147950589556427e-05,
+      "loss": 0.0764,
+      "step": 3642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.914748268762868e-05,
+      "loss": 0.0872,
+      "step": 3644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.914701478570092e-05,
+      "loss": 0.1115,
+      "step": 3646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9146546883773164e-05,
+      "loss": 0.0901,
+      "step": 3648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9146078981845403e-05,
+      "loss": 0.0947,
+      "step": 3650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9145611079917656e-05,
+      "loss": 0.0965,
+      "step": 3652
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9145143177989895e-05,
+      "loss": 0.0938,
+      "step": 3654
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.914467527606214e-05,
+      "loss": 0.1238,
+      "step": 3656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.914420737413438e-05,
+      "loss": 0.097,
+      "step": 3658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9143739472206626e-05,
+      "loss": 0.098,
+      "step": 3660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.914327157027887e-05,
+      "loss": 0.0833,
+      "step": 3662
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.914280366835112e-05,
+      "loss": 0.0798,
+      "step": 3664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.914233576642336e-05,
+      "loss": 0.0802,
+      "step": 3666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.91418678644956e-05,
+      "loss": 0.0726,
+      "step": 3668
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.914139996256785e-05,
+      "loss": 0.0939,
+      "step": 3670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9140932060640095e-05,
+      "loss": 0.0983,
+      "step": 3672
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9140464158712334e-05,
+      "loss": 0.0997,
+      "step": 3674
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.913999625678458e-05,
+      "loss": 0.0726,
+      "step": 3676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9139528354856826e-05,
+      "loss": 0.0967,
+      "step": 3678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.913906045292907e-05,
+      "loss": 0.0933,
+      "step": 3680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.913859255100131e-05,
+      "loss": 0.0727,
+      "step": 3682
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.913812464907356e-05,
+      "loss": 0.1104,
+      "step": 3684
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.91376567471458e-05,
+      "loss": 0.0989,
+      "step": 3686
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.913718884521805e-05,
+      "loss": 0.0867,
+      "step": 3688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.913672094329029e-05,
+      "loss": 0.0684,
+      "step": 3690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9136253041362534e-05,
+      "loss": 0.0805,
+      "step": 3692
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.913578513943477e-05,
+      "loss": 0.1268,
+      "step": 3694
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9135317237507026e-05,
+      "loss": 0.1143,
+      "step": 3696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9134849335579265e-05,
+      "loss": 0.119,
+      "step": 3698
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.913438143365151e-05,
+      "loss": 0.0887,
+      "step": 3700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.913391353172375e-05,
+      "loss": 0.0864,
+      "step": 3702
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9133445629795996e-05,
+      "loss": 0.088,
+      "step": 3704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.913297772786824e-05,
+      "loss": 0.1035,
+      "step": 3706
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.913250982594049e-05,
+      "loss": 0.1182,
+      "step": 3708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9132041924012727e-05,
+      "loss": 0.0731,
+      "step": 3710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.913157402208497e-05,
+      "loss": 0.0962,
+      "step": 3712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.913110612015722e-05,
+      "loss": 0.1324,
+      "step": 3714
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9130638218229464e-05,
+      "loss": 0.0934,
+      "step": 3716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9130170316301703e-05,
+      "loss": 0.0941,
+      "step": 3718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.912970241437395e-05,
+      "loss": 0.0777,
+      "step": 3720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9129234512446195e-05,
+      "loss": 0.0773,
+      "step": 3722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.912876661051844e-05,
+      "loss": 0.0969,
+      "step": 3724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.912829870859068e-05,
+      "loss": 0.0824,
+      "step": 3726
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9127830806662926e-05,
+      "loss": 0.0849,
+      "step": 3728
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.912736290473517e-05,
+      "loss": 0.0925,
+      "step": 3730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.912689500280742e-05,
+      "loss": 0.1111,
+      "step": 3732
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.912642710087966e-05,
+      "loss": 0.0961,
+      "step": 3734
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.91259591989519e-05,
+      "loss": 0.0834,
+      "step": 3736
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.912549129702414e-05,
+      "loss": 0.0897,
+      "step": 3738
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9125023395096395e-05,
+      "loss": 0.0979,
+      "step": 3740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9124555493168634e-05,
+      "loss": 0.1118,
+      "step": 3742
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.912408759124088e-05,
+      "loss": 0.0778,
+      "step": 3744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.912361968931312e-05,
+      "loss": 0.074,
+      "step": 3746
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.912315178738537e-05,
+      "loss": 0.0897,
+      "step": 3748
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.912268388545761e-05,
+      "loss": 0.0941,
+      "step": 3750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.912221598352986e-05,
+      "loss": 0.0834,
+      "step": 3752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9121748081602096e-05,
+      "loss": 0.1081,
+      "step": 3754
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.912128017967434e-05,
+      "loss": 0.0938,
+      "step": 3756
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.912081227774659e-05,
+      "loss": 0.0984,
+      "step": 3758
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9120344375818834e-05,
+      "loss": 0.099,
+      "step": 3760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.911987647389107e-05,
+      "loss": 0.0916,
+      "step": 3762
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.911940857196332e-05,
+      "loss": 0.0924,
+      "step": 3764
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9118940670035565e-05,
+      "loss": 0.0822,
+      "step": 3766
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.911847276810781e-05,
+      "loss": 0.0792,
+      "step": 3768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.911800486618005e-05,
+      "loss": 0.1024,
+      "step": 3770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9117536964252296e-05,
+      "loss": 0.1246,
+      "step": 3772
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.911706906232454e-05,
+      "loss": 0.0895,
+      "step": 3774
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.911660116039679e-05,
+      "loss": 0.0821,
+      "step": 3776
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9116133258469027e-05,
+      "loss": 0.0959,
+      "step": 3778
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.911566535654127e-05,
+      "loss": 0.0906,
+      "step": 3780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.911519745461352e-05,
+      "loss": 0.0815,
+      "step": 3782
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9114729552685764e-05,
+      "loss": 0.1142,
+      "step": 3784
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9114261650758003e-05,
+      "loss": 0.0801,
+      "step": 3786
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.911379374883025e-05,
+      "loss": 0.07,
+      "step": 3788
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.911332584690249e-05,
+      "loss": 0.0859,
+      "step": 3790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.911285794497474e-05,
+      "loss": 0.084,
+      "step": 3792
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.911239004304698e-05,
+      "loss": 0.1333,
+      "step": 3794
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9111922141119226e-05,
+      "loss": 0.0816,
+      "step": 3796
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9111454239191465e-05,
+      "loss": 0.0963,
+      "step": 3798
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.911098633726371e-05,
+      "loss": 0.1001,
+      "step": 3800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.911051843533596e-05,
+      "loss": 0.0969,
+      "step": 3802
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9110050533408196e-05,
+      "loss": 0.0817,
+      "step": 3804
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.910958263148044e-05,
+      "loss": 0.1269,
+      "step": 3806
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.910911472955269e-05,
+      "loss": 0.1191,
+      "step": 3808
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9108646827624934e-05,
+      "loss": 0.1094,
+      "step": 3810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.910817892569717e-05,
+      "loss": 0.0941,
+      "step": 3812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.910771102376942e-05,
+      "loss": 0.1328,
+      "step": 3814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9107243121841665e-05,
+      "loss": 0.0866,
+      "step": 3816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.910677521991391e-05,
+      "loss": 0.1208,
+      "step": 3818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.910630731798615e-05,
+      "loss": 0.0855,
+      "step": 3820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9105839416058396e-05,
+      "loss": 0.0963,
+      "step": 3822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9105371514130635e-05,
+      "loss": 0.0908,
+      "step": 3824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.910490361220289e-05,
+      "loss": 0.0924,
+      "step": 3826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.910443571027513e-05,
+      "loss": 0.0927,
+      "step": 3828
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.910396780834737e-05,
+      "loss": 0.1203,
+      "step": 3830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.910349990641961e-05,
+      "loss": 0.0944,
+      "step": 3832
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.910303200449186e-05,
+      "loss": 0.0923,
+      "step": 3834
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9102564102564104e-05,
+      "loss": 0.09,
+      "step": 3836
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.910209620063635e-05,
+      "loss": 0.0838,
+      "step": 3838
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.910162829870859e-05,
+      "loss": 0.0979,
+      "step": 3840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9101160396780835e-05,
+      "loss": 0.0925,
+      "step": 3842
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.910069249485308e-05,
+      "loss": 0.0911,
+      "step": 3844
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9100224592925327e-05,
+      "loss": 0.0723,
+      "step": 3846
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9099756690997566e-05,
+      "loss": 0.0795,
+      "step": 3848
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.909928878906981e-05,
+      "loss": 0.0956,
+      "step": 3850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.909882088714206e-05,
+      "loss": 0.0885,
+      "step": 3852
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9098352985214303e-05,
+      "loss": 0.1086,
+      "step": 3854
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.909788508328654e-05,
+      "loss": 0.0614,
+      "step": 3856
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.909741718135879e-05,
+      "loss": 0.1023,
+      "step": 3858
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9096949279431034e-05,
+      "loss": 0.1041,
+      "step": 3860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.909648137750328e-05,
+      "loss": 0.0934,
+      "step": 3862
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.909601347557552e-05,
+      "loss": 0.0782,
+      "step": 3864
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9095545573647765e-05,
+      "loss": 0.1001,
+      "step": 3866
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9095077671720005e-05,
+      "loss": 0.082,
+      "step": 3868
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.909460976979226e-05,
+      "loss": 0.0836,
+      "step": 3870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9094141867864496e-05,
+      "loss": 0.0753,
+      "step": 3872
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.909367396593674e-05,
+      "loss": 0.0935,
+      "step": 3874
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.909320606400898e-05,
+      "loss": 0.093,
+      "step": 3876
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9092738162081234e-05,
+      "loss": 0.0815,
+      "step": 3878
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.909227026015347e-05,
+      "loss": 0.0994,
+      "step": 3880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.909180235822572e-05,
+      "loss": 0.0939,
+      "step": 3882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.909133445629796e-05,
+      "loss": 0.078,
+      "step": 3884
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9090866554370204e-05,
+      "loss": 0.0785,
+      "step": 3886
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.909039865244245e-05,
+      "loss": 0.1125,
+      "step": 3888
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9089930750514696e-05,
+      "loss": 0.0668,
+      "step": 3890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9089462848586935e-05,
+      "loss": 0.1003,
+      "step": 3892
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.908899494665918e-05,
+      "loss": 0.0932,
+      "step": 3894
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.908852704473143e-05,
+      "loss": 0.095,
+      "step": 3896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.908805914280367e-05,
+      "loss": 0.1071,
+      "step": 3898
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.908759124087591e-05,
+      "loss": 0.0721,
+      "step": 3900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.908712333894816e-05,
+      "loss": 0.0861,
+      "step": 3902
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9086655437020404e-05,
+      "loss": 0.0828,
+      "step": 3904
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.908618753509265e-05,
+      "loss": 0.0956,
+      "step": 3906
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.908571963316489e-05,
+      "loss": 0.0569,
+      "step": 3908
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9085251731237135e-05,
+      "loss": 0.1102,
+      "step": 3910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.908478382930938e-05,
+      "loss": 0.0839,
+      "step": 3912
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.908431592738163e-05,
+      "loss": 0.093,
+      "step": 3914
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9083848025453866e-05,
+      "loss": 0.0756,
+      "step": 3916
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.908338012352611e-05,
+      "loss": 0.0863,
+      "step": 3918
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.908291222159835e-05,
+      "loss": 0.0755,
+      "step": 3920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9082444319670603e-05,
+      "loss": 0.0991,
+      "step": 3922
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.908197641774284e-05,
+      "loss": 0.0932,
+      "step": 3924
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.908150851581509e-05,
+      "loss": 0.1011,
+      "step": 3926
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.908104061388733e-05,
+      "loss": 0.1346,
+      "step": 3928
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.908057271195958e-05,
+      "loss": 0.1027,
+      "step": 3930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.908010481003182e-05,
+      "loss": 0.062,
+      "step": 3932
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9079636908104065e-05,
+      "loss": 0.11,
+      "step": 3934
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9079169006176305e-05,
+      "loss": 0.0779,
+      "step": 3936
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.907870110424855e-05,
+      "loss": 0.0715,
+      "step": 3938
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9078233202320796e-05,
+      "loss": 0.0872,
+      "step": 3940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.907776530039304e-05,
+      "loss": 0.0855,
+      "step": 3942
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.907729739846528e-05,
+      "loss": 0.0987,
+      "step": 3944
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.907682949653753e-05,
+      "loss": 0.0932,
+      "step": 3946
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.907636159460977e-05,
+      "loss": 0.0953,
+      "step": 3948
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.907589369268202e-05,
+      "loss": 0.0986,
+      "step": 3950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.907542579075426e-05,
+      "loss": 0.0859,
+      "step": 3952
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9074957888826504e-05,
+      "loss": 0.0915,
+      "step": 3954
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.907448998689875e-05,
+      "loss": 0.0753,
+      "step": 3956
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9074022084970996e-05,
+      "loss": 0.0887,
+      "step": 3958
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9073554183043235e-05,
+      "loss": 0.1035,
+      "step": 3960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.907308628111548e-05,
+      "loss": 0.1388,
+      "step": 3962
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.907261837918773e-05,
+      "loss": 0.0965,
+      "step": 3964
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.907215047725997e-05,
+      "loss": 0.1309,
+      "step": 3966
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.907168257533221e-05,
+      "loss": 0.1106,
+      "step": 3968
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.907121467340446e-05,
+      "loss": 0.0858,
+      "step": 3970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.90707467714767e-05,
+      "loss": 0.0735,
+      "step": 3972
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.907027886954895e-05,
+      "loss": 0.093,
+      "step": 3974
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.906981096762119e-05,
+      "loss": 0.1556,
+      "step": 3976
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9069343065693435e-05,
+      "loss": 0.0656,
+      "step": 3978
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9068875163765674e-05,
+      "loss": 0.0825,
+      "step": 3980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.906840726183792e-05,
+      "loss": 0.0945,
+      "step": 3982
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9067939359910166e-05,
+      "loss": 0.0892,
+      "step": 3984
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.906747145798241e-05,
+      "loss": 0.089,
+      "step": 3986
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.906700355605465e-05,
+      "loss": 0.1058,
+      "step": 3988
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.90665356541269e-05,
+      "loss": 0.0915,
+      "step": 3990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.906606775219914e-05,
+      "loss": 0.0894,
+      "step": 3992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.906559985027139e-05,
+      "loss": 0.0821,
+      "step": 3994
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.906513194834363e-05,
+      "loss": 0.081,
+      "step": 3996
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9064664046415874e-05,
+      "loss": 0.096,
+      "step": 3998
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.906419614448812e-05,
+      "loss": 0.0845,
+      "step": 4000
+    },
+    {
+      "epoch": 0.56,
+      "eval_gen_len": 33.3103,
+      "eval_loss": 1.031044363975525,
+      "eval_meteor": 0.0499,
+      "eval_runtime": 17.245,
+      "eval_samples_per_second": 3.363,
+      "eval_steps_per_second": 0.464,
+      "step": 4000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9063728242560365e-05,
+      "loss": 0.0821,
+      "step": 4002
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9063260340632605e-05,
+      "loss": 0.0703,
+      "step": 4004
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.906279243870485e-05,
+      "loss": 0.0763,
+      "step": 4006
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9062324536777096e-05,
+      "loss": 0.083,
+      "step": 4008
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.906185663484934e-05,
+      "loss": 0.0703,
+      "step": 4010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.906138873292158e-05,
+      "loss": 0.104,
+      "step": 4012
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.906092083099383e-05,
+      "loss": 0.0943,
+      "step": 4014
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9060452929066066e-05,
+      "loss": 0.086,
+      "step": 4016
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.905998502713832e-05,
+      "loss": 0.0977,
+      "step": 4018
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.905951712521056e-05,
+      "loss": 0.1004,
+      "step": 4020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9059049223282804e-05,
+      "loss": 0.1224,
+      "step": 4022
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.905858132135504e-05,
+      "loss": 0.094,
+      "step": 4024
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9058113419427296e-05,
+      "loss": 0.0824,
+      "step": 4026
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9057645517499535e-05,
+      "loss": 0.1008,
+      "step": 4028
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.905717761557178e-05,
+      "loss": 0.1034,
+      "step": 4030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.905670971364402e-05,
+      "loss": 0.1208,
+      "step": 4032
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9056241811716266e-05,
+      "loss": 0.0767,
+      "step": 4034
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.905577390978851e-05,
+      "loss": 0.1051,
+      "step": 4036
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.905530600786076e-05,
+      "loss": 0.1172,
+      "step": 4038
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9054838105933e-05,
+      "loss": 0.1016,
+      "step": 4040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.905437020400524e-05,
+      "loss": 0.09,
+      "step": 4042
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.905390230207749e-05,
+      "loss": 0.0693,
+      "step": 4044
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9053434400149735e-05,
+      "loss": 0.0866,
+      "step": 4046
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9052966498221974e-05,
+      "loss": 0.1154,
+      "step": 4048
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.905249859629422e-05,
+      "loss": 0.1125,
+      "step": 4050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9052030694366466e-05,
+      "loss": 0.0926,
+      "step": 4052
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9051562792438705e-05,
+      "loss": 0.0813,
+      "step": 4054
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.905109489051095e-05,
+      "loss": 0.104,
+      "step": 4056
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.905062698858319e-05,
+      "loss": 0.0821,
+      "step": 4058
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.905015908665544e-05,
+      "loss": 0.0744,
+      "step": 4060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.904969118472768e-05,
+      "loss": 0.1002,
+      "step": 4062
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.904922328279993e-05,
+      "loss": 0.0927,
+      "step": 4064
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.904875538087217e-05,
+      "loss": 0.0597,
+      "step": 4066
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.904828747894441e-05,
+      "loss": 0.0818,
+      "step": 4068
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.904781957701666e-05,
+      "loss": 0.0856,
+      "step": 4070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9047351675088905e-05,
+      "loss": 0.0839,
+      "step": 4072
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9046883773161144e-05,
+      "loss": 0.0904,
+      "step": 4074
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.904641587123339e-05,
+      "loss": 0.1198,
+      "step": 4076
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9045947969305636e-05,
+      "loss": 0.0997,
+      "step": 4078
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.904548006737788e-05,
+      "loss": 0.091,
+      "step": 4080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.904501216545012e-05,
+      "loss": 0.0827,
+      "step": 4082
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9044544263522367e-05,
+      "loss": 0.0801,
+      "step": 4084
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.904407636159461e-05,
+      "loss": 0.099,
+      "step": 4086
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.904360845966686e-05,
+      "loss": 0.1048,
+      "step": 4088
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.90431405577391e-05,
+      "loss": 0.089,
+      "step": 4090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.904267265581134e-05,
+      "loss": 0.0893,
+      "step": 4092
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.904220475388359e-05,
+      "loss": 0.0834,
+      "step": 4094
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9041736851955835e-05,
+      "loss": 0.0814,
+      "step": 4096
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9041268950028074e-05,
+      "loss": 0.0911,
+      "step": 4098
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.904080104810032e-05,
+      "loss": 0.1082,
+      "step": 4100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.904033314617256e-05,
+      "loss": 0.1039,
+      "step": 4102
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903986524424481e-05,
+      "loss": 0.0719,
+      "step": 4104
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903939734231705e-05,
+      "loss": 0.086,
+      "step": 4106
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.90389294403893e-05,
+      "loss": 0.0852,
+      "step": 4108
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9038461538461536e-05,
+      "loss": 0.1041,
+      "step": 4110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903799363653378e-05,
+      "loss": 0.1025,
+      "step": 4112
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903752573460603e-05,
+      "loss": 0.1016,
+      "step": 4114
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9037057832678274e-05,
+      "loss": 0.1134,
+      "step": 4116
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903658993075051e-05,
+      "loss": 0.1048,
+      "step": 4118
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903612202882276e-05,
+      "loss": 0.0856,
+      "step": 4120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9035654126895005e-05,
+      "loss": 0.0737,
+      "step": 4122
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903518622496725e-05,
+      "loss": 0.0931,
+      "step": 4124
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903471832303949e-05,
+      "loss": 0.0875,
+      "step": 4126
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9034250421111736e-05,
+      "loss": 0.0841,
+      "step": 4128
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903378251918398e-05,
+      "loss": 0.1084,
+      "step": 4130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903331461725623e-05,
+      "loss": 0.0762,
+      "step": 4132
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903284671532847e-05,
+      "loss": 0.1076,
+      "step": 4134
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903237881340071e-05,
+      "loss": 0.0912,
+      "step": 4136
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903191091147296e-05,
+      "loss": 0.0866,
+      "step": 4138
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9031443009545205e-05,
+      "loss": 0.092,
+      "step": 4140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9030975107617444e-05,
+      "loss": 0.1047,
+      "step": 4142
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903050720568969e-05,
+      "loss": 0.0932,
+      "step": 4144
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903003930376193e-05,
+      "loss": 0.1197,
+      "step": 4146
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.902957140183418e-05,
+      "loss": 0.1017,
+      "step": 4148
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.902910349990642e-05,
+      "loss": 0.0966,
+      "step": 4150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9028635597978667e-05,
+      "loss": 0.1375,
+      "step": 4152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9028167696050906e-05,
+      "loss": 0.096,
+      "step": 4154
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.902769979412316e-05,
+      "loss": 0.0814,
+      "step": 4156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.90272318921954e-05,
+      "loss": 0.0854,
+      "step": 4158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9026763990267643e-05,
+      "loss": 0.0935,
+      "step": 4160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.902629608833988e-05,
+      "loss": 0.0972,
+      "step": 4162
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.902582818641213e-05,
+      "loss": 0.1046,
+      "step": 4164
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9025360284484374e-05,
+      "loss": 0.1404,
+      "step": 4166
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.902489238255662e-05,
+      "loss": 0.0898,
+      "step": 4168
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.902442448062886e-05,
+      "loss": 0.082,
+      "step": 4170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9023956578701105e-05,
+      "loss": 0.1063,
+      "step": 4172
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.902348867677335e-05,
+      "loss": 0.0762,
+      "step": 4174
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.90230207748456e-05,
+      "loss": 0.1002,
+      "step": 4176
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9022552872917836e-05,
+      "loss": 0.0935,
+      "step": 4178
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.902208497099008e-05,
+      "loss": 0.1117,
+      "step": 4180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.902161706906233e-05,
+      "loss": 0.1026,
+      "step": 4182
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9021149167134574e-05,
+      "loss": 0.0976,
+      "step": 4184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.902068126520681e-05,
+      "loss": 0.1024,
+      "step": 4186
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.902021336327906e-05,
+      "loss": 0.0984,
+      "step": 4188
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9019745461351305e-05,
+      "loss": 0.0916,
+      "step": 4190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.901927755942355e-05,
+      "loss": 0.1315,
+      "step": 4192
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.901880965749579e-05,
+      "loss": 0.0902,
+      "step": 4194
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9018341755568036e-05,
+      "loss": 0.1066,
+      "step": 4196
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9017873853640275e-05,
+      "loss": 0.0916,
+      "step": 4198
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.901740595171253e-05,
+      "loss": 0.1009,
+      "step": 4200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.901693804978477e-05,
+      "loss": 0.099,
+      "step": 4202
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.901647014785701e-05,
+      "loss": 0.1022,
+      "step": 4204
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.901600224592925e-05,
+      "loss": 0.1264,
+      "step": 4206
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9015534344001505e-05,
+      "loss": 0.0855,
+      "step": 4208
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9015066442073744e-05,
+      "loss": 0.0807,
+      "step": 4210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.901459854014599e-05,
+      "loss": 0.098,
+      "step": 4212
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.901413063821823e-05,
+      "loss": 0.0941,
+      "step": 4214
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9013662736290475e-05,
+      "loss": 0.0892,
+      "step": 4216
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.901319483436272e-05,
+      "loss": 0.086,
+      "step": 4218
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9012726932434967e-05,
+      "loss": 0.0854,
+      "step": 4220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9012259030507206e-05,
+      "loss": 0.0884,
+      "step": 4222
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.901179112857945e-05,
+      "loss": 0.14,
+      "step": 4224
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.90113232266517e-05,
+      "loss": 0.1044,
+      "step": 4226
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9010855324723943e-05,
+      "loss": 0.1168,
+      "step": 4228
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.901038742279618e-05,
+      "loss": 0.099,
+      "step": 4230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.900991952086843e-05,
+      "loss": 0.0921,
+      "step": 4232
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9009451618940674e-05,
+      "loss": 0.1101,
+      "step": 4234
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.900898371701292e-05,
+      "loss": 0.0668,
+      "step": 4236
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.900851581508516e-05,
+      "loss": 0.1024,
+      "step": 4238
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9008047913157405e-05,
+      "loss": 0.1071,
+      "step": 4240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.900758001122965e-05,
+      "loss": 0.0846,
+      "step": 4242
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.90071121093019e-05,
+      "loss": 0.0732,
+      "step": 4244
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9006644207374136e-05,
+      "loss": 0.0998,
+      "step": 4246
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.900617630544638e-05,
+      "loss": 0.1033,
+      "step": 4248
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.900570840351862e-05,
+      "loss": 0.0899,
+      "step": 4250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9005240501590874e-05,
+      "loss": 0.09,
+      "step": 4252
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.900477259966311e-05,
+      "loss": 0.1022,
+      "step": 4254
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.900430469773536e-05,
+      "loss": 0.0846,
+      "step": 4256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.90038367958076e-05,
+      "loss": 0.1168,
+      "step": 4258
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9003368893879844e-05,
+      "loss": 0.094,
+      "step": 4260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.900290099195209e-05,
+      "loss": 0.0767,
+      "step": 4262
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9002433090024336e-05,
+      "loss": 0.0952,
+      "step": 4264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9001965188096575e-05,
+      "loss": 0.0958,
+      "step": 4266
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.900149728616882e-05,
+      "loss": 0.0928,
+      "step": 4268
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.900102938424107e-05,
+      "loss": 0.0999,
+      "step": 4270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.900056148231331e-05,
+      "loss": 0.1167,
+      "step": 4272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.900009358038555e-05,
+      "loss": 0.0872,
+      "step": 4274
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.89996256784578e-05,
+      "loss": 0.0778,
+      "step": 4276
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8999157776530044e-05,
+      "loss": 0.097,
+      "step": 4278
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.899868987460229e-05,
+      "loss": 0.0768,
+      "step": 4280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.899822197267453e-05,
+      "loss": 0.0844,
+      "step": 4282
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8997754070746775e-05,
+      "loss": 0.1015,
+      "step": 4284
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.899728616881902e-05,
+      "loss": 0.0848,
+      "step": 4286
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8996818266891267e-05,
+      "loss": 0.0916,
+      "step": 4288
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8996350364963506e-05,
+      "loss": 0.091,
+      "step": 4290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.899588246303575e-05,
+      "loss": 0.085,
+      "step": 4292
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.899541456110799e-05,
+      "loss": 0.0828,
+      "step": 4294
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8994946659180243e-05,
+      "loss": 0.0947,
+      "step": 4296
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.899447875725248e-05,
+      "loss": 0.1177,
+      "step": 4298
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.899401085532473e-05,
+      "loss": 0.0947,
+      "step": 4300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.899354295339697e-05,
+      "loss": 0.0855,
+      "step": 4302
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.899307505146922e-05,
+      "loss": 0.0957,
+      "step": 4304
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.899260714954146e-05,
+      "loss": 0.0863,
+      "step": 4306
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.89921392476137e-05,
+      "loss": 0.1037,
+      "step": 4308
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8991671345685944e-05,
+      "loss": 0.1227,
+      "step": 4310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.899120344375819e-05,
+      "loss": 0.0743,
+      "step": 4312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8990735541830436e-05,
+      "loss": 0.1011,
+      "step": 4314
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8990267639902675e-05,
+      "loss": 0.0948,
+      "step": 4316
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.898979973797492e-05,
+      "loss": 0.0914,
+      "step": 4318
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.898933183604717e-05,
+      "loss": 0.0866,
+      "step": 4320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.898886393411941e-05,
+      "loss": 0.0906,
+      "step": 4322
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.898839603219165e-05,
+      "loss": 0.0916,
+      "step": 4324
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.89879281302639e-05,
+      "loss": 0.0887,
+      "step": 4326
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.898746022833614e-05,
+      "loss": 0.0896,
+      "step": 4328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.898699232640839e-05,
+      "loss": 0.1005,
+      "step": 4330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.898652442448063e-05,
+      "loss": 0.0842,
+      "step": 4332
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8986056522552875e-05,
+      "loss": 0.0873,
+      "step": 4334
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8985588620625114e-05,
+      "loss": 0.1558,
+      "step": 4336
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.898512071869737e-05,
+      "loss": 0.0988,
+      "step": 4338
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8984652816769606e-05,
+      "loss": 0.0885,
+      "step": 4340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.898418491484185e-05,
+      "loss": 0.0964,
+      "step": 4342
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.898371701291409e-05,
+      "loss": 0.0886,
+      "step": 4344
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.898324911098634e-05,
+      "loss": 0.0924,
+      "step": 4346
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.898278120905858e-05,
+      "loss": 0.1097,
+      "step": 4348
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.898231330713083e-05,
+      "loss": 0.0865,
+      "step": 4350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.898184540520307e-05,
+      "loss": 0.1014,
+      "step": 4352
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8981377503275314e-05,
+      "loss": 0.1189,
+      "step": 4354
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.898090960134756e-05,
+      "loss": 0.0735,
+      "step": 4356
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8980441699419806e-05,
+      "loss": 0.0957,
+      "step": 4358
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8979973797492045e-05,
+      "loss": 0.1077,
+      "step": 4360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.897950589556429e-05,
+      "loss": 0.0805,
+      "step": 4362
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.897903799363654e-05,
+      "loss": 0.0992,
+      "step": 4364
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.897857009170878e-05,
+      "loss": 0.1078,
+      "step": 4366
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.897810218978102e-05,
+      "loss": 0.0871,
+      "step": 4368
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.897763428785327e-05,
+      "loss": 0.0985,
+      "step": 4370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8977166385925514e-05,
+      "loss": 0.0938,
+      "step": 4372
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.897669848399776e-05,
+      "loss": 0.1007,
+      "step": 4374
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.897623058207e-05,
+      "loss": 0.0793,
+      "step": 4376
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8975762680142244e-05,
+      "loss": 0.0948,
+      "step": 4378
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8975294778214484e-05,
+      "loss": 0.0967,
+      "step": 4380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8974826876286736e-05,
+      "loss": 0.0931,
+      "step": 4382
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8974358974358975e-05,
+      "loss": 0.0887,
+      "step": 4384
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.897389107243122e-05,
+      "loss": 0.0804,
+      "step": 4386
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.897342317050346e-05,
+      "loss": 0.0771,
+      "step": 4388
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.897295526857571e-05,
+      "loss": 0.0911,
+      "step": 4390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.897248736664795e-05,
+      "loss": 0.0988,
+      "step": 4392
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.89720194647202e-05,
+      "loss": 0.0744,
+      "step": 4394
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.897155156279244e-05,
+      "loss": 0.0785,
+      "step": 4396
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.897108366086468e-05,
+      "loss": 0.0776,
+      "step": 4398
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.897061575893693e-05,
+      "loss": 0.0814,
+      "step": 4400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8970147857009175e-05,
+      "loss": 0.097,
+      "step": 4402
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8969679955081414e-05,
+      "loss": 0.0763,
+      "step": 4404
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.896921205315366e-05,
+      "loss": 0.1057,
+      "step": 4406
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8968744151225906e-05,
+      "loss": 0.0981,
+      "step": 4408
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.896827624929815e-05,
+      "loss": 0.0961,
+      "step": 4410
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.896780834737039e-05,
+      "loss": 0.0917,
+      "step": 4412
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.896734044544264e-05,
+      "loss": 0.0941,
+      "step": 4414
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.896687254351488e-05,
+      "loss": 0.0752,
+      "step": 4416
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.896640464158713e-05,
+      "loss": 0.1232,
+      "step": 4418
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.896593673965937e-05,
+      "loss": 0.0823,
+      "step": 4420
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8965468837731614e-05,
+      "loss": 0.0992,
+      "step": 4422
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.896500093580385e-05,
+      "loss": 0.0677,
+      "step": 4424
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8964533033876106e-05,
+      "loss": 0.0863,
+      "step": 4426
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8964065131948345e-05,
+      "loss": 0.1048,
+      "step": 4428
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.896359723002059e-05,
+      "loss": 0.0841,
+      "step": 4430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.896312932809283e-05,
+      "loss": 0.1079,
+      "step": 4432
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.896266142616508e-05,
+      "loss": 0.0667,
+      "step": 4434
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.896219352423732e-05,
+      "loss": 0.0898,
+      "step": 4436
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.896172562230957e-05,
+      "loss": 0.1009,
+      "step": 4438
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.896125772038181e-05,
+      "loss": 0.0881,
+      "step": 4440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.896078981845405e-05,
+      "loss": 0.0873,
+      "step": 4442
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.89603219165263e-05,
+      "loss": 0.1066,
+      "step": 4444
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8959854014598545e-05,
+      "loss": 0.1005,
+      "step": 4446
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8959386112670784e-05,
+      "loss": 0.0918,
+      "step": 4448
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.895891821074303e-05,
+      "loss": 0.0812,
+      "step": 4450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8958450308815275e-05,
+      "loss": 0.0732,
+      "step": 4452
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.895798240688752e-05,
+      "loss": 0.1159,
+      "step": 4454
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.895751450495976e-05,
+      "loss": 0.1127,
+      "step": 4456
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8957046603032006e-05,
+      "loss": 0.0782,
+      "step": 4458
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.895657870110425e-05,
+      "loss": 0.1067,
+      "step": 4460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.89561107991765e-05,
+      "loss": 0.078,
+      "step": 4462
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.895564289724874e-05,
+      "loss": 0.1023,
+      "step": 4464
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.895517499532098e-05,
+      "loss": 0.0884,
+      "step": 4466
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.895470709339323e-05,
+      "loss": 0.0833,
+      "step": 4468
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8954239191465475e-05,
+      "loss": 0.0966,
+      "step": 4470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8953771289537714e-05,
+      "loss": 0.0709,
+      "step": 4472
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.895330338760996e-05,
+      "loss": 0.0861,
+      "step": 4474
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.89528354856822e-05,
+      "loss": 0.1094,
+      "step": 4476
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.895236758375445e-05,
+      "loss": 0.1049,
+      "step": 4478
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.895189968182669e-05,
+      "loss": 0.0909,
+      "step": 4480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.895143177989894e-05,
+      "loss": 0.0886,
+      "step": 4482
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8950963877971176e-05,
+      "loss": 0.112,
+      "step": 4484
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.895049597604343e-05,
+      "loss": 0.0861,
+      "step": 4486
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.895002807411567e-05,
+      "loss": 0.086,
+      "step": 4488
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8949560172187914e-05,
+      "loss": 0.0954,
+      "step": 4490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.894909227026015e-05,
+      "loss": 0.1179,
+      "step": 4492
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.89486243683324e-05,
+      "loss": 0.0842,
+      "step": 4494
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8948156466404645e-05,
+      "loss": 0.1064,
+      "step": 4496
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.894768856447689e-05,
+      "loss": 0.1053,
+      "step": 4498
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.894722066254913e-05,
+      "loss": 0.0951,
+      "step": 4500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8946752760621376e-05,
+      "loss": 0.1047,
+      "step": 4502
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.894628485869362e-05,
+      "loss": 0.0911,
+      "step": 4504
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.894581695676587e-05,
+      "loss": 0.0806,
+      "step": 4506
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.894534905483811e-05,
+      "loss": 0.0861,
+      "step": 4508
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.894488115291035e-05,
+      "loss": 0.09,
+      "step": 4510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.89444132509826e-05,
+      "loss": 0.0681,
+      "step": 4512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8943945349054845e-05,
+      "loss": 0.1114,
+      "step": 4514
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8943477447127084e-05,
+      "loss": 0.1009,
+      "step": 4516
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.894300954519933e-05,
+      "loss": 0.0783,
+      "step": 4518
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8942541643271575e-05,
+      "loss": 0.0956,
+      "step": 4520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.894207374134382e-05,
+      "loss": 0.0938,
+      "step": 4522
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.894160583941606e-05,
+      "loss": 0.0783,
+      "step": 4524
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8941137937488306e-05,
+      "loss": 0.095,
+      "step": 4526
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8940670035560546e-05,
+      "loss": 0.1184,
+      "step": 4528
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.89402021336328e-05,
+      "loss": 0.0978,
+      "step": 4530
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.893973423170504e-05,
+      "loss": 0.1171,
+      "step": 4532
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.893926632977728e-05,
+      "loss": 0.0834,
+      "step": 4534
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.893879842784952e-05,
+      "loss": 0.1069,
+      "step": 4536
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.893833052592177e-05,
+      "loss": 0.1139,
+      "step": 4538
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8937862623994014e-05,
+      "loss": 0.1091,
+      "step": 4540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.893739472206626e-05,
+      "loss": 0.1117,
+      "step": 4542
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.89369268201385e-05,
+      "loss": 0.1039,
+      "step": 4544
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8936458918210745e-05,
+      "loss": 0.0875,
+      "step": 4546
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.893599101628299e-05,
+      "loss": 0.0898,
+      "step": 4548
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.893552311435524e-05,
+      "loss": 0.0719,
+      "step": 4550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8935055212427476e-05,
+      "loss": 0.1235,
+      "step": 4552
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.893458731049972e-05,
+      "loss": 0.0776,
+      "step": 4554
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.893411940857197e-05,
+      "loss": 0.0873,
+      "step": 4556
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8933651506644214e-05,
+      "loss": 0.0984,
+      "step": 4558
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.893318360471645e-05,
+      "loss": 0.0908,
+      "step": 4560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.893271570278869e-05,
+      "loss": 0.0766,
+      "step": 4562
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8932247800860945e-05,
+      "loss": 0.0991,
+      "step": 4564
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8931779898933184e-05,
+      "loss": 0.0929,
+      "step": 4566
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.893131199700543e-05,
+      "loss": 0.0673,
+      "step": 4568
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.893084409507767e-05,
+      "loss": 0.0776,
+      "step": 4570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8930376193149915e-05,
+      "loss": 0.0835,
+      "step": 4572
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.892990829122216e-05,
+      "loss": 0.1216,
+      "step": 4574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.892944038929441e-05,
+      "loss": 0.0918,
+      "step": 4576
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8928972487366646e-05,
+      "loss": 0.1165,
+      "step": 4578
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.892850458543889e-05,
+      "loss": 0.0949,
+      "step": 4580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.892803668351114e-05,
+      "loss": 0.106,
+      "step": 4582
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8927568781583384e-05,
+      "loss": 0.0911,
+      "step": 4584
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.892710087965562e-05,
+      "loss": 0.0676,
+      "step": 4586
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.892663297772787e-05,
+      "loss": 0.0986,
+      "step": 4588
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8926165075800115e-05,
+      "loss": 0.1082,
+      "step": 4590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.892569717387236e-05,
+      "loss": 0.104,
+      "step": 4592
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.89252292719446e-05,
+      "loss": 0.0883,
+      "step": 4594
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8924761370016846e-05,
+      "loss": 0.1203,
+      "step": 4596
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.892429346808909e-05,
+      "loss": 0.1259,
+      "step": 4598
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.892382556616134e-05,
+      "loss": 0.088,
+      "step": 4600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8923357664233577e-05,
+      "loss": 0.0992,
+      "step": 4602
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.892288976230582e-05,
+      "loss": 0.0931,
+      "step": 4604
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.892242186037806e-05,
+      "loss": 0.112,
+      "step": 4606
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8921953958450314e-05,
+      "loss": 0.1198,
+      "step": 4608
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8921486056522553e-05,
+      "loss": 0.0877,
+      "step": 4610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.89210181545948e-05,
+      "loss": 0.0897,
+      "step": 4612
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.892055025266704e-05,
+      "loss": 0.093,
+      "step": 4614
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.892008235073929e-05,
+      "loss": 0.1275,
+      "step": 4616
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.891961444881153e-05,
+      "loss": 0.1018,
+      "step": 4618
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8919146546883776e-05,
+      "loss": 0.1001,
+      "step": 4620
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8918678644956015e-05,
+      "loss": 0.0741,
+      "step": 4622
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.891821074302826e-05,
+      "loss": 0.0864,
+      "step": 4624
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.891774284110051e-05,
+      "loss": 0.098,
+      "step": 4626
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.891727493917275e-05,
+      "loss": 0.0803,
+      "step": 4628
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.891680703724499e-05,
+      "loss": 0.0959,
+      "step": 4630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.891633913531724e-05,
+      "loss": 0.0874,
+      "step": 4632
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8915871233389484e-05,
+      "loss": 0.0949,
+      "step": 4634
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.891540333146173e-05,
+      "loss": 0.0782,
+      "step": 4636
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.891493542953397e-05,
+      "loss": 0.0949,
+      "step": 4638
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8914467527606215e-05,
+      "loss": 0.1179,
+      "step": 4640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.891399962567846e-05,
+      "loss": 0.0777,
+      "step": 4642
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.891353172375071e-05,
+      "loss": 0.0822,
+      "step": 4644
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8913063821822946e-05,
+      "loss": 0.1081,
+      "step": 4646
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.891259591989519e-05,
+      "loss": 0.139,
+      "step": 4648
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.891212801796744e-05,
+      "loss": 0.0901,
+      "step": 4650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8911660116039684e-05,
+      "loss": 0.0992,
+      "step": 4652
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.891119221411192e-05,
+      "loss": 0.1104,
+      "step": 4654
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.891072431218417e-05,
+      "loss": 0.1043,
+      "step": 4656
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.891025641025641e-05,
+      "loss": 0.0848,
+      "step": 4658
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.890978850832866e-05,
+      "loss": 0.1208,
+      "step": 4660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.89093206064009e-05,
+      "loss": 0.0921,
+      "step": 4662
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8908852704473146e-05,
+      "loss": 0.1036,
+      "step": 4664
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8908384802545385e-05,
+      "loss": 0.1074,
+      "step": 4666
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.890791690061764e-05,
+      "loss": 0.0866,
+      "step": 4668
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8907448998689877e-05,
+      "loss": 0.077,
+      "step": 4670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.890698109676212e-05,
+      "loss": 0.1032,
+      "step": 4672
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.890651319483436e-05,
+      "loss": 0.0909,
+      "step": 4674
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.890604529290661e-05,
+      "loss": 0.0973,
+      "step": 4676
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8905577390978853e-05,
+      "loss": 0.1134,
+      "step": 4678
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.89051094890511e-05,
+      "loss": 0.0955,
+      "step": 4680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.890464158712334e-05,
+      "loss": 0.1156,
+      "step": 4682
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8904173685195584e-05,
+      "loss": 0.1031,
+      "step": 4684
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.890370578326783e-05,
+      "loss": 0.1189,
+      "step": 4686
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8903237881340076e-05,
+      "loss": 0.0778,
+      "step": 4688
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8902769979412315e-05,
+      "loss": 0.0846,
+      "step": 4690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.890230207748456e-05,
+      "loss": 0.1086,
+      "step": 4692
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.890183417555681e-05,
+      "loss": 0.1226,
+      "step": 4694
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.890136627362905e-05,
+      "loss": 0.1065,
+      "step": 4696
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.890089837170129e-05,
+      "loss": 0.0925,
+      "step": 4698
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.890043046977354e-05,
+      "loss": 0.0848,
+      "step": 4700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.889996256784578e-05,
+      "loss": 0.1009,
+      "step": 4702
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.889949466591803e-05,
+      "loss": 0.0826,
+      "step": 4704
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.889902676399027e-05,
+      "loss": 0.0922,
+      "step": 4706
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8898558862062515e-05,
+      "loss": 0.1119,
+      "step": 4708
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8898090960134754e-05,
+      "loss": 0.0901,
+      "step": 4710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.889762305820701e-05,
+      "loss": 0.1133,
+      "step": 4712
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8897155156279246e-05,
+      "loss": 0.085,
+      "step": 4714
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.889668725435149e-05,
+      "loss": 0.1013,
+      "step": 4716
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.889621935242373e-05,
+      "loss": 0.0785,
+      "step": 4718
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.889575145049598e-05,
+      "loss": 0.0972,
+      "step": 4720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.889528354856822e-05,
+      "loss": 0.0839,
+      "step": 4722
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.889481564664047e-05,
+      "loss": 0.0845,
+      "step": 4724
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.889434774471271e-05,
+      "loss": 0.112,
+      "step": 4726
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8893879842784954e-05,
+      "loss": 0.0875,
+      "step": 4728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.88934119408572e-05,
+      "loss": 0.0809,
+      "step": 4730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8892944038929446e-05,
+      "loss": 0.1014,
+      "step": 4732
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8892476137001685e-05,
+      "loss": 0.1008,
+      "step": 4734
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.889200823507393e-05,
+      "loss": 0.0968,
+      "step": 4736
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8891540333146177e-05,
+      "loss": 0.104,
+      "step": 4738
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.889107243121842e-05,
+      "loss": 0.0895,
+      "step": 4740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.889060452929066e-05,
+      "loss": 0.1038,
+      "step": 4742
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.889013662736291e-05,
+      "loss": 0.0948,
+      "step": 4744
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8889668725435153e-05,
+      "loss": 0.0859,
+      "step": 4746
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.88892008235074e-05,
+      "loss": 0.0925,
+      "step": 4748
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.888873292157964e-05,
+      "loss": 0.0894,
+      "step": 4750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8888265019651884e-05,
+      "loss": 0.1084,
+      "step": 4752
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8887797117724124e-05,
+      "loss": 0.1075,
+      "step": 4754
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8887329215796376e-05,
+      "loss": 0.0874,
+      "step": 4756
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8886861313868615e-05,
+      "loss": 0.1107,
+      "step": 4758
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.888639341194086e-05,
+      "loss": 0.1032,
+      "step": 4760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.88859255100131e-05,
+      "loss": 0.1123,
+      "step": 4762
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.888545760808535e-05,
+      "loss": 0.075,
+      "step": 4764
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.888498970615759e-05,
+      "loss": 0.0872,
+      "step": 4766
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.888452180422984e-05,
+      "loss": 0.1139,
+      "step": 4768
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.888405390230208e-05,
+      "loss": 0.077,
+      "step": 4770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.888358600037432e-05,
+      "loss": 0.1046,
+      "step": 4772
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.888311809844657e-05,
+      "loss": 0.0873,
+      "step": 4774
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8882650196518815e-05,
+      "loss": 0.0995,
+      "step": 4776
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8882182294591054e-05,
+      "loss": 0.0791,
+      "step": 4778
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.88817143926633e-05,
+      "loss": 0.0981,
+      "step": 4780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8881246490735546e-05,
+      "loss": 0.0895,
+      "step": 4782
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.888077858880779e-05,
+      "loss": 0.0972,
+      "step": 4784
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.888031068688003e-05,
+      "loss": 0.1236,
+      "step": 4786
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.887984278495228e-05,
+      "loss": 0.097,
+      "step": 4788
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.887937488302452e-05,
+      "loss": 0.1093,
+      "step": 4790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.887890698109677e-05,
+      "loss": 0.0803,
+      "step": 4792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.887843907916901e-05,
+      "loss": 0.0826,
+      "step": 4794
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8877971177241254e-05,
+      "loss": 0.1125,
+      "step": 4796
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.88775032753135e-05,
+      "loss": 0.105,
+      "step": 4798
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8877035373385746e-05,
+      "loss": 0.0959,
+      "step": 4800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8876567471457985e-05,
+      "loss": 0.0936,
+      "step": 4802
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.887609956953023e-05,
+      "loss": 0.1086,
+      "step": 4804
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.887563166760247e-05,
+      "loss": 0.0829,
+      "step": 4806
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.887516376567472e-05,
+      "loss": 0.1008,
+      "step": 4808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.887469586374696e-05,
+      "loss": 0.0936,
+      "step": 4810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.88742279618192e-05,
+      "loss": 0.0756,
+      "step": 4812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.887376005989145e-05,
+      "loss": 0.0609,
+      "step": 4814
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.887329215796369e-05,
+      "loss": 0.1086,
+      "step": 4816
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.887282425603594e-05,
+      "loss": 0.102,
+      "step": 4818
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.887235635410818e-05,
+      "loss": 0.0932,
+      "step": 4820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8871888452180424e-05,
+      "loss": 0.0812,
+      "step": 4822
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.887142055025267e-05,
+      "loss": 0.0932,
+      "step": 4824
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8870952648324915e-05,
+      "loss": 0.0684,
+      "step": 4826
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8870484746397155e-05,
+      "loss": 0.0867,
+      "step": 4828
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.88700168444694e-05,
+      "loss": 0.1134,
+      "step": 4830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8869548942541646e-05,
+      "loss": 0.0709,
+      "step": 4832
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.886908104061389e-05,
+      "loss": 0.0993,
+      "step": 4834
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.886861313868613e-05,
+      "loss": 0.0925,
+      "step": 4836
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.886814523675838e-05,
+      "loss": 0.074,
+      "step": 4838
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8867677334830616e-05,
+      "loss": 0.0871,
+      "step": 4840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.886720943290287e-05,
+      "loss": 0.1135,
+      "step": 4842
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.886674153097511e-05,
+      "loss": 0.0792,
+      "step": 4844
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8866273629047354e-05,
+      "loss": 0.098,
+      "step": 4846
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.886580572711959e-05,
+      "loss": 0.0822,
+      "step": 4848
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.886533782519184e-05,
+      "loss": 0.1011,
+      "step": 4850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8864869923264085e-05,
+      "loss": 0.0857,
+      "step": 4852
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.886440202133633e-05,
+      "loss": 0.1509,
+      "step": 4854
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.886393411940857e-05,
+      "loss": 0.098,
+      "step": 4856
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8863466217480816e-05,
+      "loss": 0.0951,
+      "step": 4858
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.886299831555306e-05,
+      "loss": 0.0924,
+      "step": 4860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.886253041362531e-05,
+      "loss": 0.0884,
+      "step": 4862
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.886206251169755e-05,
+      "loss": 0.1034,
+      "step": 4864
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.886159460976979e-05,
+      "loss": 0.0821,
+      "step": 4866
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.886112670784204e-05,
+      "loss": 0.1115,
+      "step": 4868
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8860658805914285e-05,
+      "loss": 0.0928,
+      "step": 4870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8860190903986524e-05,
+      "loss": 0.116,
+      "step": 4872
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.885972300205877e-05,
+      "loss": 0.1248,
+      "step": 4874
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8859255100131016e-05,
+      "loss": 0.0932,
+      "step": 4876
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.885878719820326e-05,
+      "loss": 0.0847,
+      "step": 4878
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.88583192962755e-05,
+      "loss": 0.1006,
+      "step": 4880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.885785139434775e-05,
+      "loss": 0.0924,
+      "step": 4882
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8857383492419986e-05,
+      "loss": 0.1081,
+      "step": 4884
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.885691559049224e-05,
+      "loss": 0.1089,
+      "step": 4886
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.885644768856448e-05,
+      "loss": 0.0778,
+      "step": 4888
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8855979786636724e-05,
+      "loss": 0.0876,
+      "step": 4890
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.885551188470896e-05,
+      "loss": 0.0956,
+      "step": 4892
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8855043982781215e-05,
+      "loss": 0.0968,
+      "step": 4894
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8854576080853455e-05,
+      "loss": 0.1055,
+      "step": 4896
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.88541081789257e-05,
+      "loss": 0.0718,
+      "step": 4898
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.885364027699794e-05,
+      "loss": 0.1171,
+      "step": 4900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8853172375070186e-05,
+      "loss": 0.0821,
+      "step": 4902
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.885270447314243e-05,
+      "loss": 0.083,
+      "step": 4904
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.885223657121468e-05,
+      "loss": 0.1034,
+      "step": 4906
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8851768669286916e-05,
+      "loss": 0.1082,
+      "step": 4908
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.885130076735916e-05,
+      "loss": 0.0854,
+      "step": 4910
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.885083286543141e-05,
+      "loss": 0.0951,
+      "step": 4912
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8850364963503654e-05,
+      "loss": 0.1106,
+      "step": 4914
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.884989706157589e-05,
+      "loss": 0.1112,
+      "step": 4916
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.884942915964814e-05,
+      "loss": 0.0852,
+      "step": 4918
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8848961257720385e-05,
+      "loss": 0.0912,
+      "step": 4920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.884849335579263e-05,
+      "loss": 0.0921,
+      "step": 4922
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.884802545386487e-05,
+      "loss": 0.0897,
+      "step": 4924
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8847557551937116e-05,
+      "loss": 0.1344,
+      "step": 4926
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.884708965000936e-05,
+      "loss": 0.0962,
+      "step": 4928
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.884662174808161e-05,
+      "loss": 0.1081,
+      "step": 4930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.884615384615385e-05,
+      "loss": 0.0939,
+      "step": 4932
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.884568594422609e-05,
+      "loss": 0.097,
+      "step": 4934
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.884521804229833e-05,
+      "loss": 0.0804,
+      "step": 4936
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8844750140370585e-05,
+      "loss": 0.1433,
+      "step": 4938
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8844282238442824e-05,
+      "loss": 0.0986,
+      "step": 4940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.884381433651507e-05,
+      "loss": 0.0981,
+      "step": 4942
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.884334643458731e-05,
+      "loss": 0.0859,
+      "step": 4944
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.884287853265956e-05,
+      "loss": 0.1135,
+      "step": 4946
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.88424106307318e-05,
+      "loss": 0.1019,
+      "step": 4948
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.884194272880405e-05,
+      "loss": 0.1109,
+      "step": 4950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8841474826876286e-05,
+      "loss": 0.1208,
+      "step": 4952
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.884100692494853e-05,
+      "loss": 0.0752,
+      "step": 4954
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.884053902302078e-05,
+      "loss": 0.1325,
+      "step": 4956
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8840071121093024e-05,
+      "loss": 0.111,
+      "step": 4958
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.883960321916526e-05,
+      "loss": 0.087,
+      "step": 4960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.883913531723751e-05,
+      "loss": 0.0767,
+      "step": 4962
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8838667415309755e-05,
+      "loss": 0.0867,
+      "step": 4964
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8838199513382e-05,
+      "loss": 0.0836,
+      "step": 4966
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.883773161145424e-05,
+      "loss": 0.084,
+      "step": 4968
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8837263709526486e-05,
+      "loss": 0.1198,
+      "step": 4970
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.883679580759873e-05,
+      "loss": 0.0794,
+      "step": 4972
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.883632790567098e-05,
+      "loss": 0.0856,
+      "step": 4974
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8835860003743216e-05,
+      "loss": 0.0822,
+      "step": 4976
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.883539210181546e-05,
+      "loss": 0.1034,
+      "step": 4978
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.883492419988771e-05,
+      "loss": 0.1047,
+      "step": 4980
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8834456297959954e-05,
+      "loss": 0.1017,
+      "step": 4982
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.883398839603219e-05,
+      "loss": 0.1045,
+      "step": 4984
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.883352049410444e-05,
+      "loss": 0.0984,
+      "step": 4986
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.883305259217668e-05,
+      "loss": 0.0999,
+      "step": 4988
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.883258469024893e-05,
+      "loss": 0.0997,
+      "step": 4990
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.883211678832117e-05,
+      "loss": 0.0962,
+      "step": 4992
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8831648886393416e-05,
+      "loss": 0.0899,
+      "step": 4994
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8831180984465655e-05,
+      "loss": 0.0797,
+      "step": 4996
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.88307130825379e-05,
+      "loss": 0.0936,
+      "step": 4998
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.883024518061015e-05,
+      "loss": 0.0877,
+      "step": 5000
+    },
+    {
+      "epoch": 0.7,
+      "eval_gen_len": 30.7586,
+      "eval_loss": 1.0278133153915405,
+      "eval_meteor": 0.0429,
+      "eval_runtime": 14.877,
+      "eval_samples_per_second": 3.899,
+      "eval_steps_per_second": 0.538,
+      "step": 5000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.882977727868239e-05,
+      "loss": 0.1066,
+      "step": 5002
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.882930937675463e-05,
+      "loss": 0.0858,
+      "step": 5004
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.882884147482688e-05,
+      "loss": 0.092,
+      "step": 5006
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8828373572899124e-05,
+      "loss": 0.09,
+      "step": 5008
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.882790567097137e-05,
+      "loss": 0.1196,
+      "step": 5010
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.882743776904361e-05,
+      "loss": 0.1112,
+      "step": 5012
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8826969867115855e-05,
+      "loss": 0.102,
+      "step": 5014
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.88265019651881e-05,
+      "loss": 0.0905,
+      "step": 5016
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.882603406326035e-05,
+      "loss": 0.1077,
+      "step": 5018
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8825566161332586e-05,
+      "loss": 0.1116,
+      "step": 5020
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.882509825940483e-05,
+      "loss": 0.1122,
+      "step": 5022
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.882463035747708e-05,
+      "loss": 0.0991,
+      "step": 5024
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8824162455549324e-05,
+      "loss": 0.0988,
+      "step": 5026
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.882369455362156e-05,
+      "loss": 0.0812,
+      "step": 5028
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.882322665169381e-05,
+      "loss": 0.0862,
+      "step": 5030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.882275874976605e-05,
+      "loss": 0.09,
+      "step": 5032
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.88222908478383e-05,
+      "loss": 0.0951,
+      "step": 5034
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.882182294591054e-05,
+      "loss": 0.0941,
+      "step": 5036
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8821355043982786e-05,
+      "loss": 0.1444,
+      "step": 5038
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8820887142055025e-05,
+      "loss": 0.1041,
+      "step": 5040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.882041924012728e-05,
+      "loss": 0.1071,
+      "step": 5042
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8819951338199517e-05,
+      "loss": 0.0689,
+      "step": 5044
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.881948343627176e-05,
+      "loss": 0.0874,
+      "step": 5046
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8819015534344e-05,
+      "loss": 0.1425,
+      "step": 5048
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.881854763241625e-05,
+      "loss": 0.0879,
+      "step": 5050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.881807973048849e-05,
+      "loss": 0.0951,
+      "step": 5052
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.881761182856074e-05,
+      "loss": 0.0918,
+      "step": 5054
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.881714392663298e-05,
+      "loss": 0.0966,
+      "step": 5056
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8816676024705224e-05,
+      "loss": 0.0977,
+      "step": 5058
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.881620812277747e-05,
+      "loss": 0.0908,
+      "step": 5060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8815740220849716e-05,
+      "loss": 0.0909,
+      "step": 5062
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8815272318921955e-05,
+      "loss": 0.0929,
+      "step": 5064
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8814804416994194e-05,
+      "loss": 0.0875,
+      "step": 5066
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.881433651506645e-05,
+      "loss": 0.0916,
+      "step": 5068
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8813868613138686e-05,
+      "loss": 0.1227,
+      "step": 5070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.881340071121093e-05,
+      "loss": 0.1054,
+      "step": 5072
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.881293280928317e-05,
+      "loss": 0.0916,
+      "step": 5074
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8812464907355424e-05,
+      "loss": 0.0883,
+      "step": 5076
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.881199700542766e-05,
+      "loss": 0.098,
+      "step": 5078
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.881152910349991e-05,
+      "loss": 0.0925,
+      "step": 5080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.881106120157215e-05,
+      "loss": 0.0824,
+      "step": 5082
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8810593299644394e-05,
+      "loss": 0.0913,
+      "step": 5084
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.881012539771664e-05,
+      "loss": 0.078,
+      "step": 5086
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8809657495788886e-05,
+      "loss": 0.0972,
+      "step": 5088
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8809189593861125e-05,
+      "loss": 0.0802,
+      "step": 5090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.880872169193337e-05,
+      "loss": 0.0854,
+      "step": 5092
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.880825379000562e-05,
+      "loss": 0.0963,
+      "step": 5094
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.880778588807786e-05,
+      "loss": 0.1212,
+      "step": 5096
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.88073179861501e-05,
+      "loss": 0.0881,
+      "step": 5098
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.880685008422235e-05,
+      "loss": 0.1133,
+      "step": 5100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8806382182294594e-05,
+      "loss": 0.0829,
+      "step": 5102
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.880591428036684e-05,
+      "loss": 0.0888,
+      "step": 5104
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.880544637843908e-05,
+      "loss": 0.1003,
+      "step": 5106
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8804978476511325e-05,
+      "loss": 0.0716,
+      "step": 5108
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.880451057458357e-05,
+      "loss": 0.1152,
+      "step": 5110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8804042672655817e-05,
+      "loss": 0.082,
+      "step": 5112
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8803574770728056e-05,
+      "loss": 0.0875,
+      "step": 5114
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.88031068688003e-05,
+      "loss": 0.1069,
+      "step": 5116
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.880263896687254e-05,
+      "loss": 0.0879,
+      "step": 5118
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8802171064944793e-05,
+      "loss": 0.0848,
+      "step": 5120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.880170316301703e-05,
+      "loss": 0.0749,
+      "step": 5122
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.880123526108928e-05,
+      "loss": 0.0772,
+      "step": 5124
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.880076735916152e-05,
+      "loss": 0.0929,
+      "step": 5126
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8800299457233763e-05,
+      "loss": 0.1225,
+      "step": 5128
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.879983155530601e-05,
+      "loss": 0.0864,
+      "step": 5130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8799363653378255e-05,
+      "loss": 0.099,
+      "step": 5132
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8798895751450494e-05,
+      "loss": 0.0949,
+      "step": 5134
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.879842784952274e-05,
+      "loss": 0.0723,
+      "step": 5136
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8797959947594986e-05,
+      "loss": 0.1157,
+      "step": 5138
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.879749204566723e-05,
+      "loss": 0.0795,
+      "step": 5140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.879702414373947e-05,
+      "loss": 0.111,
+      "step": 5142
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.879655624181172e-05,
+      "loss": 0.1169,
+      "step": 5144
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.879608833988396e-05,
+      "loss": 0.0893,
+      "step": 5146
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.879562043795621e-05,
+      "loss": 0.1175,
+      "step": 5148
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.879515253602845e-05,
+      "loss": 0.1116,
+      "step": 5150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8794684634100694e-05,
+      "loss": 0.1105,
+      "step": 5152
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.879421673217294e-05,
+      "loss": 0.1006,
+      "step": 5154
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8793748830245186e-05,
+      "loss": 0.0852,
+      "step": 5156
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8793280928317425e-05,
+      "loss": 0.0972,
+      "step": 5158
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.879281302638967e-05,
+      "loss": 0.1039,
+      "step": 5160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.879234512446191e-05,
+      "loss": 0.1021,
+      "step": 5162
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.879187722253416e-05,
+      "loss": 0.1008,
+      "step": 5164
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.87914093206064e-05,
+      "loss": 0.0823,
+      "step": 5166
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.879094141867865e-05,
+      "loss": 0.0925,
+      "step": 5168
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.879047351675089e-05,
+      "loss": 0.1156,
+      "step": 5170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.879000561482314e-05,
+      "loss": 0.1213,
+      "step": 5172
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.878953771289538e-05,
+      "loss": 0.1397,
+      "step": 5174
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8789069810967625e-05,
+      "loss": 0.0847,
+      "step": 5176
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8788601909039864e-05,
+      "loss": 0.0956,
+      "step": 5178
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.878813400711211e-05,
+      "loss": 0.1024,
+      "step": 5180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8787666105184356e-05,
+      "loss": 0.0907,
+      "step": 5182
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.87871982032566e-05,
+      "loss": 0.0927,
+      "step": 5184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.878673030132884e-05,
+      "loss": 0.0882,
+      "step": 5186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.878626239940109e-05,
+      "loss": 0.0587,
+      "step": 5188
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.878579449747333e-05,
+      "loss": 0.0883,
+      "step": 5190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.878532659554558e-05,
+      "loss": 0.0874,
+      "step": 5192
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.878485869361782e-05,
+      "loss": 0.0828,
+      "step": 5194
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8784390791690064e-05,
+      "loss": 0.064,
+      "step": 5196
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.878392288976231e-05,
+      "loss": 0.0796,
+      "step": 5198
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8783454987834555e-05,
+      "loss": 0.0894,
+      "step": 5200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8782987085906794e-05,
+      "loss": 0.0972,
+      "step": 5202
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.878251918397904e-05,
+      "loss": 0.1182,
+      "step": 5204
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8782051282051286e-05,
+      "loss": 0.0831,
+      "step": 5206
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.878158338012353e-05,
+      "loss": 0.096,
+      "step": 5208
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.878111547819577e-05,
+      "loss": 0.0795,
+      "step": 5210
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.878064757626802e-05,
+      "loss": 0.1084,
+      "step": 5212
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8780179674340256e-05,
+      "loss": 0.0898,
+      "step": 5214
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.877971177241251e-05,
+      "loss": 0.095,
+      "step": 5216
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.877924387048475e-05,
+      "loss": 0.0821,
+      "step": 5218
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8778775968556994e-05,
+      "loss": 0.1048,
+      "step": 5220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.877830806662923e-05,
+      "loss": 0.0875,
+      "step": 5222
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8777840164701486e-05,
+      "loss": 0.0855,
+      "step": 5224
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8777372262773725e-05,
+      "loss": 0.0792,
+      "step": 5226
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.877690436084597e-05,
+      "loss": 0.1225,
+      "step": 5228
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.877643645891821e-05,
+      "loss": 0.1503,
+      "step": 5230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8775968556990456e-05,
+      "loss": 0.0773,
+      "step": 5232
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.87755006550627e-05,
+      "loss": 0.1163,
+      "step": 5234
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.877503275313495e-05,
+      "loss": 0.0889,
+      "step": 5236
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.877456485120719e-05,
+      "loss": 0.1115,
+      "step": 5238
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.877409694927943e-05,
+      "loss": 0.1228,
+      "step": 5240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.877362904735168e-05,
+      "loss": 0.0851,
+      "step": 5242
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8773161145423925e-05,
+      "loss": 0.105,
+      "step": 5244
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8772693243496164e-05,
+      "loss": 0.0942,
+      "step": 5246
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.877222534156841e-05,
+      "loss": 0.0972,
+      "step": 5248
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8771757439640656e-05,
+      "loss": 0.1146,
+      "step": 5250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.87712895377129e-05,
+      "loss": 0.1119,
+      "step": 5252
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.877082163578514e-05,
+      "loss": 0.1169,
+      "step": 5254
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.877035373385739e-05,
+      "loss": 0.1061,
+      "step": 5256
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.876988583192963e-05,
+      "loss": 0.0853,
+      "step": 5258
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.876941793000188e-05,
+      "loss": 0.1577,
+      "step": 5260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.876895002807412e-05,
+      "loss": 0.0861,
+      "step": 5262
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8768482126146364e-05,
+      "loss": 0.0989,
+      "step": 5264
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.87680142242186e-05,
+      "loss": 0.1048,
+      "step": 5266
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8767546322290855e-05,
+      "loss": 0.088,
+      "step": 5268
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8767078420363094e-05,
+      "loss": 0.1005,
+      "step": 5270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.876661051843534e-05,
+      "loss": 0.0915,
+      "step": 5272
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.876614261650758e-05,
+      "loss": 0.0972,
+      "step": 5274
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8765674714579825e-05,
+      "loss": 0.0944,
+      "step": 5276
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.876520681265207e-05,
+      "loss": 0.0973,
+      "step": 5278
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.876473891072432e-05,
+      "loss": 0.0953,
+      "step": 5280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8764271008796556e-05,
+      "loss": 0.0862,
+      "step": 5282
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.87638031068688e-05,
+      "loss": 0.098,
+      "step": 5284
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.876333520494105e-05,
+      "loss": 0.1076,
+      "step": 5286
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8762867303013294e-05,
+      "loss": 0.0939,
+      "step": 5288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.876239940108553e-05,
+      "loss": 0.0821,
+      "step": 5290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.876193149915778e-05,
+      "loss": 0.0927,
+      "step": 5292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8761463597230025e-05,
+      "loss": 0.0995,
+      "step": 5294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.876099569530227e-05,
+      "loss": 0.0986,
+      "step": 5296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.876052779337451e-05,
+      "loss": 0.0788,
+      "step": 5298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8760059891446756e-05,
+      "loss": 0.0971,
+      "step": 5300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8759591989519e-05,
+      "loss": 0.0936,
+      "step": 5302
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.875912408759125e-05,
+      "loss": 0.1128,
+      "step": 5304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.875865618566349e-05,
+      "loss": 0.1094,
+      "step": 5306
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.875818828373573e-05,
+      "loss": 0.1193,
+      "step": 5308
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.875772038180797e-05,
+      "loss": 0.0843,
+      "step": 5310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8757252479880225e-05,
+      "loss": 0.0942,
+      "step": 5312
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8756784577952464e-05,
+      "loss": 0.1054,
+      "step": 5314
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.875631667602471e-05,
+      "loss": 0.1116,
+      "step": 5316
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.875584877409695e-05,
+      "loss": 0.0969,
+      "step": 5318
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8755380872169195e-05,
+      "loss": 0.099,
+      "step": 5320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.875491297024144e-05,
+      "loss": 0.1027,
+      "step": 5322
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.875444506831368e-05,
+      "loss": 0.0683,
+      "step": 5324
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8753977166385926e-05,
+      "loss": 0.1208,
+      "step": 5326
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.875350926445817e-05,
+      "loss": 0.0801,
+      "step": 5328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.875304136253042e-05,
+      "loss": 0.0926,
+      "step": 5330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.875257346060266e-05,
+      "loss": 0.0755,
+      "step": 5332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.87521055586749e-05,
+      "loss": 0.0812,
+      "step": 5334
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.875163765674715e-05,
+      "loss": 0.1073,
+      "step": 5336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8751169754819395e-05,
+      "loss": 0.1058,
+      "step": 5338
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8750701852891634e-05,
+      "loss": 0.0831,
+      "step": 5340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.875023395096388e-05,
+      "loss": 0.1217,
+      "step": 5342
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.874976604903612e-05,
+      "loss": 0.0912,
+      "step": 5344
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.874929814710837e-05,
+      "loss": 0.0806,
+      "step": 5346
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.874883024518061e-05,
+      "loss": 0.0736,
+      "step": 5348
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8748362343252856e-05,
+      "loss": 0.1044,
+      "step": 5350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8747894441325096e-05,
+      "loss": 0.0607,
+      "step": 5352
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.874742653939735e-05,
+      "loss": 0.0997,
+      "step": 5354
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.874695863746959e-05,
+      "loss": 0.1129,
+      "step": 5356
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.874649073554183e-05,
+      "loss": 0.1079,
+      "step": 5358
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.874602283361407e-05,
+      "loss": 0.1107,
+      "step": 5360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.874555493168632e-05,
+      "loss": 0.1014,
+      "step": 5362
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8745087029758564e-05,
+      "loss": 0.0788,
+      "step": 5364
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.874461912783081e-05,
+      "loss": 0.0861,
+      "step": 5366
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.874415122590305e-05,
+      "loss": 0.0941,
+      "step": 5368
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8743683323975295e-05,
+      "loss": 0.1025,
+      "step": 5370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.874321542204754e-05,
+      "loss": 0.0818,
+      "step": 5372
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.874274752011979e-05,
+      "loss": 0.0861,
+      "step": 5374
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8742279618192026e-05,
+      "loss": 0.0949,
+      "step": 5376
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.874181171626427e-05,
+      "loss": 0.0917,
+      "step": 5378
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.874134381433652e-05,
+      "loss": 0.0911,
+      "step": 5380
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8740875912408764e-05,
+      "loss": 0.1092,
+      "step": 5382
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8740408010481e-05,
+      "loss": 0.0843,
+      "step": 5384
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.873994010855325e-05,
+      "loss": 0.1067,
+      "step": 5386
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8739472206625495e-05,
+      "loss": 0.1009,
+      "step": 5388
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.873900430469774e-05,
+      "loss": 0.0771,
+      "step": 5390
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.873853640276998e-05,
+      "loss": 0.0855,
+      "step": 5392
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8738068500842226e-05,
+      "loss": 0.0999,
+      "step": 5394
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8737600598914465e-05,
+      "loss": 0.0723,
+      "step": 5396
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.873713269698672e-05,
+      "loss": 0.0921,
+      "step": 5398
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.873666479505896e-05,
+      "loss": 0.1074,
+      "step": 5400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.87361968931312e-05,
+      "loss": 0.1037,
+      "step": 5402
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.873572899120344e-05,
+      "loss": 0.1177,
+      "step": 5404
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.873526108927569e-05,
+      "loss": 0.0802,
+      "step": 5406
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8734793187347934e-05,
+      "loss": 0.1014,
+      "step": 5408
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.873432528542018e-05,
+      "loss": 0.1004,
+      "step": 5410
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.873385738349242e-05,
+      "loss": 0.0965,
+      "step": 5412
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8733389481564665e-05,
+      "loss": 0.1153,
+      "step": 5414
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.873292157963691e-05,
+      "loss": 0.0954,
+      "step": 5416
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8732453677709156e-05,
+      "loss": 0.1108,
+      "step": 5418
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8731985775781396e-05,
+      "loss": 0.0833,
+      "step": 5420
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.873151787385364e-05,
+      "loss": 0.092,
+      "step": 5422
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.873104997192589e-05,
+      "loss": 0.0987,
+      "step": 5424
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.873058206999813e-05,
+      "loss": 0.1232,
+      "step": 5426
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.873011416807037e-05,
+      "loss": 0.0961,
+      "step": 5428
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.872964626614262e-05,
+      "loss": 0.0977,
+      "step": 5430
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8729178364214864e-05,
+      "loss": 0.1228,
+      "step": 5432
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.872871046228711e-05,
+      "loss": 0.0832,
+      "step": 5434
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.872824256035935e-05,
+      "loss": 0.0934,
+      "step": 5436
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8727774658431595e-05,
+      "loss": 0.1066,
+      "step": 5438
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8727306756503834e-05,
+      "loss": 0.0877,
+      "step": 5440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.872683885457609e-05,
+      "loss": 0.0959,
+      "step": 5442
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8726370952648326e-05,
+      "loss": 0.1042,
+      "step": 5444
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.872590305072057e-05,
+      "loss": 0.1263,
+      "step": 5446
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.872543514879281e-05,
+      "loss": 0.1156,
+      "step": 5448
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8724967246865064e-05,
+      "loss": 0.102,
+      "step": 5450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.87244993449373e-05,
+      "loss": 0.0916,
+      "step": 5452
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.872403144300955e-05,
+      "loss": 0.1195,
+      "step": 5454
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.872356354108179e-05,
+      "loss": 0.0908,
+      "step": 5456
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8723095639154034e-05,
+      "loss": 0.0879,
+      "step": 5458
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.872262773722628e-05,
+      "loss": 0.0976,
+      "step": 5460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8722159835298526e-05,
+      "loss": 0.0783,
+      "step": 5462
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8721691933370765e-05,
+      "loss": 0.1068,
+      "step": 5464
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.872122403144301e-05,
+      "loss": 0.0795,
+      "step": 5466
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.872075612951526e-05,
+      "loss": 0.0846,
+      "step": 5468
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.87202882275875e-05,
+      "loss": 0.0909,
+      "step": 5470
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.871982032565974e-05,
+      "loss": 0.0914,
+      "step": 5472
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.871935242373199e-05,
+      "loss": 0.144,
+      "step": 5474
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8718884521804234e-05,
+      "loss": 0.0967,
+      "step": 5476
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.871841661987648e-05,
+      "loss": 0.077,
+      "step": 5478
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.871794871794872e-05,
+      "loss": 0.0871,
+      "step": 5480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8717480816020965e-05,
+      "loss": 0.1024,
+      "step": 5482
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.871701291409321e-05,
+      "loss": 0.1155,
+      "step": 5484
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8716545012165456e-05,
+      "loss": 0.1163,
+      "step": 5486
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8716077110237696e-05,
+      "loss": 0.1136,
+      "step": 5488
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.871560920830994e-05,
+      "loss": 0.1093,
+      "step": 5490
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.871514130638218e-05,
+      "loss": 0.0954,
+      "step": 5492
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.871467340445443e-05,
+      "loss": 0.0912,
+      "step": 5494
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.871420550252667e-05,
+      "loss": 0.1163,
+      "step": 5496
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.871373760059892e-05,
+      "loss": 0.1029,
+      "step": 5498
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.871326969867116e-05,
+      "loss": 0.0773,
+      "step": 5500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.871280179674341e-05,
+      "loss": 0.0908,
+      "step": 5502
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.871233389481565e-05,
+      "loss": 0.0809,
+      "step": 5504
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8711865992887895e-05,
+      "loss": 0.1013,
+      "step": 5506
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8711398090960134e-05,
+      "loss": 0.0955,
+      "step": 5508
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.871093018903238e-05,
+      "loss": 0.0844,
+      "step": 5510
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8710462287104626e-05,
+      "loss": 0.1063,
+      "step": 5512
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.870999438517687e-05,
+      "loss": 0.0918,
+      "step": 5514
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.870952648324911e-05,
+      "loss": 0.0901,
+      "step": 5516
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.870905858132136e-05,
+      "loss": 0.0831,
+      "step": 5518
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.87085906793936e-05,
+      "loss": 0.0877,
+      "step": 5520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.870812277746585e-05,
+      "loss": 0.1052,
+      "step": 5522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.870765487553809e-05,
+      "loss": 0.0827,
+      "step": 5524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8707186973610334e-05,
+      "loss": 0.1171,
+      "step": 5526
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.870671907168258e-05,
+      "loss": 0.0911,
+      "step": 5528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8706251169754826e-05,
+      "loss": 0.129,
+      "step": 5530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8705783267827065e-05,
+      "loss": 0.0777,
+      "step": 5532
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.870531536589931e-05,
+      "loss": 0.0703,
+      "step": 5534
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.870484746397156e-05,
+      "loss": 0.1071,
+      "step": 5536
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.87043795620438e-05,
+      "loss": 0.0731,
+      "step": 5538
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.870391166011604e-05,
+      "loss": 0.0921,
+      "step": 5540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.870344375818829e-05,
+      "loss": 0.0842,
+      "step": 5542
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.870297585626053e-05,
+      "loss": 0.1026,
+      "step": 5544
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.870250795433278e-05,
+      "loss": 0.0941,
+      "step": 5546
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.870204005240502e-05,
+      "loss": 0.0942,
+      "step": 5548
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8701572150477265e-05,
+      "loss": 0.0914,
+      "step": 5550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8701104248549504e-05,
+      "loss": 0.1367,
+      "step": 5552
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.870063634662175e-05,
+      "loss": 0.0841,
+      "step": 5554
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8700168444693996e-05,
+      "loss": 0.1086,
+      "step": 5556
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.869970054276624e-05,
+      "loss": 0.086,
+      "step": 5558
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.869923264083848e-05,
+      "loss": 0.1065,
+      "step": 5560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8698764738910727e-05,
+      "loss": 0.0951,
+      "step": 5562
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.869829683698297e-05,
+      "loss": 0.0806,
+      "step": 5564
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.869782893505522e-05,
+      "loss": 0.1128,
+      "step": 5566
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.869736103312746e-05,
+      "loss": 0.1105,
+      "step": 5568
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8696893131199703e-05,
+      "loss": 0.0936,
+      "step": 5570
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.869642522927195e-05,
+      "loss": 0.0906,
+      "step": 5572
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.869595732734419e-05,
+      "loss": 0.0979,
+      "step": 5574
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8695489425416434e-05,
+      "loss": 0.1293,
+      "step": 5576
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8695021523488674e-05,
+      "loss": 0.0945,
+      "step": 5578
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8694553621560926e-05,
+      "loss": 0.1071,
+      "step": 5580
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8694085719633165e-05,
+      "loss": 0.0819,
+      "step": 5582
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.869361781770541e-05,
+      "loss": 0.09,
+      "step": 5584
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.869314991577765e-05,
+      "loss": 0.094,
+      "step": 5586
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8692682013849896e-05,
+      "loss": 0.0905,
+      "step": 5588
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.869221411192214e-05,
+      "loss": 0.1133,
+      "step": 5590
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.869174620999439e-05,
+      "loss": 0.1095,
+      "step": 5592
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.869127830806663e-05,
+      "loss": 0.1052,
+      "step": 5594
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.869081040613887e-05,
+      "loss": 0.0891,
+      "step": 5596
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.869034250421112e-05,
+      "loss": 0.1034,
+      "step": 5598
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.8689874602283365e-05,
+      "loss": 0.0836,
+      "step": 5600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.8689406700355604e-05,
+      "loss": 0.0989,
+      "step": 5602
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.868893879842785e-05,
+      "loss": 0.0892,
+      "step": 5604
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.8688470896500096e-05,
+      "loss": 0.0902,
+      "step": 5606
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.868800299457234e-05,
+      "loss": 0.1037,
+      "step": 5608
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.868753509264458e-05,
+      "loss": 0.1106,
+      "step": 5610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.868706719071683e-05,
+      "loss": 0.1128,
+      "step": 5612
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.868659928878907e-05,
+      "loss": 0.1263,
+      "step": 5614
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.868613138686132e-05,
+      "loss": 0.1297,
+      "step": 5616
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.868566348493356e-05,
+      "loss": 0.116,
+      "step": 5618
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.8685195583005804e-05,
+      "loss": 0.1371,
+      "step": 5620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.868472768107804e-05,
+      "loss": 0.1183,
+      "step": 5622
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.8684259779150296e-05,
+      "loss": 0.0773,
+      "step": 5624
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.8683791877222535e-05,
+      "loss": 0.1158,
+      "step": 5626
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.868332397529478e-05,
+      "loss": 0.0743,
+      "step": 5628
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.868285607336702e-05,
+      "loss": 0.0896,
+      "step": 5630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.868238817143927e-05,
+      "loss": 0.1097,
+      "step": 5632
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.868192026951151e-05,
+      "loss": 0.0834,
+      "step": 5634
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.868145236758376e-05,
+      "loss": 0.0808,
+      "step": 5636
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.8680984465656e-05,
+      "loss": 0.0934,
+      "step": 5638
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.868051656372824e-05,
+      "loss": 0.1005,
+      "step": 5640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.868004866180049e-05,
+      "loss": 0.1217,
+      "step": 5642
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.8679580759872734e-05,
+      "loss": 0.0884,
+      "step": 5644
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.8679112857944974e-05,
+      "loss": 0.1019,
+      "step": 5646
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.867864495601722e-05,
+      "loss": 0.0945,
+      "step": 5648
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.8678177054089465e-05,
+      "loss": 0.093,
+      "step": 5650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.867770915216171e-05,
+      "loss": 0.1392,
+      "step": 5652
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.867724125023395e-05,
+      "loss": 0.0792,
+      "step": 5654
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.8676773348306196e-05,
+      "loss": 0.1005,
+      "step": 5656
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.867630544637844e-05,
+      "loss": 0.1035,
+      "step": 5658
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.867583754445069e-05,
+      "loss": 0.085,
+      "step": 5660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.867536964252293e-05,
+      "loss": 0.1051,
+      "step": 5662
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.867490174059517e-05,
+      "loss": 0.0851,
+      "step": 5664
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.867443383866742e-05,
+      "loss": 0.0912,
+      "step": 5666
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8673965936739665e-05,
+      "loss": 0.0964,
+      "step": 5668
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8673498034811904e-05,
+      "loss": 0.0948,
+      "step": 5670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.867303013288415e-05,
+      "loss": 0.1255,
+      "step": 5672
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.867256223095639e-05,
+      "loss": 0.0945,
+      "step": 5674
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.867209432902864e-05,
+      "loss": 0.1198,
+      "step": 5676
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.867162642710088e-05,
+      "loss": 0.0898,
+      "step": 5678
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.867115852517313e-05,
+      "loss": 0.0981,
+      "step": 5680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8670690623245366e-05,
+      "loss": 0.0937,
+      "step": 5682
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.867022272131762e-05,
+      "loss": 0.0942,
+      "step": 5684
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.866975481938986e-05,
+      "loss": 0.0784,
+      "step": 5686
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8669286917462104e-05,
+      "loss": 0.132,
+      "step": 5688
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.866881901553434e-05,
+      "loss": 0.1304,
+      "step": 5690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.866835111360659e-05,
+      "loss": 0.0966,
+      "step": 5692
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8667883211678835e-05,
+      "loss": 0.1298,
+      "step": 5694
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.866741530975108e-05,
+      "loss": 0.0615,
+      "step": 5696
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.866694740782332e-05,
+      "loss": 0.075,
+      "step": 5698
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8666479505895566e-05,
+      "loss": 0.11,
+      "step": 5700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.866601160396781e-05,
+      "loss": 0.0993,
+      "step": 5702
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.866554370204006e-05,
+      "loss": 0.0881,
+      "step": 5704
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.86650758001123e-05,
+      "loss": 0.0967,
+      "step": 5706
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.866460789818454e-05,
+      "loss": 0.1145,
+      "step": 5708
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.866413999625679e-05,
+      "loss": 0.1109,
+      "step": 5710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8663672094329034e-05,
+      "loss": 0.1026,
+      "step": 5712
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8663204192401274e-05,
+      "loss": 0.1205,
+      "step": 5714
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.866273629047352e-05,
+      "loss": 0.0781,
+      "step": 5716
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.866226838854576e-05,
+      "loss": 0.1263,
+      "step": 5718
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.866180048661801e-05,
+      "loss": 0.0782,
+      "step": 5720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.866133258469025e-05,
+      "loss": 0.092,
+      "step": 5722
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8660864682762496e-05,
+      "loss": 0.084,
+      "step": 5724
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8660396780834735e-05,
+      "loss": 0.0845,
+      "step": 5726
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.865992887890699e-05,
+      "loss": 0.1016,
+      "step": 5728
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.865946097697923e-05,
+      "loss": 0.1108,
+      "step": 5730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.865899307505147e-05,
+      "loss": 0.1022,
+      "step": 5732
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.865852517312371e-05,
+      "loss": 0.0797,
+      "step": 5734
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.865805727119596e-05,
+      "loss": 0.1102,
+      "step": 5736
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.8657589369268204e-05,
+      "loss": 0.1081,
+      "step": 5738
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.865712146734045e-05,
+      "loss": 0.1162,
+      "step": 5740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.865665356541269e-05,
+      "loss": 0.1079,
+      "step": 5742
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.8656185663484935e-05,
+      "loss": 0.0622,
+      "step": 5744
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.865571776155718e-05,
+      "loss": 0.0808,
+      "step": 5746
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.865524985962943e-05,
+      "loss": 0.0969,
+      "step": 5748
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.8654781957701666e-05,
+      "loss": 0.1134,
+      "step": 5750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.865431405577391e-05,
+      "loss": 0.0911,
+      "step": 5752
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.865384615384616e-05,
+      "loss": 0.1119,
+      "step": 5754
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.8653378251918404e-05,
+      "loss": 0.1267,
+      "step": 5756
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.865291034999064e-05,
+      "loss": 0.1022,
+      "step": 5758
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.865244244806289e-05,
+      "loss": 0.0947,
+      "step": 5760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.8651974546135135e-05,
+      "loss": 0.0856,
+      "step": 5762
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.865150664420738e-05,
+      "loss": 0.1083,
+      "step": 5764
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.865103874227962e-05,
+      "loss": 0.1219,
+      "step": 5766
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.8650570840351866e-05,
+      "loss": 0.1151,
+      "step": 5768
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.8650102938424105e-05,
+      "loss": 0.1197,
+      "step": 5770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.864963503649636e-05,
+      "loss": 0.0811,
+      "step": 5772
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.86491671345686e-05,
+      "loss": 0.093,
+      "step": 5774
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.864869923264084e-05,
+      "loss": 0.1262,
+      "step": 5776
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.864823133071308e-05,
+      "loss": 0.1196,
+      "step": 5778
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.8647763428785334e-05,
+      "loss": 0.0947,
+      "step": 5780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.8647295526857574e-05,
+      "loss": 0.1022,
+      "step": 5782
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.864682762492982e-05,
+      "loss": 0.1471,
+      "step": 5784
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.864635972300206e-05,
+      "loss": 0.0992,
+      "step": 5786
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.8645891821074305e-05,
+      "loss": 0.0914,
+      "step": 5788
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.864542391914655e-05,
+      "loss": 0.1023,
+      "step": 5790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.8644956017218796e-05,
+      "loss": 0.1051,
+      "step": 5792
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.8644488115291036e-05,
+      "loss": 0.1129,
+      "step": 5794
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.864402021336328e-05,
+      "loss": 0.0977,
+      "step": 5796
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.864355231143553e-05,
+      "loss": 0.0916,
+      "step": 5798
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.864308440950777e-05,
+      "loss": 0.0867,
+      "step": 5800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.864261650758001e-05,
+      "loss": 0.0936,
+      "step": 5802
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.864214860565226e-05,
+      "loss": 0.1163,
+      "step": 5804
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.8641680703724504e-05,
+      "loss": 0.104,
+      "step": 5806
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.864121280179675e-05,
+      "loss": 0.1155,
+      "step": 5808
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.864074489986899e-05,
+      "loss": 0.098,
+      "step": 5810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.8640276997941235e-05,
+      "loss": 0.0928,
+      "step": 5812
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.863980909601348e-05,
+      "loss": 0.0958,
+      "step": 5814
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.863934119408573e-05,
+      "loss": 0.0983,
+      "step": 5816
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.8638873292157966e-05,
+      "loss": 0.0974,
+      "step": 5818
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.863840539023021e-05,
+      "loss": 0.1004,
+      "step": 5820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.863793748830245e-05,
+      "loss": 0.14,
+      "step": 5822
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.8637469586374704e-05,
+      "loss": 0.0913,
+      "step": 5824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.863700168444694e-05,
+      "loss": 0.0883,
+      "step": 5826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.863653378251918e-05,
+      "loss": 0.1027,
+      "step": 5828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.863606588059143e-05,
+      "loss": 0.0953,
+      "step": 5830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.8635597978663674e-05,
+      "loss": 0.0811,
+      "step": 5832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.863513007673592e-05,
+      "loss": 0.1174,
+      "step": 5834
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.863466217480816e-05,
+      "loss": 0.1065,
+      "step": 5836
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.8634194272880405e-05,
+      "loss": 0.1064,
+      "step": 5838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.863372637095265e-05,
+      "loss": 0.0954,
+      "step": 5840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.86332584690249e-05,
+      "loss": 0.1126,
+      "step": 5842
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.8632790567097136e-05,
+      "loss": 0.0907,
+      "step": 5844
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.863232266516938e-05,
+      "loss": 0.111,
+      "step": 5846
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.863185476324163e-05,
+      "loss": 0.092,
+      "step": 5848
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.8631386861313874e-05,
+      "loss": 0.0969,
+      "step": 5850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.863091895938611e-05,
+      "loss": 0.1048,
+      "step": 5852
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.863045105745836e-05,
+      "loss": 0.098,
+      "step": 5854
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.86299831555306e-05,
+      "loss": 0.1183,
+      "step": 5856
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.862951525360285e-05,
+      "loss": 0.0794,
+      "step": 5858
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.862904735167509e-05,
+      "loss": 0.1104,
+      "step": 5860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.8628579449747336e-05,
+      "loss": 0.1069,
+      "step": 5862
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.8628111547819575e-05,
+      "loss": 0.1049,
+      "step": 5864
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.862764364589182e-05,
+      "loss": 0.0684,
+      "step": 5866
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.8627175743964066e-05,
+      "loss": 0.1266,
+      "step": 5868
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.862670784203631e-05,
+      "loss": 0.11,
+      "step": 5870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.862623994010855e-05,
+      "loss": 0.1415,
+      "step": 5872
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.86257720381808e-05,
+      "loss": 0.0985,
+      "step": 5874
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.862530413625304e-05,
+      "loss": 0.0979,
+      "step": 5876
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.862483623432529e-05,
+      "loss": 0.0997,
+      "step": 5878
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.862436833239753e-05,
+      "loss": 0.0902,
+      "step": 5880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.8623900430469774e-05,
+      "loss": 0.0743,
+      "step": 5882
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.862343252854202e-05,
+      "loss": 0.1087,
+      "step": 5884
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.8622964626614266e-05,
+      "loss": 0.1133,
+      "step": 5886
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.8622496724686505e-05,
+      "loss": 0.089,
+      "step": 5888
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.862202882275875e-05,
+      "loss": 0.0954,
+      "step": 5890
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.8621560920831e-05,
+      "loss": 0.097,
+      "step": 5892
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.862109301890324e-05,
+      "loss": 0.1184,
+      "step": 5894
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.862062511697548e-05,
+      "loss": 0.0975,
+      "step": 5896
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.862015721504773e-05,
+      "loss": 0.1071,
+      "step": 5898
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.861968931311997e-05,
+      "loss": 0.069,
+      "step": 5900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.861922141119222e-05,
+      "loss": 0.1376,
+      "step": 5902
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.861875350926446e-05,
+      "loss": 0.1246,
+      "step": 5904
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.8618285607336705e-05,
+      "loss": 0.1001,
+      "step": 5906
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.8617817705408944e-05,
+      "loss": 0.1003,
+      "step": 5908
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.86173498034812e-05,
+      "loss": 0.1234,
+      "step": 5910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.8616881901553436e-05,
+      "loss": 0.0929,
+      "step": 5912
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.861641399962568e-05,
+      "loss": 0.0917,
+      "step": 5914
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.861594609769792e-05,
+      "loss": 0.0919,
+      "step": 5916
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.861547819577017e-05,
+      "loss": 0.0958,
+      "step": 5918
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.861501029384241e-05,
+      "loss": 0.1047,
+      "step": 5920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.861454239191466e-05,
+      "loss": 0.0816,
+      "step": 5922
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.86140744899869e-05,
+      "loss": 0.1162,
+      "step": 5924
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.8613606588059144e-05,
+      "loss": 0.0917,
+      "step": 5926
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.861313868613139e-05,
+      "loss": 0.1072,
+      "step": 5928
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.8612670784203636e-05,
+      "loss": 0.1359,
+      "step": 5930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.8612202882275875e-05,
+      "loss": 0.1258,
+      "step": 5932
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.861173498034812e-05,
+      "loss": 0.1104,
+      "step": 5934
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.8611267078420366e-05,
+      "loss": 0.0921,
+      "step": 5936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.861079917649261e-05,
+      "loss": 0.0894,
+      "step": 5938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.861033127456485e-05,
+      "loss": 0.086,
+      "step": 5940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.86098633726371e-05,
+      "loss": 0.1098,
+      "step": 5942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.860939547070934e-05,
+      "loss": 0.0945,
+      "step": 5944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.860892756878159e-05,
+      "loss": 0.083,
+      "step": 5946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.860845966685383e-05,
+      "loss": 0.0899,
+      "step": 5948
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.8607991764926074e-05,
+      "loss": 0.1089,
+      "step": 5950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.8607523862998313e-05,
+      "loss": 0.0862,
+      "step": 5952
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.8607055961070566e-05,
+      "loss": 0.1048,
+      "step": 5954
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.8606588059142805e-05,
+      "loss": 0.1208,
+      "step": 5956
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.860612015721505e-05,
+      "loss": 0.1038,
+      "step": 5958
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.860565225528729e-05,
+      "loss": 0.0831,
+      "step": 5960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.860518435335954e-05,
+      "loss": 0.0993,
+      "step": 5962
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.860471645143178e-05,
+      "loss": 0.1206,
+      "step": 5964
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.860424854950403e-05,
+      "loss": 0.0931,
+      "step": 5966
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.860378064757627e-05,
+      "loss": 0.0725,
+      "step": 5968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.860331274564851e-05,
+      "loss": 0.0996,
+      "step": 5970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.860284484372076e-05,
+      "loss": 0.0704,
+      "step": 5972
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.8602376941793005e-05,
+      "loss": 0.0899,
+      "step": 5974
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.8601909039865244e-05,
+      "loss": 0.0883,
+      "step": 5976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.860144113793749e-05,
+      "loss": 0.0819,
+      "step": 5978
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.8600973236009736e-05,
+      "loss": 0.0999,
+      "step": 5980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.860050533408198e-05,
+      "loss": 0.0846,
+      "step": 5982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.860003743215422e-05,
+      "loss": 0.0805,
+      "step": 5984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.859956953022647e-05,
+      "loss": 0.1264,
+      "step": 5986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.859910162829871e-05,
+      "loss": 0.1262,
+      "step": 5988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.859863372637096e-05,
+      "loss": 0.1123,
+      "step": 5990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.85981658244432e-05,
+      "loss": 0.106,
+      "step": 5992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.8597697922515444e-05,
+      "loss": 0.0991,
+      "step": 5994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.859723002058769e-05,
+      "loss": 0.0839,
+      "step": 5996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.8596762118659936e-05,
+      "loss": 0.0766,
+      "step": 5998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.8596294216732175e-05,
+      "loss": 0.1062,
+      "step": 6000
+    },
+    {
+      "epoch": 0.84,
+      "eval_gen_len": 29.3793,
+      "eval_loss": 1.022807002067566,
+      "eval_meteor": 0.0513,
+      "eval_runtime": 14.225,
+      "eval_samples_per_second": 4.077,
+      "eval_steps_per_second": 0.562,
+      "step": 6000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.859582631480442e-05,
+      "loss": 0.0607,
+      "step": 6002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.859535841287666e-05,
+      "loss": 0.0952,
+      "step": 6004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.859489051094891e-05,
+      "loss": 0.1087,
+      "step": 6006
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.859442260902115e-05,
+      "loss": 0.1181,
+      "step": 6008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.85939547070934e-05,
+      "loss": 0.0786,
+      "step": 6010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.8593486805165637e-05,
+      "loss": 0.0841,
+      "step": 6012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.859301890323788e-05,
+      "loss": 0.1031,
+      "step": 6014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.859255100131013e-05,
+      "loss": 0.092,
+      "step": 6016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.8592083099382374e-05,
+      "loss": 0.1029,
+      "step": 6018
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.8591615197454613e-05,
+      "loss": 0.0929,
+      "step": 6020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.859114729552686e-05,
+      "loss": 0.0853,
+      "step": 6022
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.8590679393599105e-05,
+      "loss": 0.0855,
+      "step": 6024
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.859021149167135e-05,
+      "loss": 0.0849,
+      "step": 6026
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.858974358974359e-05,
+      "loss": 0.0936,
+      "step": 6028
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.8589275687815836e-05,
+      "loss": 0.1235,
+      "step": 6030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.858880778588808e-05,
+      "loss": 0.0925,
+      "step": 6032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.858833988396033e-05,
+      "loss": 0.1205,
+      "step": 6034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.858787198203257e-05,
+      "loss": 0.1217,
+      "step": 6036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.858740408010481e-05,
+      "loss": 0.0957,
+      "step": 6038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.858693617817706e-05,
+      "loss": 0.0844,
+      "step": 6040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.8586468276249305e-05,
+      "loss": 0.0872,
+      "step": 6042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.8586000374321544e-05,
+      "loss": 0.1021,
+      "step": 6044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.858553247239379e-05,
+      "loss": 0.1088,
+      "step": 6046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.858506457046603e-05,
+      "loss": 0.0879,
+      "step": 6048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.858459666853828e-05,
+      "loss": 0.1001,
+      "step": 6050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.858412876661052e-05,
+      "loss": 0.1154,
+      "step": 6052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.858366086468277e-05,
+      "loss": 0.0824,
+      "step": 6054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.8583192962755006e-05,
+      "loss": 0.0981,
+      "step": 6056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.858272506082726e-05,
+      "loss": 0.0853,
+      "step": 6058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.85822571588995e-05,
+      "loss": 0.0906,
+      "step": 6060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.8581789256971744e-05,
+      "loss": 0.1102,
+      "step": 6062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.858132135504398e-05,
+      "loss": 0.0784,
+      "step": 6064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.858085345311623e-05,
+      "loss": 0.0725,
+      "step": 6066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.8580385551188475e-05,
+      "loss": 0.1081,
+      "step": 6068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.857991764926072e-05,
+      "loss": 0.1199,
+      "step": 6070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.857944974733296e-05,
+      "loss": 0.1477,
+      "step": 6072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.8578981845405206e-05,
+      "loss": 0.1329,
+      "step": 6074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.857851394347745e-05,
+      "loss": 0.0915,
+      "step": 6076
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.857804604154969e-05,
+      "loss": 0.0836,
+      "step": 6078
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.857757813962194e-05,
+      "loss": 0.1161,
+      "step": 6080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.8577110237694176e-05,
+      "loss": 0.0971,
+      "step": 6082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.857664233576643e-05,
+      "loss": 0.1114,
+      "step": 6084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.857617443383867e-05,
+      "loss": 0.11,
+      "step": 6086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.8575706531910913e-05,
+      "loss": 0.1008,
+      "step": 6088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.857523862998315e-05,
+      "loss": 0.1019,
+      "step": 6090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8574770728055405e-05,
+      "loss": 0.1183,
+      "step": 6092
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8574302826127644e-05,
+      "loss": 0.1048,
+      "step": 6094
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.857383492419989e-05,
+      "loss": 0.0978,
+      "step": 6096
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.857336702227213e-05,
+      "loss": 0.1271,
+      "step": 6098
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8572899120344375e-05,
+      "loss": 0.1231,
+      "step": 6100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.857243121841662e-05,
+      "loss": 0.075,
+      "step": 6102
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.857196331648887e-05,
+      "loss": 0.0971,
+      "step": 6104
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8571495414561106e-05,
+      "loss": 0.0998,
+      "step": 6106
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.857102751263335e-05,
+      "loss": 0.1032,
+      "step": 6108
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.85705596107056e-05,
+      "loss": 0.0914,
+      "step": 6110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8570091708777844e-05,
+      "loss": 0.1036,
+      "step": 6112
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.856962380685008e-05,
+      "loss": 0.1222,
+      "step": 6114
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.856915590492233e-05,
+      "loss": 0.1505,
+      "step": 6116
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8568688002994575e-05,
+      "loss": 0.1288,
+      "step": 6118
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.856822010106682e-05,
+      "loss": 0.0904,
+      "step": 6120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.856775219913906e-05,
+      "loss": 0.0823,
+      "step": 6122
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8567284297211306e-05,
+      "loss": 0.1094,
+      "step": 6124
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.856681639528355e-05,
+      "loss": 0.0855,
+      "step": 6126
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.85663484933558e-05,
+      "loss": 0.1334,
+      "step": 6128
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.856588059142804e-05,
+      "loss": 0.1046,
+      "step": 6130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.856541268950028e-05,
+      "loss": 0.0929,
+      "step": 6132
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.856494478757252e-05,
+      "loss": 0.1016,
+      "step": 6134
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8564476885644775e-05,
+      "loss": 0.0838,
+      "step": 6136
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8564008983717014e-05,
+      "loss": 0.0816,
+      "step": 6138
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.856354108178926e-05,
+      "loss": 0.11,
+      "step": 6140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.85630731798615e-05,
+      "loss": 0.1227,
+      "step": 6142
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8562605277933745e-05,
+      "loss": 0.0896,
+      "step": 6144
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.856213737600599e-05,
+      "loss": 0.1126,
+      "step": 6146
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.856166947407824e-05,
+      "loss": 0.0829,
+      "step": 6148
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8561201572150476e-05,
+      "loss": 0.1106,
+      "step": 6150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.856073367022272e-05,
+      "loss": 0.1107,
+      "step": 6152
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.856026576829497e-05,
+      "loss": 0.1127,
+      "step": 6154
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8559797866367214e-05,
+      "loss": 0.0964,
+      "step": 6156
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.855932996443945e-05,
+      "loss": 0.1157,
+      "step": 6158
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.85588620625117e-05,
+      "loss": 0.0899,
+      "step": 6160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8558394160583944e-05,
+      "loss": 0.119,
+      "step": 6162
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.855792625865619e-05,
+      "loss": 0.0888,
+      "step": 6164
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.855745835672843e-05,
+      "loss": 0.1033,
+      "step": 6166
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.8556990454800675e-05,
+      "loss": 0.1036,
+      "step": 6168
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.855652255287292e-05,
+      "loss": 0.1108,
+      "step": 6170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.855605465094517e-05,
+      "loss": 0.0931,
+      "step": 6172
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.8555586749017406e-05,
+      "loss": 0.119,
+      "step": 6174
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.855511884708965e-05,
+      "loss": 0.096,
+      "step": 6176
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.855465094516189e-05,
+      "loss": 0.1296,
+      "step": 6178
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.8554183043234144e-05,
+      "loss": 0.1038,
+      "step": 6180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.855371514130638e-05,
+      "loss": 0.1,
+      "step": 6182
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.855324723937863e-05,
+      "loss": 0.0973,
+      "step": 6184
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.855277933745087e-05,
+      "loss": 0.0748,
+      "step": 6186
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.855231143552312e-05,
+      "loss": 0.0792,
+      "step": 6188
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.855184353359536e-05,
+      "loss": 0.0941,
+      "step": 6190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.8551375631667606e-05,
+      "loss": 0.0827,
+      "step": 6192
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.8550907729739845e-05,
+      "loss": 0.1037,
+      "step": 6194
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.855043982781209e-05,
+      "loss": 0.105,
+      "step": 6196
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.854997192588434e-05,
+      "loss": 0.1088,
+      "step": 6198
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.854950402395658e-05,
+      "loss": 0.1072,
+      "step": 6200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.854903612202882e-05,
+      "loss": 0.1163,
+      "step": 6202
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.854856822010107e-05,
+      "loss": 0.0877,
+      "step": 6204
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.8548100318173314e-05,
+      "loss": 0.114,
+      "step": 6206
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.854763241624556e-05,
+      "loss": 0.1165,
+      "step": 6208
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.85471645143178e-05,
+      "loss": 0.1088,
+      "step": 6210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.8546696612390045e-05,
+      "loss": 0.0901,
+      "step": 6212
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.854622871046229e-05,
+      "loss": 0.0794,
+      "step": 6214
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.854576080853454e-05,
+      "loss": 0.1076,
+      "step": 6216
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.8545292906606776e-05,
+      "loss": 0.1255,
+      "step": 6218
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.854482500467902e-05,
+      "loss": 0.0943,
+      "step": 6220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.854435710275127e-05,
+      "loss": 0.1077,
+      "step": 6222
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.8543889200823514e-05,
+      "loss": 0.1009,
+      "step": 6224
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.854342129889575e-05,
+      "loss": 0.1059,
+      "step": 6226
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.8542953396968e-05,
+      "loss": 0.0911,
+      "step": 6228
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.854248549504024e-05,
+      "loss": 0.183,
+      "step": 6230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.854201759311249e-05,
+      "loss": 0.1078,
+      "step": 6232
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.854154969118473e-05,
+      "loss": 0.0954,
+      "step": 6234
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.8541081789256975e-05,
+      "loss": 0.1076,
+      "step": 6236
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.8540613887329215e-05,
+      "loss": 0.0982,
+      "step": 6238
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.854014598540147e-05,
+      "loss": 0.1291,
+      "step": 6240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.8539678083473706e-05,
+      "loss": 0.0829,
+      "step": 6242
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.853921018154595e-05,
+      "loss": 0.0662,
+      "step": 6244
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.853874227961819e-05,
+      "loss": 0.0739,
+      "step": 6246
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.853827437769044e-05,
+      "loss": 0.0794,
+      "step": 6248
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.853780647576268e-05,
+      "loss": 0.1045,
+      "step": 6250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.853733857383493e-05,
+      "loss": 0.0904,
+      "step": 6252
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.853687067190717e-05,
+      "loss": 0.077,
+      "step": 6254
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.8536402769979414e-05,
+      "loss": 0.0991,
+      "step": 6256
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.853593486805166e-05,
+      "loss": 0.1022,
+      "step": 6258
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.8535466966123906e-05,
+      "loss": 0.106,
+      "step": 6260
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.8534999064196145e-05,
+      "loss": 0.1008,
+      "step": 6262
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.853453116226839e-05,
+      "loss": 0.0948,
+      "step": 6264
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.853406326034064e-05,
+      "loss": 0.0963,
+      "step": 6266
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.853359535841288e-05,
+      "loss": 0.0881,
+      "step": 6268
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.853312745648512e-05,
+      "loss": 0.0933,
+      "step": 6270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.853265955455737e-05,
+      "loss": 0.0912,
+      "step": 6272
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.8532191652629614e-05,
+      "loss": 0.1104,
+      "step": 6274
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.853172375070186e-05,
+      "loss": 0.11,
+      "step": 6276
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.85312558487741e-05,
+      "loss": 0.0975,
+      "step": 6278
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.8530787946846345e-05,
+      "loss": 0.1034,
+      "step": 6280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.8530320044918584e-05,
+      "loss": 0.0773,
+      "step": 6282
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.852985214299084e-05,
+      "loss": 0.1249,
+      "step": 6284
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.8529384241063076e-05,
+      "loss": 0.1355,
+      "step": 6286
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.852891633913532e-05,
+      "loss": 0.0809,
+      "step": 6288
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.852844843720756e-05,
+      "loss": 0.1087,
+      "step": 6290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.852798053527981e-05,
+      "loss": 0.0923,
+      "step": 6292
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.852751263335205e-05,
+      "loss": 0.1128,
+      "step": 6294
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.85270447314243e-05,
+      "loss": 0.0799,
+      "step": 6296
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.852657682949654e-05,
+      "loss": 0.0949,
+      "step": 6298
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.8526108927568784e-05,
+      "loss": 0.1028,
+      "step": 6300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.852564102564103e-05,
+      "loss": 0.1062,
+      "step": 6302
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.8525173123713275e-05,
+      "loss": 0.0819,
+      "step": 6304
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.8524705221785515e-05,
+      "loss": 0.0895,
+      "step": 6306
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.852423731985776e-05,
+      "loss": 0.0985,
+      "step": 6308
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.8523769417930006e-05,
+      "loss": 0.0971,
+      "step": 6310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.852330151600225e-05,
+      "loss": 0.1381,
+      "step": 6312
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.852283361407449e-05,
+      "loss": 0.0967,
+      "step": 6314
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.852236571214674e-05,
+      "loss": 0.1021,
+      "step": 6316
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.852189781021898e-05,
+      "loss": 0.0963,
+      "step": 6318
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.852142990829123e-05,
+      "loss": 0.1066,
+      "step": 6320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.852096200636347e-05,
+      "loss": 0.0817,
+      "step": 6322
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.8520494104435714e-05,
+      "loss": 0.1203,
+      "step": 6324
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.8520026202507953e-05,
+      "loss": 0.0768,
+      "step": 6326
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.8519558300580206e-05,
+      "loss": 0.0735,
+      "step": 6328
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.8519090398652445e-05,
+      "loss": 0.1131,
+      "step": 6330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.8518622496724684e-05,
+      "loss": 0.0946,
+      "step": 6332
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.851815459479693e-05,
+      "loss": 0.1166,
+      "step": 6334
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.8517686692869176e-05,
+      "loss": 0.0827,
+      "step": 6336
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.851721879094142e-05,
+      "loss": 0.0983,
+      "step": 6338
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.851675088901366e-05,
+      "loss": 0.0972,
+      "step": 6340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.851628298708591e-05,
+      "loss": 0.0934,
+      "step": 6342
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.851581508515815e-05,
+      "loss": 0.1255,
+      "step": 6344
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.85153471832304e-05,
+      "loss": 0.0948,
+      "step": 6346
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.851487928130264e-05,
+      "loss": 0.0938,
+      "step": 6348
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.8514411379374884e-05,
+      "loss": 0.0958,
+      "step": 6350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.851394347744713e-05,
+      "loss": 0.1154,
+      "step": 6352
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.8513475575519376e-05,
+      "loss": 0.1278,
+      "step": 6354
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.8513007673591615e-05,
+      "loss": 0.1073,
+      "step": 6356
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.851253977166386e-05,
+      "loss": 0.109,
+      "step": 6358
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.85120718697361e-05,
+      "loss": 0.102,
+      "step": 6360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.851160396780835e-05,
+      "loss": 0.0921,
+      "step": 6362
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.851113606588059e-05,
+      "loss": 0.0813,
+      "step": 6364
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.851066816395284e-05,
+      "loss": 0.1047,
+      "step": 6366
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.851020026202508e-05,
+      "loss": 0.1126,
+      "step": 6368
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.850973236009733e-05,
+      "loss": 0.0892,
+      "step": 6370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.850926445816957e-05,
+      "loss": 0.1046,
+      "step": 6372
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.8508796556241815e-05,
+      "loss": 0.1093,
+      "step": 6374
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8508328654314054e-05,
+      "loss": 0.1002,
+      "step": 6376
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.85078607523863e-05,
+      "loss": 0.1065,
+      "step": 6378
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8507392850458546e-05,
+      "loss": 0.1215,
+      "step": 6380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.850692494853079e-05,
+      "loss": 0.0916,
+      "step": 6382
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.850645704660303e-05,
+      "loss": 0.0933,
+      "step": 6384
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8505989144675277e-05,
+      "loss": 0.1038,
+      "step": 6386
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.850552124274752e-05,
+      "loss": 0.0875,
+      "step": 6388
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.850505334081977e-05,
+      "loss": 0.1225,
+      "step": 6390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.850458543889201e-05,
+      "loss": 0.0956,
+      "step": 6392
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8504117536964253e-05,
+      "loss": 0.1178,
+      "step": 6394
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.85036496350365e-05,
+      "loss": 0.1011,
+      "step": 6396
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8503181733108745e-05,
+      "loss": 0.0821,
+      "step": 6398
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8502713831180984e-05,
+      "loss": 0.1073,
+      "step": 6400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.850224592925323e-05,
+      "loss": 0.0952,
+      "step": 6402
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8501778027325476e-05,
+      "loss": 0.1331,
+      "step": 6404
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.850131012539772e-05,
+      "loss": 0.0767,
+      "step": 6406
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.850084222346996e-05,
+      "loss": 0.1026,
+      "step": 6408
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.850037432154221e-05,
+      "loss": 0.0974,
+      "step": 6410
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8499906419614446e-05,
+      "loss": 0.1189,
+      "step": 6412
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.84994385176867e-05,
+      "loss": 0.1029,
+      "step": 6414
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.849897061575894e-05,
+      "loss": 0.0899,
+      "step": 6416
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8498502713831184e-05,
+      "loss": 0.0949,
+      "step": 6418
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.849803481190342e-05,
+      "loss": 0.0998,
+      "step": 6420
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.849756690997567e-05,
+      "loss": 0.0758,
+      "step": 6422
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8497099008047915e-05,
+      "loss": 0.1262,
+      "step": 6424
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.849663110612016e-05,
+      "loss": 0.0781,
+      "step": 6426
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.84961632041924e-05,
+      "loss": 0.0878,
+      "step": 6428
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8495695302264646e-05,
+      "loss": 0.0719,
+      "step": 6430
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.849522740033689e-05,
+      "loss": 0.0843,
+      "step": 6432
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.849475949840914e-05,
+      "loss": 0.0928,
+      "step": 6434
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.849429159648138e-05,
+      "loss": 0.1238,
+      "step": 6436
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.849382369455362e-05,
+      "loss": 0.113,
+      "step": 6438
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.849335579262587e-05,
+      "loss": 0.0953,
+      "step": 6440
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8492887890698115e-05,
+      "loss": 0.1073,
+      "step": 6442
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8492419988770354e-05,
+      "loss": 0.1078,
+      "step": 6444
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.84919520868426e-05,
+      "loss": 0.0983,
+      "step": 6446
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8491484184914846e-05,
+      "loss": 0.087,
+      "step": 6448
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.849101628298709e-05,
+      "loss": 0.1162,
+      "step": 6450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.849054838105933e-05,
+      "loss": 0.1125,
+      "step": 6452
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8490080479131577e-05,
+      "loss": 0.1088,
+      "step": 6454
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8489612577203816e-05,
+      "loss": 0.1248,
+      "step": 6456
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.848914467527607e-05,
+      "loss": 0.0978,
+      "step": 6458
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.848867677334831e-05,
+      "loss": 0.0851,
+      "step": 6460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8488208871420553e-05,
+      "loss": 0.1121,
+      "step": 6462
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.848774096949279e-05,
+      "loss": 0.0838,
+      "step": 6464
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8487273067565045e-05,
+      "loss": 0.1132,
+      "step": 6466
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8486805165637284e-05,
+      "loss": 0.1201,
+      "step": 6468
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.848633726370953e-05,
+      "loss": 0.0927,
+      "step": 6470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.848586936178177e-05,
+      "loss": 0.1031,
+      "step": 6472
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8485401459854015e-05,
+      "loss": 0.1185,
+      "step": 6474
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.848493355792626e-05,
+      "loss": 0.1226,
+      "step": 6476
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.848446565599851e-05,
+      "loss": 0.1114,
+      "step": 6478
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8483997754070746e-05,
+      "loss": 0.1083,
+      "step": 6480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.848352985214299e-05,
+      "loss": 0.1019,
+      "step": 6482
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.848306195021524e-05,
+      "loss": 0.077,
+      "step": 6484
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8482594048287484e-05,
+      "loss": 0.1112,
+      "step": 6486
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.848212614635972e-05,
+      "loss": 0.0828,
+      "step": 6488
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.848165824443197e-05,
+      "loss": 0.1204,
+      "step": 6490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8481190342504215e-05,
+      "loss": 0.1105,
+      "step": 6492
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.848072244057646e-05,
+      "loss": 0.0913,
+      "step": 6494
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.84802545386487e-05,
+      "loss": 0.1125,
+      "step": 6496
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8479786636720946e-05,
+      "loss": 0.1119,
+      "step": 6498
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.847931873479319e-05,
+      "loss": 0.106,
+      "step": 6500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.847885083286544e-05,
+      "loss": 0.0962,
+      "step": 6502
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.847838293093768e-05,
+      "loss": 0.1283,
+      "step": 6504
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.847791502900992e-05,
+      "loss": 0.0884,
+      "step": 6506
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.847744712708216e-05,
+      "loss": 0.0893,
+      "step": 6508
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8476979225154415e-05,
+      "loss": 0.1104,
+      "step": 6510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8476511323226654e-05,
+      "loss": 0.0902,
+      "step": 6512
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.84760434212989e-05,
+      "loss": 0.1052,
+      "step": 6514
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.847557551937114e-05,
+      "loss": 0.0942,
+      "step": 6516
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.847510761744339e-05,
+      "loss": 0.1106,
+      "step": 6518
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.847463971551563e-05,
+      "loss": 0.0741,
+      "step": 6520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8474171813587877e-05,
+      "loss": 0.0991,
+      "step": 6522
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8473703911660116e-05,
+      "loss": 0.1039,
+      "step": 6524
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.847323600973236e-05,
+      "loss": 0.1115,
+      "step": 6526
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.847276810780461e-05,
+      "loss": 0.1064,
+      "step": 6528
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8472300205876853e-05,
+      "loss": 0.1291,
+      "step": 6530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.847183230394909e-05,
+      "loss": 0.1032,
+      "step": 6532
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.847136440202134e-05,
+      "loss": 0.1156,
+      "step": 6534
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8470896500093584e-05,
+      "loss": 0.1356,
+      "step": 6536
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.847042859816583e-05,
+      "loss": 0.1375,
+      "step": 6538
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.846996069623807e-05,
+      "loss": 0.0808,
+      "step": 6540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8469492794310315e-05,
+      "loss": 0.1261,
+      "step": 6542
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.846902489238256e-05,
+      "loss": 0.0996,
+      "step": 6544
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.846855699045481e-05,
+      "loss": 0.1103,
+      "step": 6546
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8468089088527046e-05,
+      "loss": 0.1061,
+      "step": 6548
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.846762118659929e-05,
+      "loss": 0.0861,
+      "step": 6550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.846715328467154e-05,
+      "loss": 0.0728,
+      "step": 6552
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8466685382743784e-05,
+      "loss": 0.1171,
+      "step": 6554
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.846621748081602e-05,
+      "loss": 0.092,
+      "step": 6556
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.846574957888827e-05,
+      "loss": 0.1062,
+      "step": 6558
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.846528167696051e-05,
+      "loss": 0.0762,
+      "step": 6560
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.846481377503276e-05,
+      "loss": 0.0735,
+      "step": 6562
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8464345873105e-05,
+      "loss": 0.0945,
+      "step": 6564
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8463877971177246e-05,
+      "loss": 0.1233,
+      "step": 6566
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8463410069249485e-05,
+      "loss": 0.1,
+      "step": 6568
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.846294216732173e-05,
+      "loss": 0.0977,
+      "step": 6570
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.846247426539398e-05,
+      "loss": 0.0987,
+      "step": 6572
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.846200636346622e-05,
+      "loss": 0.1063,
+      "step": 6574
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.846153846153846e-05,
+      "loss": 0.0971,
+      "step": 6576
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.846107055961071e-05,
+      "loss": 0.1042,
+      "step": 6578
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8460602657682954e-05,
+      "loss": 0.1063,
+      "step": 6580
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.84601347557552e-05,
+      "loss": 0.1105,
+      "step": 6582
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.845966685382744e-05,
+      "loss": 0.1092,
+      "step": 6584
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8459198951899685e-05,
+      "loss": 0.0794,
+      "step": 6586
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.845873104997193e-05,
+      "loss": 0.1062,
+      "step": 6588
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.845826314804417e-05,
+      "loss": 0.1065,
+      "step": 6590
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8457795246116416e-05,
+      "loss": 0.1123,
+      "step": 6592
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8457327344188655e-05,
+      "loss": 0.1104,
+      "step": 6594
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.845685944226091e-05,
+      "loss": 0.0636,
+      "step": 6596
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.845639154033315e-05,
+      "loss": 0.0963,
+      "step": 6598
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.845592363840539e-05,
+      "loss": 0.1153,
+      "step": 6600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.845545573647763e-05,
+      "loss": 0.1124,
+      "step": 6602
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.845498783454988e-05,
+      "loss": 0.167,
+      "step": 6604
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8454519932622124e-05,
+      "loss": 0.0995,
+      "step": 6606
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.845405203069437e-05,
+      "loss": 0.1159,
+      "step": 6608
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.845358412876661e-05,
+      "loss": 0.1221,
+      "step": 6610
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8453116226838855e-05,
+      "loss": 0.0812,
+      "step": 6612
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.84526483249111e-05,
+      "loss": 0.0912,
+      "step": 6614
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8452180422983346e-05,
+      "loss": 0.125,
+      "step": 6616
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8451712521055585e-05,
+      "loss": 0.1155,
+      "step": 6618
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.845124461912783e-05,
+      "loss": 0.0926,
+      "step": 6620
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.845077671720008e-05,
+      "loss": 0.1366,
+      "step": 6622
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.845030881527232e-05,
+      "loss": 0.1135,
+      "step": 6624
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.844984091334456e-05,
+      "loss": 0.1239,
+      "step": 6626
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.844937301141681e-05,
+      "loss": 0.1278,
+      "step": 6628
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8448905109489054e-05,
+      "loss": 0.0896,
+      "step": 6630
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.84484372075613e-05,
+      "loss": 0.0938,
+      "step": 6632
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.844796930563354e-05,
+      "loss": 0.0911,
+      "step": 6634
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8447501403705785e-05,
+      "loss": 0.1133,
+      "step": 6636
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8447033501778024e-05,
+      "loss": 0.1028,
+      "step": 6638
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.844656559985028e-05,
+      "loss": 0.1258,
+      "step": 6640
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8446097697922516e-05,
+      "loss": 0.1117,
+      "step": 6642
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.844562979599476e-05,
+      "loss": 0.0975,
+      "step": 6644
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8445161894067e-05,
+      "loss": 0.1121,
+      "step": 6646
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8444693992139254e-05,
+      "loss": 0.1198,
+      "step": 6648
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.844422609021149e-05,
+      "loss": 0.1173,
+      "step": 6650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.844375818828374e-05,
+      "loss": 0.1062,
+      "step": 6652
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.844329028635598e-05,
+      "loss": 0.0994,
+      "step": 6654
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8442822384428224e-05,
+      "loss": 0.1192,
+      "step": 6656
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.844235448250047e-05,
+      "loss": 0.0775,
+      "step": 6658
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8441886580572716e-05,
+      "loss": 0.1088,
+      "step": 6660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8441418678644955e-05,
+      "loss": 0.1271,
+      "step": 6662
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.84409507767172e-05,
+      "loss": 0.0869,
+      "step": 6664
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.844048287478945e-05,
+      "loss": 0.1133,
+      "step": 6666
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.844001497286169e-05,
+      "loss": 0.0873,
+      "step": 6668
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.843954707093393e-05,
+      "loss": 0.0897,
+      "step": 6670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.843907916900618e-05,
+      "loss": 0.0942,
+      "step": 6672
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8438611267078424e-05,
+      "loss": 0.1154,
+      "step": 6674
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.843814336515067e-05,
+      "loss": 0.106,
+      "step": 6676
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.843767546322291e-05,
+      "loss": 0.1179,
+      "step": 6678
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8437207561295155e-05,
+      "loss": 0.1076,
+      "step": 6680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.84367396593674e-05,
+      "loss": 0.1158,
+      "step": 6682
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8436271757439646e-05,
+      "loss": 0.1141,
+      "step": 6684
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8435803855511885e-05,
+      "loss": 0.1287,
+      "step": 6686
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.843533595358413e-05,
+      "loss": 0.0818,
+      "step": 6688
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.843486805165637e-05,
+      "loss": 0.1097,
+      "step": 6690
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.843440014972862e-05,
+      "loss": 0.0919,
+      "step": 6692
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.843393224780086e-05,
+      "loss": 0.0963,
+      "step": 6694
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.843346434587311e-05,
+      "loss": 0.1151,
+      "step": 6696
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.843299644394535e-05,
+      "loss": 0.1007,
+      "step": 6698
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.843252854201759e-05,
+      "loss": 0.1192,
+      "step": 6700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.843206064008984e-05,
+      "loss": 0.0704,
+      "step": 6702
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8431592738162085e-05,
+      "loss": 0.1191,
+      "step": 6704
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8431124836234324e-05,
+      "loss": 0.0758,
+      "step": 6706
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.843065693430657e-05,
+      "loss": 0.12,
+      "step": 6708
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8430189032378816e-05,
+      "loss": 0.1071,
+      "step": 6710
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.842972113045106e-05,
+      "loss": 0.1074,
+      "step": 6712
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.84292532285233e-05,
+      "loss": 0.0955,
+      "step": 6714
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.842878532659555e-05,
+      "loss": 0.0992,
+      "step": 6716
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.842831742466779e-05,
+      "loss": 0.1008,
+      "step": 6718
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.842784952274004e-05,
+      "loss": 0.1133,
+      "step": 6720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.842738162081228e-05,
+      "loss": 0.1024,
+      "step": 6722
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8426913718884524e-05,
+      "loss": 0.0731,
+      "step": 6724
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.842644581695677e-05,
+      "loss": 0.0925,
+      "step": 6726
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8425977915029016e-05,
+      "loss": 0.0857,
+      "step": 6728
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8425510013101255e-05,
+      "loss": 0.0843,
+      "step": 6730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.84250421111735e-05,
+      "loss": 0.106,
+      "step": 6732
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.842457420924574e-05,
+      "loss": 0.1088,
+      "step": 6734
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.842410630731799e-05,
+      "loss": 0.097,
+      "step": 6736
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.842363840539023e-05,
+      "loss": 0.1064,
+      "step": 6738
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.842317050346248e-05,
+      "loss": 0.1006,
+      "step": 6740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.842270260153472e-05,
+      "loss": 0.1257,
+      "step": 6742
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.842223469960697e-05,
+      "loss": 0.1,
+      "step": 6744
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.842176679767921e-05,
+      "loss": 0.1103,
+      "step": 6746
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8421298895751455e-05,
+      "loss": 0.108,
+      "step": 6748
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8420830993823694e-05,
+      "loss": 0.0737,
+      "step": 6750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.842036309189594e-05,
+      "loss": 0.1007,
+      "step": 6752
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8419895189968186e-05,
+      "loss": 0.0944,
+      "step": 6754
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.841942728804043e-05,
+      "loss": 0.1202,
+      "step": 6756
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.841895938611267e-05,
+      "loss": 0.0799,
+      "step": 6758
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8418491484184916e-05,
+      "loss": 0.0949,
+      "step": 6760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.841802358225716e-05,
+      "loss": 0.1105,
+      "step": 6762
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.841755568032941e-05,
+      "loss": 0.0744,
+      "step": 6764
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.841708777840165e-05,
+      "loss": 0.1234,
+      "step": 6766
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.841661987647389e-05,
+      "loss": 0.092,
+      "step": 6768
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.841615197454614e-05,
+      "loss": 0.0982,
+      "step": 6770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8415684072618385e-05,
+      "loss": 0.1087,
+      "step": 6772
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8415216170690624e-05,
+      "loss": 0.0769,
+      "step": 6774
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.841474826876287e-05,
+      "loss": 0.0998,
+      "step": 6776
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8414280366835116e-05,
+      "loss": 0.097,
+      "step": 6778
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.841381246490736e-05,
+      "loss": 0.1062,
+      "step": 6780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.84133445629796e-05,
+      "loss": 0.1102,
+      "step": 6782
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.841287666105185e-05,
+      "loss": 0.0784,
+      "step": 6784
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8412408759124086e-05,
+      "loss": 0.0953,
+      "step": 6786
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.841194085719634e-05,
+      "loss": 0.0952,
+      "step": 6788
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.841147295526858e-05,
+      "loss": 0.1155,
+      "step": 6790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8411005053340824e-05,
+      "loss": 0.0963,
+      "step": 6792
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.841053715141306e-05,
+      "loss": 0.1182,
+      "step": 6794
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8410069249485316e-05,
+      "loss": 0.0928,
+      "step": 6796
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8409601347557555e-05,
+      "loss": 0.097,
+      "step": 6798
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.84091334456298e-05,
+      "loss": 0.0953,
+      "step": 6800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.840866554370204e-05,
+      "loss": 0.0926,
+      "step": 6802
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8408197641774286e-05,
+      "loss": 0.074,
+      "step": 6804
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.840772973984653e-05,
+      "loss": 0.0794,
+      "step": 6806
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.840726183791878e-05,
+      "loss": 0.1061,
+      "step": 6808
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.840679393599102e-05,
+      "loss": 0.108,
+      "step": 6810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.840632603406326e-05,
+      "loss": 0.0953,
+      "step": 6812
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.840585813213551e-05,
+      "loss": 0.0895,
+      "step": 6814
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8405390230207755e-05,
+      "loss": 0.097,
+      "step": 6816
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8404922328279994e-05,
+      "loss": 0.0926,
+      "step": 6818
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.840445442635224e-05,
+      "loss": 0.1324,
+      "step": 6820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8403986524424486e-05,
+      "loss": 0.0832,
+      "step": 6822
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.840351862249673e-05,
+      "loss": 0.097,
+      "step": 6824
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.840305072056897e-05,
+      "loss": 0.1154,
+      "step": 6826
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8402582818641216e-05,
+      "loss": 0.1211,
+      "step": 6828
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.840211491671346e-05,
+      "loss": 0.105,
+      "step": 6830
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.840164701478571e-05,
+      "loss": 0.0973,
+      "step": 6832
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.840117911285795e-05,
+      "loss": 0.0933,
+      "step": 6834
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.840071121093019e-05,
+      "loss": 0.0872,
+      "step": 6836
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.840024330900243e-05,
+      "loss": 0.1034,
+      "step": 6838
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.839977540707468e-05,
+      "loss": 0.0869,
+      "step": 6840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8399307505146924e-05,
+      "loss": 0.0798,
+      "step": 6842
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8398839603219163e-05,
+      "loss": 0.0998,
+      "step": 6844
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.839837170129141e-05,
+      "loss": 0.0968,
+      "step": 6846
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8397903799363655e-05,
+      "loss": 0.1173,
+      "step": 6848
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.83974358974359e-05,
+      "loss": 0.0919,
+      "step": 6850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.839696799550814e-05,
+      "loss": 0.1159,
+      "step": 6852
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8396500093580386e-05,
+      "loss": 0.0714,
+      "step": 6854
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.839603219165263e-05,
+      "loss": 0.0924,
+      "step": 6856
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.839556428972488e-05,
+      "loss": 0.1054,
+      "step": 6858
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.839509638779712e-05,
+      "loss": 0.1118,
+      "step": 6860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.839462848586936e-05,
+      "loss": 0.0879,
+      "step": 6862
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.839416058394161e-05,
+      "loss": 0.1111,
+      "step": 6864
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8393692682013855e-05,
+      "loss": 0.0892,
+      "step": 6866
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8393224780086094e-05,
+      "loss": 0.1077,
+      "step": 6868
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.839275687815834e-05,
+      "loss": 0.1028,
+      "step": 6870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.839228897623058e-05,
+      "loss": 0.1059,
+      "step": 6872
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.839182107430283e-05,
+      "loss": 0.0976,
+      "step": 6874
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.839135317237507e-05,
+      "loss": 0.0977,
+      "step": 6876
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.839088527044732e-05,
+      "loss": 0.0917,
+      "step": 6878
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8390417368519556e-05,
+      "loss": 0.0875,
+      "step": 6880
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.83899494665918e-05,
+      "loss": 0.1239,
+      "step": 6882
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.838948156466405e-05,
+      "loss": 0.0962,
+      "step": 6884
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8389013662736294e-05,
+      "loss": 0.105,
+      "step": 6886
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.838854576080853e-05,
+      "loss": 0.102,
+      "step": 6888
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.838807785888078e-05,
+      "loss": 0.0977,
+      "step": 6890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8387609956953025e-05,
+      "loss": 0.1037,
+      "step": 6892
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.838714205502527e-05,
+      "loss": 0.0885,
+      "step": 6894
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.838667415309751e-05,
+      "loss": 0.0917,
+      "step": 6896
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8386206251169756e-05,
+      "loss": 0.1001,
+      "step": 6898
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8385738349242e-05,
+      "loss": 0.1313,
+      "step": 6900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.838527044731425e-05,
+      "loss": 0.1169,
+      "step": 6902
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8384802545386487e-05,
+      "loss": 0.1195,
+      "step": 6904
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.838433464345873e-05,
+      "loss": 0.0921,
+      "step": 6906
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.838386674153098e-05,
+      "loss": 0.1008,
+      "step": 6908
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8383398839603224e-05,
+      "loss": 0.0865,
+      "step": 6910
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8382930937675463e-05,
+      "loss": 0.0884,
+      "step": 6912
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.838246303574771e-05,
+      "loss": 0.1269,
+      "step": 6914
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.838199513381995e-05,
+      "loss": 0.0834,
+      "step": 6916
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.83815272318922e-05,
+      "loss": 0.0979,
+      "step": 6918
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.838105932996444e-05,
+      "loss": 0.0706,
+      "step": 6920
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8380591428036686e-05,
+      "loss": 0.1109,
+      "step": 6922
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8380123526108925e-05,
+      "loss": 0.0822,
+      "step": 6924
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.837965562418118e-05,
+      "loss": 0.103,
+      "step": 6926
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.837918772225342e-05,
+      "loss": 0.1162,
+      "step": 6928
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.837871982032566e-05,
+      "loss": 0.0904,
+      "step": 6930
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.83782519183979e-05,
+      "loss": 0.0865,
+      "step": 6932
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.837778401647015e-05,
+      "loss": 0.1013,
+      "step": 6934
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8377316114542394e-05,
+      "loss": 0.0997,
+      "step": 6936
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.837684821261464e-05,
+      "loss": 0.096,
+      "step": 6938
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.837638031068688e-05,
+      "loss": 0.1282,
+      "step": 6940
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8375912408759125e-05,
+      "loss": 0.081,
+      "step": 6942
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.837544450683137e-05,
+      "loss": 0.0763,
+      "step": 6944
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.837497660490362e-05,
+      "loss": 0.0851,
+      "step": 6946
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8374508702975856e-05,
+      "loss": 0.1228,
+      "step": 6948
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.83740408010481e-05,
+      "loss": 0.1094,
+      "step": 6950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.837357289912035e-05,
+      "loss": 0.1124,
+      "step": 6952
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8373104997192594e-05,
+      "loss": 0.0812,
+      "step": 6954
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.837263709526483e-05,
+      "loss": 0.0816,
+      "step": 6956
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.837216919333708e-05,
+      "loss": 0.1111,
+      "step": 6958
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8371701291409325e-05,
+      "loss": 0.1044,
+      "step": 6960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.837123338948157e-05,
+      "loss": 0.1142,
+      "step": 6962
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.837076548755381e-05,
+      "loss": 0.1034,
+      "step": 6964
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8370297585626056e-05,
+      "loss": 0.1176,
+      "step": 6966
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8369829683698295e-05,
+      "loss": 0.0962,
+      "step": 6968
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.836936178177055e-05,
+      "loss": 0.0932,
+      "step": 6970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.836889387984279e-05,
+      "loss": 0.1057,
+      "step": 6972
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.836842597791503e-05,
+      "loss": 0.1024,
+      "step": 6974
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.836795807598727e-05,
+      "loss": 0.1027,
+      "step": 6976
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8367490174059524e-05,
+      "loss": 0.1228,
+      "step": 6978
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8367022272131763e-05,
+      "loss": 0.0971,
+      "step": 6980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.836655437020401e-05,
+      "loss": 0.1094,
+      "step": 6982
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.836608646827625e-05,
+      "loss": 0.0743,
+      "step": 6984
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8365618566348494e-05,
+      "loss": 0.0925,
+      "step": 6986
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.836515066442074e-05,
+      "loss": 0.1026,
+      "step": 6988
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8364682762492986e-05,
+      "loss": 0.1369,
+      "step": 6990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8364214860565225e-05,
+      "loss": 0.1028,
+      "step": 6992
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.836374695863747e-05,
+      "loss": 0.1015,
+      "step": 6994
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.836327905670972e-05,
+      "loss": 0.104,
+      "step": 6996
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.836281115478196e-05,
+      "loss": 0.0902,
+      "step": 6998
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.83623432528542e-05,
+      "loss": 0.1163,
+      "step": 7000
+    },
+    {
+      "epoch": 0.98,
+      "eval_gen_len": 29.0862,
+      "eval_loss": 1.0126045942306519,
+      "eval_meteor": 0.0443,
+      "eval_runtime": 14.844,
+      "eval_samples_per_second": 3.907,
+      "eval_steps_per_second": 0.539,
+      "step": 7000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.836187535092645e-05,
+      "loss": 0.1137,
+      "step": 7002
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8361407448998694e-05,
+      "loss": 0.0969,
+      "step": 7004
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.836093954707094e-05,
+      "loss": 0.0938,
+      "step": 7006
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.836047164514318e-05,
+      "loss": 0.0878,
+      "step": 7008
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8360003743215425e-05,
+      "loss": 0.0865,
+      "step": 7010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8359535841287664e-05,
+      "loss": 0.1052,
+      "step": 7012
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.835906793935992e-05,
+      "loss": 0.0923,
+      "step": 7014
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8358600037432156e-05,
+      "loss": 0.1033,
+      "step": 7016
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.83581321355044e-05,
+      "loss": 0.099,
+      "step": 7018
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.835766423357664e-05,
+      "loss": 0.119,
+      "step": 7020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8357196331648894e-05,
+      "loss": 0.1074,
+      "step": 7022
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.835672842972113e-05,
+      "loss": 0.1074,
+      "step": 7024
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.835626052779338e-05,
+      "loss": 0.104,
+      "step": 7026
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.835579262586562e-05,
+      "loss": 0.0993,
+      "step": 7028
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8355324723937864e-05,
+      "loss": 0.085,
+      "step": 7030
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.835485682201011e-05,
+      "loss": 0.0896,
+      "step": 7032
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8354388920082356e-05,
+      "loss": 0.0996,
+      "step": 7034
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8353921018154595e-05,
+      "loss": 0.1204,
+      "step": 7036
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.835345311622684e-05,
+      "loss": 0.0781,
+      "step": 7038
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.835298521429909e-05,
+      "loss": 0.1396,
+      "step": 7040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.835251731237133e-05,
+      "loss": 0.1755,
+      "step": 7042
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.835204941044357e-05,
+      "loss": 0.0937,
+      "step": 7044
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.835158150851582e-05,
+      "loss": 0.0942,
+      "step": 7046
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8351113606588064e-05,
+      "loss": 0.1169,
+      "step": 7048
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.835064570466031e-05,
+      "loss": 0.0962,
+      "step": 7050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.835017780273255e-05,
+      "loss": 0.1254,
+      "step": 7052
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8349709900804794e-05,
+      "loss": 0.1053,
+      "step": 7054
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.834924199887704e-05,
+      "loss": 0.1188,
+      "step": 7056
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8348774096949286e-05,
+      "loss": 0.1038,
+      "step": 7058
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8348306195021525e-05,
+      "loss": 0.1068,
+      "step": 7060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.834783829309377e-05,
+      "loss": 0.1112,
+      "step": 7062
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.834737039116601e-05,
+      "loss": 0.1352,
+      "step": 7064
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.834690248923826e-05,
+      "loss": 0.1129,
+      "step": 7066
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.83464345873105e-05,
+      "loss": 0.1124,
+      "step": 7068
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.834596668538275e-05,
+      "loss": 0.1069,
+      "step": 7070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.834549878345499e-05,
+      "loss": 0.1167,
+      "step": 7072
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.834503088152724e-05,
+      "loss": 0.0909,
+      "step": 7074
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.834456297959948e-05,
+      "loss": 0.1275,
+      "step": 7076
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8344095077671725e-05,
+      "loss": 0.096,
+      "step": 7078
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8343627175743964e-05,
+      "loss": 0.1105,
+      "step": 7080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.834315927381621e-05,
+      "loss": 0.1029,
+      "step": 7082
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8342691371888456e-05,
+      "loss": 0.0912,
+      "step": 7084
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.83422234699607e-05,
+      "loss": 0.0942,
+      "step": 7086
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.834175556803294e-05,
+      "loss": 0.0853,
+      "step": 7088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.834128766610519e-05,
+      "loss": 0.0912,
+      "step": 7090
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.834081976417743e-05,
+      "loss": 0.0714,
+      "step": 7092
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.834035186224967e-05,
+      "loss": 0.0945,
+      "step": 7094
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.833988396032192e-05,
+      "loss": 0.1026,
+      "step": 7096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.833941605839416e-05,
+      "loss": 0.1034,
+      "step": 7098
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.833894815646641e-05,
+      "loss": 0.0926,
+      "step": 7100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.833848025453865e-05,
+      "loss": 0.085,
+      "step": 7102
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8338012352610895e-05,
+      "loss": 0.1048,
+      "step": 7104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8337544450683134e-05,
+      "loss": 0.1001,
+      "step": 7106
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.833707654875539e-05,
+      "loss": 0.0964,
+      "step": 7108
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8336608646827626e-05,
+      "loss": 0.1004,
+      "step": 7110
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.833614074489987e-05,
+      "loss": 0.0852,
+      "step": 7112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.833567284297211e-05,
+      "loss": 0.0964,
+      "step": 7114
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.833520494104436e-05,
+      "loss": 0.0953,
+      "step": 7116
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.83347370391166e-05,
+      "loss": 0.1235,
+      "step": 7118
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.833426913718885e-05,
+      "loss": 0.0937,
+      "step": 7120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.833380123526109e-05,
+      "loss": 0.1229,
+      "step": 7122
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8333567284297214e-05,
+      "loss": 0.1498,
+      "step": 7124
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.833309938236946e-05,
+      "loss": 0.0874,
+      "step": 7126
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8332631480441706e-05,
+      "loss": 0.0612,
+      "step": 7128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8332163578513945e-05,
+      "loss": 0.0723,
+      "step": 7130
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.833169567658619e-05,
+      "loss": 0.0633,
+      "step": 7132
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.833122777465843e-05,
+      "loss": 0.068,
+      "step": 7134
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.833075987273068e-05,
+      "loss": 0.0644,
+      "step": 7136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.833029197080292e-05,
+      "loss": 0.059,
+      "step": 7138
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.832982406887517e-05,
+      "loss": 0.0491,
+      "step": 7140
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.832935616694741e-05,
+      "loss": 0.0526,
+      "step": 7142
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.832888826501965e-05,
+      "loss": 0.0508,
+      "step": 7144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.83284203630919e-05,
+      "loss": 0.0626,
+      "step": 7146
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8327952461164145e-05,
+      "loss": 0.0754,
+      "step": 7148
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8327484559236384e-05,
+      "loss": 0.0685,
+      "step": 7150
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.832701665730863e-05,
+      "loss": 0.0698,
+      "step": 7152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8326548755380876e-05,
+      "loss": 0.0732,
+      "step": 7154
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.832608085345312e-05,
+      "loss": 0.0628,
+      "step": 7156
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.832561295152536e-05,
+      "loss": 0.075,
+      "step": 7158
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8325145049597607e-05,
+      "loss": 0.0582,
+      "step": 7160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.832467714766985e-05,
+      "loss": 0.063,
+      "step": 7162
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.83242092457421e-05,
+      "loss": 0.0574,
+      "step": 7164
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.832374134381434e-05,
+      "loss": 0.057,
+      "step": 7166
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8323273441886583e-05,
+      "loss": 0.0757,
+      "step": 7168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.832280553995883e-05,
+      "loss": 0.0547,
+      "step": 7170
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8322337638031075e-05,
+      "loss": 0.0701,
+      "step": 7172
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8321869736103314e-05,
+      "loss": 0.0587,
+      "step": 7174
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.832140183417556e-05,
+      "loss": 0.0661,
+      "step": 7176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.83209339322478e-05,
+      "loss": 0.0653,
+      "step": 7178
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.832046603032005e-05,
+      "loss": 0.0517,
+      "step": 7180
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.831999812839229e-05,
+      "loss": 0.0517,
+      "step": 7182
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.831953022646454e-05,
+      "loss": 0.0689,
+      "step": 7184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8319062324536776e-05,
+      "loss": 0.0664,
+      "step": 7186
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.831859442260903e-05,
+      "loss": 0.0978,
+      "step": 7188
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.831812652068127e-05,
+      "loss": 0.0796,
+      "step": 7190
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8317658618753514e-05,
+      "loss": 0.0552,
+      "step": 7192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.831719071682575e-05,
+      "loss": 0.0539,
+      "step": 7194
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8316722814898e-05,
+      "loss": 0.0512,
+      "step": 7196
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8316254912970245e-05,
+      "loss": 0.0468,
+      "step": 7198
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.831578701104249e-05,
+      "loss": 0.0708,
+      "step": 7200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.831531910911473e-05,
+      "loss": 0.0638,
+      "step": 7202
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8314851207186976e-05,
+      "loss": 0.0721,
+      "step": 7204
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.831438330525922e-05,
+      "loss": 0.1117,
+      "step": 7206
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.831391540333147e-05,
+      "loss": 0.0617,
+      "step": 7208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.831344750140371e-05,
+      "loss": 0.0466,
+      "step": 7210
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.831297959947595e-05,
+      "loss": 0.064,
+      "step": 7212
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.83125116975482e-05,
+      "loss": 0.0715,
+      "step": 7214
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8312043795620445e-05,
+      "loss": 0.0638,
+      "step": 7216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8311575893692684e-05,
+      "loss": 0.0467,
+      "step": 7218
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.831110799176492e-05,
+      "loss": 0.0772,
+      "step": 7220
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8310640089837176e-05,
+      "loss": 0.0694,
+      "step": 7222
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8310172187909415e-05,
+      "loss": 0.0674,
+      "step": 7224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.830970428598166e-05,
+      "loss": 0.0755,
+      "step": 7226
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.83092363840539e-05,
+      "loss": 0.0662,
+      "step": 7228
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8308768482126146e-05,
+      "loss": 0.0614,
+      "step": 7230
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.830830058019839e-05,
+      "loss": 0.0579,
+      "step": 7232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.830783267827064e-05,
+      "loss": 0.0823,
+      "step": 7234
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.830736477634288e-05,
+      "loss": 0.0653,
+      "step": 7236
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.830689687441512e-05,
+      "loss": 0.0713,
+      "step": 7238
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.830642897248737e-05,
+      "loss": 0.0644,
+      "step": 7240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8305961070559614e-05,
+      "loss": 0.0711,
+      "step": 7242
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8305493168631854e-05,
+      "loss": 0.0618,
+      "step": 7244
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.83050252667041e-05,
+      "loss": 0.0537,
+      "step": 7246
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8304557364776345e-05,
+      "loss": 0.0449,
+      "step": 7248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.830408946284859e-05,
+      "loss": 0.0807,
+      "step": 7250
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.830362156092083e-05,
+      "loss": 0.0597,
+      "step": 7252
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8303153658993076e-05,
+      "loss": 0.067,
+      "step": 7254
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.830268575706532e-05,
+      "loss": 0.0769,
+      "step": 7256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.830221785513757e-05,
+      "loss": 0.0526,
+      "step": 7258
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.830174995320981e-05,
+      "loss": 0.0624,
+      "step": 7260
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.830128205128205e-05,
+      "loss": 0.0553,
+      "step": 7262
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.830081414935429e-05,
+      "loss": 0.0978,
+      "step": 7264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8300346247426545e-05,
+      "loss": 0.0651,
+      "step": 7266
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8299878345498784e-05,
+      "loss": 0.0584,
+      "step": 7268
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.829941044357103e-05,
+      "loss": 0.066,
+      "step": 7270
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.829894254164327e-05,
+      "loss": 0.0597,
+      "step": 7272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.829847463971552e-05,
+      "loss": 0.0579,
+      "step": 7274
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.829800673778776e-05,
+      "loss": 0.0657,
+      "step": 7276
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.829753883586001e-05,
+      "loss": 0.0558,
+      "step": 7278
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8297070933932246e-05,
+      "loss": 0.0477,
+      "step": 7280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.829660303200449e-05,
+      "loss": 0.0681,
+      "step": 7282
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.829613513007674e-05,
+      "loss": 0.0579,
+      "step": 7284
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8295667228148984e-05,
+      "loss": 0.0725,
+      "step": 7286
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.829519932622122e-05,
+      "loss": 0.0675,
+      "step": 7288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.829473142429347e-05,
+      "loss": 0.0747,
+      "step": 7290
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8294263522365715e-05,
+      "loss": 0.061,
+      "step": 7292
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.829379562043796e-05,
+      "loss": 0.0767,
+      "step": 7294
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.82933277185102e-05,
+      "loss": 0.0563,
+      "step": 7296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.8292859816582446e-05,
+      "loss": 0.0656,
+      "step": 7298
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.829239191465469e-05,
+      "loss": 0.0674,
+      "step": 7300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.829192401272694e-05,
+      "loss": 0.0565,
+      "step": 7302
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.829145611079918e-05,
+      "loss": 0.0848,
+      "step": 7304
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.829098820887142e-05,
+      "loss": 0.0647,
+      "step": 7306
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.829052030694366e-05,
+      "loss": 0.1279,
+      "step": 7308
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8290052405015914e-05,
+      "loss": 0.1035,
+      "step": 7310
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8289584503088154e-05,
+      "loss": 0.0651,
+      "step": 7312
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.82891166011604e-05,
+      "loss": 0.0533,
+      "step": 7314
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.828864869923264e-05,
+      "loss": 0.0539,
+      "step": 7316
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.828818079730489e-05,
+      "loss": 0.0557,
+      "step": 7318
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.828771289537713e-05,
+      "loss": 0.0614,
+      "step": 7320
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8287244993449376e-05,
+      "loss": 0.0697,
+      "step": 7322
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8286777091521616e-05,
+      "loss": 0.0604,
+      "step": 7324
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.828630918959386e-05,
+      "loss": 0.0769,
+      "step": 7326
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.828584128766611e-05,
+      "loss": 0.0574,
+      "step": 7328
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.828537338573835e-05,
+      "loss": 0.0731,
+      "step": 7330
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.828490548381059e-05,
+      "loss": 0.0723,
+      "step": 7332
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.828443758188284e-05,
+      "loss": 0.0887,
+      "step": 7334
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8283969679955084e-05,
+      "loss": 0.0518,
+      "step": 7336
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.828350177802733e-05,
+      "loss": 0.0574,
+      "step": 7338
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.828303387609957e-05,
+      "loss": 0.0467,
+      "step": 7340
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8282565974171815e-05,
+      "loss": 0.0675,
+      "step": 7342
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.828209807224406e-05,
+      "loss": 0.0528,
+      "step": 7344
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.828163017031631e-05,
+      "loss": 0.0507,
+      "step": 7346
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8281162268388546e-05,
+      "loss": 0.0674,
+      "step": 7348
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.828069436646079e-05,
+      "loss": 0.069,
+      "step": 7350
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.828022646453304e-05,
+      "loss": 0.0562,
+      "step": 7352
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8279758562605284e-05,
+      "loss": 0.0706,
+      "step": 7354
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.827929066067752e-05,
+      "loss": 0.0788,
+      "step": 7356
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.827882275874977e-05,
+      "loss": 0.057,
+      "step": 7358
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.827835485682201e-05,
+      "loss": 0.0555,
+      "step": 7360
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.827788695489426e-05,
+      "loss": 0.0682,
+      "step": 7362
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.82774190529665e-05,
+      "loss": 0.0704,
+      "step": 7364
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8276951151038746e-05,
+      "loss": 0.0572,
+      "step": 7366
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.8276483249110985e-05,
+      "loss": 0.0653,
+      "step": 7368
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.827601534718324e-05,
+      "loss": 0.0529,
+      "step": 7370
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.827554744525548e-05,
+      "loss": 0.0839,
+      "step": 7372
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.827507954332772e-05,
+      "loss": 0.054,
+      "step": 7374
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.827461164139996e-05,
+      "loss": 0.0571,
+      "step": 7376
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.827414373947221e-05,
+      "loss": 0.0714,
+      "step": 7378
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8273675837544454e-05,
+      "loss": 0.0643,
+      "step": 7380
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.82732079356167e-05,
+      "loss": 0.0683,
+      "step": 7382
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.827274003368894e-05,
+      "loss": 0.0637,
+      "step": 7384
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8272272131761185e-05,
+      "loss": 0.0618,
+      "step": 7386
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.827180422983343e-05,
+      "loss": 0.0454,
+      "step": 7388
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8271336327905676e-05,
+      "loss": 0.0771,
+      "step": 7390
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8270868425977916e-05,
+      "loss": 0.0571,
+      "step": 7392
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.827040052405016e-05,
+      "loss": 0.0552,
+      "step": 7394
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.826993262212241e-05,
+      "loss": 0.0479,
+      "step": 7396
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.826946472019465e-05,
+      "loss": 0.0548,
+      "step": 7398
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.826899681826689e-05,
+      "loss": 0.0895,
+      "step": 7400
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.826852891633914e-05,
+      "loss": 0.061,
+      "step": 7402
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8268061014411384e-05,
+      "loss": 0.0498,
+      "step": 7404
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.826759311248363e-05,
+      "loss": 0.0818,
+      "step": 7406
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.826712521055587e-05,
+      "loss": 0.0608,
+      "step": 7408
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8266657308628115e-05,
+      "loss": 0.0578,
+      "step": 7410
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8266189406700354e-05,
+      "loss": 0.0586,
+      "step": 7412
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.826572150477261e-05,
+      "loss": 0.0822,
+      "step": 7414
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8265253602844846e-05,
+      "loss": 0.0717,
+      "step": 7416
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.826478570091709e-05,
+      "loss": 0.0689,
+      "step": 7418
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.826431779898933e-05,
+      "loss": 0.0631,
+      "step": 7420
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.826384989706158e-05,
+      "loss": 0.0611,
+      "step": 7422
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.826338199513382e-05,
+      "loss": 0.0666,
+      "step": 7424
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.826291409320607e-05,
+      "loss": 0.0727,
+      "step": 7426
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.826244619127831e-05,
+      "loss": 0.0587,
+      "step": 7428
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8261978289350554e-05,
+      "loss": 0.0682,
+      "step": 7430
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.82615103874228e-05,
+      "loss": 0.0822,
+      "step": 7432
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8261042485495046e-05,
+      "loss": 0.0738,
+      "step": 7434
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.8260574583567285e-05,
+      "loss": 0.0635,
+      "step": 7436
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.826010668163953e-05,
+      "loss": 0.0888,
+      "step": 7438
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.825963877971178e-05,
+      "loss": 0.0693,
+      "step": 7440
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.825917087778402e-05,
+      "loss": 0.0491,
+      "step": 7442
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.825870297585626e-05,
+      "loss": 0.0827,
+      "step": 7444
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.825823507392851e-05,
+      "loss": 0.0647,
+      "step": 7446
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8257767172000754e-05,
+      "loss": 0.0726,
+      "step": 7448
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8257299270073e-05,
+      "loss": 0.0706,
+      "step": 7450
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.825683136814524e-05,
+      "loss": 0.0626,
+      "step": 7452
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8256363466217485e-05,
+      "loss": 0.0643,
+      "step": 7454
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8255895564289724e-05,
+      "loss": 0.077,
+      "step": 7456
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8255427662361976e-05,
+      "loss": 0.0779,
+      "step": 7458
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8254959760434216e-05,
+      "loss": 0.0888,
+      "step": 7460
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.825449185850646e-05,
+      "loss": 0.0717,
+      "step": 7462
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.82540239565787e-05,
+      "loss": 0.0695,
+      "step": 7464
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.825355605465095e-05,
+      "loss": 0.0626,
+      "step": 7466
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.825308815272319e-05,
+      "loss": 0.0669,
+      "step": 7468
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.825262025079544e-05,
+      "loss": 0.0651,
+      "step": 7470
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.825215234886768e-05,
+      "loss": 0.0729,
+      "step": 7472
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.825168444693992e-05,
+      "loss": 0.0771,
+      "step": 7474
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.825121654501217e-05,
+      "loss": 0.06,
+      "step": 7476
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.825074864308441e-05,
+      "loss": 0.09,
+      "step": 7478
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8250280741156654e-05,
+      "loss": 0.0725,
+      "step": 7480
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.82498128392289e-05,
+      "loss": 0.0557,
+      "step": 7482
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8249344937301146e-05,
+      "loss": 0.0807,
+      "step": 7484
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8248877035373385e-05,
+      "loss": 0.0546,
+      "step": 7486
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.824840913344563e-05,
+      "loss": 0.0702,
+      "step": 7488
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.824794123151787e-05,
+      "loss": 0.056,
+      "step": 7490
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.824747332959012e-05,
+      "loss": 0.0478,
+      "step": 7492
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.824700542766236e-05,
+      "loss": 0.1009,
+      "step": 7494
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.824653752573461e-05,
+      "loss": 0.0793,
+      "step": 7496
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.824606962380685e-05,
+      "loss": 0.0483,
+      "step": 7498
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.82456017218791e-05,
+      "loss": 0.0788,
+      "step": 7500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.824513381995134e-05,
+      "loss": 0.057,
+      "step": 7502
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8244665918023585e-05,
+      "loss": 0.0492,
+      "step": 7504
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8244198016095824e-05,
+      "loss": 0.0532,
+      "step": 7506
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.824373011416807e-05,
+      "loss": 0.0536,
+      "step": 7508
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.8243262212240316e-05,
+      "loss": 0.0543,
+      "step": 7510
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.824279431031256e-05,
+      "loss": 0.062,
+      "step": 7512
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.82423264083848e-05,
+      "loss": 0.0596,
+      "step": 7514
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.824185850645705e-05,
+      "loss": 0.062,
+      "step": 7516
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.824139060452929e-05,
+      "loss": 0.0675,
+      "step": 7518
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.824092270260154e-05,
+      "loss": 0.086,
+      "step": 7520
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.824045480067378e-05,
+      "loss": 0.066,
+      "step": 7522
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8239986898746024e-05,
+      "loss": 0.0588,
+      "step": 7524
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.823951899681827e-05,
+      "loss": 0.0607,
+      "step": 7526
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8239051094890516e-05,
+      "loss": 0.0704,
+      "step": 7528
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8238583192962755e-05,
+      "loss": 0.0591,
+      "step": 7530
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8238115291035e-05,
+      "loss": 0.06,
+      "step": 7532
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8237647389107247e-05,
+      "loss": 0.0442,
+      "step": 7534
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.823717948717949e-05,
+      "loss": 0.0712,
+      "step": 7536
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.823671158525173e-05,
+      "loss": 0.0796,
+      "step": 7538
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.823624368332398e-05,
+      "loss": 0.069,
+      "step": 7540
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8235775781396217e-05,
+      "loss": 0.0548,
+      "step": 7542
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.823530787946847e-05,
+      "loss": 0.0547,
+      "step": 7544
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.823483997754071e-05,
+      "loss": 0.0631,
+      "step": 7546
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8234372075612954e-05,
+      "loss": 0.0475,
+      "step": 7548
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8233904173685193e-05,
+      "loss": 0.0955,
+      "step": 7550
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8233436271757446e-05,
+      "loss": 0.0594,
+      "step": 7552
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8232968369829685e-05,
+      "loss": 0.0649,
+      "step": 7554
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.823250046790193e-05,
+      "loss": 0.0786,
+      "step": 7556
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.823203256597417e-05,
+      "loss": 0.0765,
+      "step": 7558
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8231564664046416e-05,
+      "loss": 0.0757,
+      "step": 7560
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.823109676211866e-05,
+      "loss": 0.0754,
+      "step": 7562
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.823062886019091e-05,
+      "loss": 0.0841,
+      "step": 7564
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.823016095826315e-05,
+      "loss": 0.0519,
+      "step": 7566
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.822969305633539e-05,
+      "loss": 0.0816,
+      "step": 7568
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.822922515440764e-05,
+      "loss": 0.0679,
+      "step": 7570
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8228757252479885e-05,
+      "loss": 0.0718,
+      "step": 7572
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8228289350552124e-05,
+      "loss": 0.0536,
+      "step": 7574
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.822782144862437e-05,
+      "loss": 0.0623,
+      "step": 7576
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.8227353546696616e-05,
+      "loss": 0.0431,
+      "step": 7578
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.822688564476886e-05,
+      "loss": 0.0596,
+      "step": 7580
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.82264177428411e-05,
+      "loss": 0.0538,
+      "step": 7582
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.822594984091335e-05,
+      "loss": 0.0893,
+      "step": 7584
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.822548193898559e-05,
+      "loss": 0.0555,
+      "step": 7586
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.822501403705784e-05,
+      "loss": 0.0734,
+      "step": 7588
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.822454613513008e-05,
+      "loss": 0.0567,
+      "step": 7590
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8224078233202324e-05,
+      "loss": 0.064,
+      "step": 7592
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.822361033127456e-05,
+      "loss": 0.065,
+      "step": 7594
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8223142429346816e-05,
+      "loss": 0.0534,
+      "step": 7596
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8222674527419055e-05,
+      "loss": 0.0747,
+      "step": 7598
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.82222066254913e-05,
+      "loss": 0.0647,
+      "step": 7600
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.822173872356354e-05,
+      "loss": 0.0856,
+      "step": 7602
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8221270821635786e-05,
+      "loss": 0.0707,
+      "step": 7604
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.822080291970803e-05,
+      "loss": 0.0771,
+      "step": 7606
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.822033501778028e-05,
+      "loss": 0.0515,
+      "step": 7608
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.821986711585252e-05,
+      "loss": 0.0551,
+      "step": 7610
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.821939921392476e-05,
+      "loss": 0.0631,
+      "step": 7612
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.821893131199701e-05,
+      "loss": 0.0606,
+      "step": 7614
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8218463410069254e-05,
+      "loss": 0.0536,
+      "step": 7616
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8217995508141493e-05,
+      "loss": 0.0663,
+      "step": 7618
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.821752760621374e-05,
+      "loss": 0.0762,
+      "step": 7620
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8217059704285985e-05,
+      "loss": 0.0706,
+      "step": 7622
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.821659180235823e-05,
+      "loss": 0.0515,
+      "step": 7624
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.821612390043047e-05,
+      "loss": 0.0671,
+      "step": 7626
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8215655998502716e-05,
+      "loss": 0.0682,
+      "step": 7628
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.821518809657496e-05,
+      "loss": 0.0613,
+      "step": 7630
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.821472019464721e-05,
+      "loss": 0.0691,
+      "step": 7632
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.821425229271945e-05,
+      "loss": 0.0641,
+      "step": 7634
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.821378439079169e-05,
+      "loss": 0.0649,
+      "step": 7636
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.821331648886393e-05,
+      "loss": 0.0827,
+      "step": 7638
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8212848586936185e-05,
+      "loss": 0.0626,
+      "step": 7640
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8212380685008424e-05,
+      "loss": 0.0616,
+      "step": 7642
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.821191278308067e-05,
+      "loss": 0.0842,
+      "step": 7644
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.821144488115291e-05,
+      "loss": 0.0791,
+      "step": 7646
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.821097697922516e-05,
+      "loss": 0.0669,
+      "step": 7648
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.82105090772974e-05,
+      "loss": 0.0816,
+      "step": 7650
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.821004117536965e-05,
+      "loss": 0.0793,
+      "step": 7652
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.8209573273441886e-05,
+      "loss": 0.0845,
+      "step": 7654
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.820910537151413e-05,
+      "loss": 0.0449,
+      "step": 7656
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.820863746958638e-05,
+      "loss": 0.0511,
+      "step": 7658
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8208169567658624e-05,
+      "loss": 0.0566,
+      "step": 7660
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.820770166573086e-05,
+      "loss": 0.0652,
+      "step": 7662
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.820723376380311e-05,
+      "loss": 0.0824,
+      "step": 7664
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8206765861875355e-05,
+      "loss": 0.0643,
+      "step": 7666
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.82062979599476e-05,
+      "loss": 0.0708,
+      "step": 7668
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.820583005801984e-05,
+      "loss": 0.0625,
+      "step": 7670
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8205362156092086e-05,
+      "loss": 0.079,
+      "step": 7672
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.820489425416433e-05,
+      "loss": 0.082,
+      "step": 7674
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.820442635223658e-05,
+      "loss": 0.05,
+      "step": 7676
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.820395845030882e-05,
+      "loss": 0.0489,
+      "step": 7678
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.820349054838106e-05,
+      "loss": 0.0706,
+      "step": 7680
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.820302264645331e-05,
+      "loss": 0.0685,
+      "step": 7682
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8202554744525554e-05,
+      "loss": 0.0767,
+      "step": 7684
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8202086842597794e-05,
+      "loss": 0.0713,
+      "step": 7686
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.820161894067004e-05,
+      "loss": 0.0667,
+      "step": 7688
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.820115103874228e-05,
+      "loss": 0.0843,
+      "step": 7690
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.820068313681453e-05,
+      "loss": 0.0671,
+      "step": 7692
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.820021523488677e-05,
+      "loss": 0.0672,
+      "step": 7694
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8199747332959016e-05,
+      "loss": 0.0553,
+      "step": 7696
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8199279431031255e-05,
+      "loss": 0.0708,
+      "step": 7698
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.81988115291035e-05,
+      "loss": 0.0711,
+      "step": 7700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.819834362717575e-05,
+      "loss": 0.0792,
+      "step": 7702
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.819787572524799e-05,
+      "loss": 0.0559,
+      "step": 7704
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.819740782332023e-05,
+      "loss": 0.0718,
+      "step": 7706
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.819693992139248e-05,
+      "loss": 0.0542,
+      "step": 7708
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8196472019464724e-05,
+      "loss": 0.0631,
+      "step": 7710
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.819600411753697e-05,
+      "loss": 0.0861,
+      "step": 7712
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.819553621560921e-05,
+      "loss": 0.0579,
+      "step": 7714
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8195068313681455e-05,
+      "loss": 0.0659,
+      "step": 7716
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.81946004117537e-05,
+      "loss": 0.08,
+      "step": 7718
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.819413250982595e-05,
+      "loss": 0.0688,
+      "step": 7720
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.8193664607898186e-05,
+      "loss": 0.0613,
+      "step": 7722
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.819319670597043e-05,
+      "loss": 0.0559,
+      "step": 7724
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.819272880404268e-05,
+      "loss": 0.0671,
+      "step": 7726
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.819226090211492e-05,
+      "loss": 0.0881,
+      "step": 7728
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.819179300018716e-05,
+      "loss": 0.0615,
+      "step": 7730
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.81913250982594e-05,
+      "loss": 0.0512,
+      "step": 7732
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.819085719633165e-05,
+      "loss": 0.0774,
+      "step": 7734
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8190389294403894e-05,
+      "loss": 0.0809,
+      "step": 7736
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.818992139247614e-05,
+      "loss": 0.0522,
+      "step": 7738
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.818945349054838e-05,
+      "loss": 0.0971,
+      "step": 7740
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8188985588620625e-05,
+      "loss": 0.0542,
+      "step": 7742
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.818851768669287e-05,
+      "loss": 0.0455,
+      "step": 7744
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.818804978476512e-05,
+      "loss": 0.0589,
+      "step": 7746
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8187581882837356e-05,
+      "loss": 0.0563,
+      "step": 7748
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.81871139809096e-05,
+      "loss": 0.0941,
+      "step": 7750
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.818664607898185e-05,
+      "loss": 0.0742,
+      "step": 7752
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8186178177054094e-05,
+      "loss": 0.0508,
+      "step": 7754
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.818571027512633e-05,
+      "loss": 0.0669,
+      "step": 7756
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.818524237319858e-05,
+      "loss": 0.0881,
+      "step": 7758
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8184774471270824e-05,
+      "loss": 0.0674,
+      "step": 7760
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.818430656934307e-05,
+      "loss": 0.0609,
+      "step": 7762
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.818383866741531e-05,
+      "loss": 0.0733,
+      "step": 7764
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8183370765487555e-05,
+      "loss": 0.0565,
+      "step": 7766
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8182902863559795e-05,
+      "loss": 0.0612,
+      "step": 7768
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.818243496163205e-05,
+      "loss": 0.0705,
+      "step": 7770
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8181967059704286e-05,
+      "loss": 0.0761,
+      "step": 7772
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.818149915777653e-05,
+      "loss": 0.0654,
+      "step": 7774
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.818103125584877e-05,
+      "loss": 0.0732,
+      "step": 7776
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8180563353921024e-05,
+      "loss": 0.0572,
+      "step": 7778
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.818009545199326e-05,
+      "loss": 0.0724,
+      "step": 7780
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.817962755006551e-05,
+      "loss": 0.0695,
+      "step": 7782
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.817915964813775e-05,
+      "loss": 0.0701,
+      "step": 7784
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8178691746209994e-05,
+      "loss": 0.0616,
+      "step": 7786
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.817822384428224e-05,
+      "loss": 0.07,
+      "step": 7788
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8177755942354486e-05,
+      "loss": 0.057,
+      "step": 7790
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.8177288040426725e-05,
+      "loss": 0.0552,
+      "step": 7792
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.817682013849897e-05,
+      "loss": 0.0599,
+      "step": 7794
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.817635223657122e-05,
+      "loss": 0.0666,
+      "step": 7796
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.817588433464346e-05,
+      "loss": 0.0762,
+      "step": 7798
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.81754164327157e-05,
+      "loss": 0.078,
+      "step": 7800
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.817494853078795e-05,
+      "loss": 0.0919,
+      "step": 7802
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8174480628860194e-05,
+      "loss": 0.0589,
+      "step": 7804
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.817401272693244e-05,
+      "loss": 0.0794,
+      "step": 7806
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.817354482500468e-05,
+      "loss": 0.0589,
+      "step": 7808
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8173076923076925e-05,
+      "loss": 0.06,
+      "step": 7810
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.817260902114917e-05,
+      "loss": 0.0764,
+      "step": 7812
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.817214111922142e-05,
+      "loss": 0.0708,
+      "step": 7814
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8171673217293656e-05,
+      "loss": 0.0765,
+      "step": 7816
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.81712053153659e-05,
+      "loss": 0.0632,
+      "step": 7818
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.817073741343814e-05,
+      "loss": 0.0682,
+      "step": 7820
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8170269511510394e-05,
+      "loss": 0.0869,
+      "step": 7822
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.816980160958263e-05,
+      "loss": 0.0638,
+      "step": 7824
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.816933370765488e-05,
+      "loss": 0.0583,
+      "step": 7826
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.816886580572712e-05,
+      "loss": 0.0606,
+      "step": 7828
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.816839790379937e-05,
+      "loss": 0.068,
+      "step": 7830
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.816793000187161e-05,
+      "loss": 0.0684,
+      "step": 7832
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8167462099943855e-05,
+      "loss": 0.07,
+      "step": 7834
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8166994198016095e-05,
+      "loss": 0.0608,
+      "step": 7836
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.816652629608834e-05,
+      "loss": 0.0557,
+      "step": 7838
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8166058394160586e-05,
+      "loss": 0.071,
+      "step": 7840
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.816559049223283e-05,
+      "loss": 0.0844,
+      "step": 7842
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.816512259030507e-05,
+      "loss": 0.0665,
+      "step": 7844
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.816465468837732e-05,
+      "loss": 0.0669,
+      "step": 7846
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.816418678644956e-05,
+      "loss": 0.0813,
+      "step": 7848
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.816371888452181e-05,
+      "loss": 0.06,
+      "step": 7850
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.816325098259405e-05,
+      "loss": 0.0722,
+      "step": 7852
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8162783080666294e-05,
+      "loss": 0.062,
+      "step": 7854
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.816231517873854e-05,
+      "loss": 0.0721,
+      "step": 7856
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8161847276810786e-05,
+      "loss": 0.0635,
+      "step": 7858
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8161379374883025e-05,
+      "loss": 0.0516,
+      "step": 7860
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.816091147295527e-05,
+      "loss": 0.0586,
+      "step": 7862
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.816044357102752e-05,
+      "loss": 0.0866,
+      "step": 7864
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.815997566909976e-05,
+      "loss": 0.1013,
+      "step": 7866
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.8159507767172e-05,
+      "loss": 0.0496,
+      "step": 7868
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.815903986524425e-05,
+      "loss": 0.056,
+      "step": 7870
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.815857196331649e-05,
+      "loss": 0.0675,
+      "step": 7872
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.815810406138874e-05,
+      "loss": 0.0529,
+      "step": 7874
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.815763615946098e-05,
+      "loss": 0.0555,
+      "step": 7876
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8157168257533225e-05,
+      "loss": 0.054,
+      "step": 7878
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8156700355605464e-05,
+      "loss": 0.0417,
+      "step": 7880
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.815623245367771e-05,
+      "loss": 0.0638,
+      "step": 7882
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8155764551749956e-05,
+      "loss": 0.0669,
+      "step": 7884
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.81552966498222e-05,
+      "loss": 0.0656,
+      "step": 7886
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.815482874789444e-05,
+      "loss": 0.0671,
+      "step": 7888
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.815436084596669e-05,
+      "loss": 0.0731,
+      "step": 7890
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.815389294403893e-05,
+      "loss": 0.0734,
+      "step": 7892
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.815342504211118e-05,
+      "loss": 0.0656,
+      "step": 7894
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.815295714018342e-05,
+      "loss": 0.0782,
+      "step": 7896
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8152489238255664e-05,
+      "loss": 0.0714,
+      "step": 7898
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.815202133632791e-05,
+      "loss": 0.0672,
+      "step": 7900
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8151553434400155e-05,
+      "loss": 0.0649,
+      "step": 7902
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8151085532472395e-05,
+      "loss": 0.0899,
+      "step": 7904
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.815061763054464e-05,
+      "loss": 0.0662,
+      "step": 7906
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8150149728616886e-05,
+      "loss": 0.0721,
+      "step": 7908
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.814968182668913e-05,
+      "loss": 0.049,
+      "step": 7910
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.814921392476137e-05,
+      "loss": 0.0706,
+      "step": 7912
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.814874602283362e-05,
+      "loss": 0.0792,
+      "step": 7914
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8148278120905857e-05,
+      "loss": 0.0566,
+      "step": 7916
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.814781021897811e-05,
+      "loss": 0.0589,
+      "step": 7918
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.814734231705035e-05,
+      "loss": 0.064,
+      "step": 7920
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8146874415122594e-05,
+      "loss": 0.0704,
+      "step": 7922
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8146406513194833e-05,
+      "loss": 0.0591,
+      "step": 7924
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8145938611267086e-05,
+      "loss": 0.076,
+      "step": 7926
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8145470709339325e-05,
+      "loss": 0.0813,
+      "step": 7928
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.814500280741157e-05,
+      "loss": 0.0558,
+      "step": 7930
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.814453490548381e-05,
+      "loss": 0.0759,
+      "step": 7932
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.8144067003556056e-05,
+      "loss": 0.0678,
+      "step": 7934
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.81435991016283e-05,
+      "loss": 0.0739,
+      "step": 7936
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.814313119970055e-05,
+      "loss": 0.0725,
+      "step": 7938
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.814266329777279e-05,
+      "loss": 0.0775,
+      "step": 7940
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.814219539584503e-05,
+      "loss": 0.0732,
+      "step": 7942
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.814172749391728e-05,
+      "loss": 0.0624,
+      "step": 7944
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8141259591989525e-05,
+      "loss": 0.0711,
+      "step": 7946
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8140791690061764e-05,
+      "loss": 0.0646,
+      "step": 7948
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.814032378813401e-05,
+      "loss": 0.0741,
+      "step": 7950
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8139855886206256e-05,
+      "loss": 0.059,
+      "step": 7952
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.81393879842785e-05,
+      "loss": 0.0857,
+      "step": 7954
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.813892008235074e-05,
+      "loss": 0.0581,
+      "step": 7956
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.813845218042299e-05,
+      "loss": 0.0606,
+      "step": 7958
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.813798427849523e-05,
+      "loss": 0.0702,
+      "step": 7960
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.813751637656748e-05,
+      "loss": 0.0659,
+      "step": 7962
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.813704847463972e-05,
+      "loss": 0.0766,
+      "step": 7964
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8136580572711964e-05,
+      "loss": 0.0675,
+      "step": 7966
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.81361126707842e-05,
+      "loss": 0.0975,
+      "step": 7968
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8135644768856455e-05,
+      "loss": 0.0676,
+      "step": 7970
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8135176866928695e-05,
+      "loss": 0.0702,
+      "step": 7972
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.813470896500094e-05,
+      "loss": 0.0593,
+      "step": 7974
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.813424106307318e-05,
+      "loss": 0.0601,
+      "step": 7976
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8133773161145426e-05,
+      "loss": 0.067,
+      "step": 7978
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.813330525921767e-05,
+      "loss": 0.0551,
+      "step": 7980
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.813283735728991e-05,
+      "loss": 0.0796,
+      "step": 7982
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8132369455362157e-05,
+      "loss": 0.082,
+      "step": 7984
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.81319015534344e-05,
+      "loss": 0.066,
+      "step": 7986
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.813143365150665e-05,
+      "loss": 0.0799,
+      "step": 7988
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.813096574957889e-05,
+      "loss": 0.071,
+      "step": 7990
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8130497847651133e-05,
+      "loss": 0.0782,
+      "step": 7992
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.813002994572338e-05,
+      "loss": 0.0746,
+      "step": 7994
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8129562043795625e-05,
+      "loss": 0.088,
+      "step": 7996
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8129094141867864e-05,
+      "loss": 0.0548,
+      "step": 7998
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.812862623994011e-05,
+      "loss": 0.0478,
+      "step": 8000
+    },
+    {
+      "epoch": 1.12,
+      "eval_gen_len": 30.9483,
+      "eval_loss": 1.0468480587005615,
+      "eval_meteor": 0.0439,
+      "eval_runtime": 14.548,
+      "eval_samples_per_second": 3.987,
+      "eval_steps_per_second": 0.55,
+      "step": 8000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.812815833801235e-05,
+      "loss": 0.05,
+      "step": 8002
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.81276904360846e-05,
+      "loss": 0.0737,
+      "step": 8004
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.812722253415684e-05,
+      "loss": 0.0783,
+      "step": 8006
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.812675463222909e-05,
+      "loss": 0.0702,
+      "step": 8008
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.8126286730301326e-05,
+      "loss": 0.0799,
+      "step": 8010
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.812581882837357e-05,
+      "loss": 0.0586,
+      "step": 8012
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.812535092644582e-05,
+      "loss": 0.0582,
+      "step": 8014
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8124883024518064e-05,
+      "loss": 0.0588,
+      "step": 8016
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.81244151225903e-05,
+      "loss": 0.0802,
+      "step": 8018
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.812394722066255e-05,
+      "loss": 0.0737,
+      "step": 8020
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8123479318734795e-05,
+      "loss": 0.0783,
+      "step": 8022
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.812301141680704e-05,
+      "loss": 0.0774,
+      "step": 8024
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.812254351487928e-05,
+      "loss": 0.0643,
+      "step": 8026
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8122075612951526e-05,
+      "loss": 0.0802,
+      "step": 8028
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.812160771102377e-05,
+      "loss": 0.0561,
+      "step": 8030
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.812113980909602e-05,
+      "loss": 0.0615,
+      "step": 8032
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.812067190716826e-05,
+      "loss": 0.0594,
+      "step": 8034
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.81202040052405e-05,
+      "loss": 0.0687,
+      "step": 8036
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.811973610331275e-05,
+      "loss": 0.0634,
+      "step": 8038
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8119268201384995e-05,
+      "loss": 0.0599,
+      "step": 8040
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8118800299457234e-05,
+      "loss": 0.0585,
+      "step": 8042
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.811833239752948e-05,
+      "loss": 0.0434,
+      "step": 8044
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.811786449560172e-05,
+      "loss": 0.0796,
+      "step": 8046
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.811739659367397e-05,
+      "loss": 0.0748,
+      "step": 8048
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.811692869174621e-05,
+      "loss": 0.0529,
+      "step": 8050
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8116460789818457e-05,
+      "loss": 0.0687,
+      "step": 8052
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8115992887890696e-05,
+      "loss": 0.0586,
+      "step": 8054
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.811552498596295e-05,
+      "loss": 0.0777,
+      "step": 8056
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.811505708403519e-05,
+      "loss": 0.0652,
+      "step": 8058
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8114589182107433e-05,
+      "loss": 0.0725,
+      "step": 8060
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.811412128017967e-05,
+      "loss": 0.0585,
+      "step": 8062
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.811365337825192e-05,
+      "loss": 0.0635,
+      "step": 8064
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8113185476324164e-05,
+      "loss": 0.0741,
+      "step": 8066
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.811271757439641e-05,
+      "loss": 0.0816,
+      "step": 8068
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.811224967246865e-05,
+      "loss": 0.0608,
+      "step": 8070
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8111781770540895e-05,
+      "loss": 0.0733,
+      "step": 8072
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.811131386861314e-05,
+      "loss": 0.0923,
+      "step": 8074
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.811084596668539e-05,
+      "loss": 0.0607,
+      "step": 8076
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8110378064757626e-05,
+      "loss": 0.0649,
+      "step": 8078
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.810991016282987e-05,
+      "loss": 0.0739,
+      "step": 8080
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.810944226090212e-05,
+      "loss": 0.0613,
+      "step": 8082
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.8108974358974364e-05,
+      "loss": 0.0989,
+      "step": 8084
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.81085064570466e-05,
+      "loss": 0.0552,
+      "step": 8086
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.810803855511885e-05,
+      "loss": 0.0609,
+      "step": 8088
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8107570653191095e-05,
+      "loss": 0.0698,
+      "step": 8090
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.810710275126334e-05,
+      "loss": 0.0753,
+      "step": 8092
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.810663484933558e-05,
+      "loss": 0.1152,
+      "step": 8094
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8106166947407826e-05,
+      "loss": 0.0786,
+      "step": 8096
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8105699045480065e-05,
+      "loss": 0.0664,
+      "step": 8098
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.810523114355232e-05,
+      "loss": 0.0641,
+      "step": 8100
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.810476324162456e-05,
+      "loss": 0.0837,
+      "step": 8102
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.81042953396968e-05,
+      "loss": 0.0699,
+      "step": 8104
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.810382743776904e-05,
+      "loss": 0.0884,
+      "step": 8106
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8103359535841295e-05,
+      "loss": 0.072,
+      "step": 8108
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8102891633913534e-05,
+      "loss": 0.0612,
+      "step": 8110
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.810242373198578e-05,
+      "loss": 0.0544,
+      "step": 8112
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.810195583005802e-05,
+      "loss": 0.0678,
+      "step": 8114
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8101487928130265e-05,
+      "loss": 0.07,
+      "step": 8116
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.810102002620251e-05,
+      "loss": 0.0607,
+      "step": 8118
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8100552124274757e-05,
+      "loss": 0.0729,
+      "step": 8120
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8100084222346996e-05,
+      "loss": 0.0656,
+      "step": 8122
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.809961632041924e-05,
+      "loss": 0.0468,
+      "step": 8124
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.809914841849149e-05,
+      "loss": 0.0738,
+      "step": 8126
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8098680516563733e-05,
+      "loss": 0.0672,
+      "step": 8128
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.809821261463597e-05,
+      "loss": 0.0643,
+      "step": 8130
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.809774471270822e-05,
+      "loss": 0.0657,
+      "step": 8132
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8097276810780464e-05,
+      "loss": 0.0779,
+      "step": 8134
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.809680890885271e-05,
+      "loss": 0.0668,
+      "step": 8136
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.809634100692495e-05,
+      "loss": 0.067,
+      "step": 8138
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8095873104997195e-05,
+      "loss": 0.0589,
+      "step": 8140
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.809540520306944e-05,
+      "loss": 0.0877,
+      "step": 8142
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.809493730114169e-05,
+      "loss": 0.0608,
+      "step": 8144
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8094469399213926e-05,
+      "loss": 0.0513,
+      "step": 8146
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.809400149728617e-05,
+      "loss": 0.0618,
+      "step": 8148
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.809353359535841e-05,
+      "loss": 0.0819,
+      "step": 8150
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.8093065693430664e-05,
+      "loss": 0.1021,
+      "step": 8152
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.80925977915029e-05,
+      "loss": 0.0664,
+      "step": 8154
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.809212988957515e-05,
+      "loss": 0.074,
+      "step": 8156
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.809166198764739e-05,
+      "loss": 0.0618,
+      "step": 8158
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8091194085719634e-05,
+      "loss": 0.0829,
+      "step": 8160
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.809072618379188e-05,
+      "loss": 0.0654,
+      "step": 8162
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8090258281864126e-05,
+      "loss": 0.0744,
+      "step": 8164
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8089790379936365e-05,
+      "loss": 0.083,
+      "step": 8166
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.808932247800861e-05,
+      "loss": 0.0602,
+      "step": 8168
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.808885457608086e-05,
+      "loss": 0.0708,
+      "step": 8170
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.80883866741531e-05,
+      "loss": 0.0772,
+      "step": 8172
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.808791877222534e-05,
+      "loss": 0.0678,
+      "step": 8174
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.808745087029759e-05,
+      "loss": 0.0593,
+      "step": 8176
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8086982968369834e-05,
+      "loss": 0.0697,
+      "step": 8178
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.808651506644208e-05,
+      "loss": 0.0583,
+      "step": 8180
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.808604716451432e-05,
+      "loss": 0.1039,
+      "step": 8182
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8085579262586565e-05,
+      "loss": 0.0736,
+      "step": 8184
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.808511136065881e-05,
+      "loss": 0.0623,
+      "step": 8186
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8084643458731057e-05,
+      "loss": 0.0537,
+      "step": 8188
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8084175556803296e-05,
+      "loss": 0.0584,
+      "step": 8190
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.808370765487554e-05,
+      "loss": 0.0542,
+      "step": 8192
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.808323975294778e-05,
+      "loss": 0.0765,
+      "step": 8194
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8082771851020033e-05,
+      "loss": 0.0796,
+      "step": 8196
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.808230394909227e-05,
+      "loss": 0.0519,
+      "step": 8198
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.808183604716452e-05,
+      "loss": 0.0654,
+      "step": 8200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.808136814523676e-05,
+      "loss": 0.0686,
+      "step": 8202
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.808090024330901e-05,
+      "loss": 0.0627,
+      "step": 8204
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.808043234138125e-05,
+      "loss": 0.0727,
+      "step": 8206
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8079964439453495e-05,
+      "loss": 0.0523,
+      "step": 8208
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8079496537525735e-05,
+      "loss": 0.0563,
+      "step": 8210
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.807902863559798e-05,
+      "loss": 0.0895,
+      "step": 8212
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8078560733670226e-05,
+      "loss": 0.0592,
+      "step": 8214
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.807809283174247e-05,
+      "loss": 0.0438,
+      "step": 8216
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.807762492981471e-05,
+      "loss": 0.0556,
+      "step": 8218
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.807715702788696e-05,
+      "loss": 0.063,
+      "step": 8220
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.80766891259592e-05,
+      "loss": 0.0833,
+      "step": 8222
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.807622122403145e-05,
+      "loss": 0.0723,
+      "step": 8224
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.807575332210369e-05,
+      "loss": 0.051,
+      "step": 8226
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8075285420175934e-05,
+      "loss": 0.0813,
+      "step": 8228
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.807481751824818e-05,
+      "loss": 0.0658,
+      "step": 8230
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.807434961632042e-05,
+      "loss": 0.0587,
+      "step": 8232
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.8073881714392665e-05,
+      "loss": 0.0742,
+      "step": 8234
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.8073413812464904e-05,
+      "loss": 0.0613,
+      "step": 8236
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.807294591053716e-05,
+      "loss": 0.0605,
+      "step": 8238
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.8072478008609396e-05,
+      "loss": 0.0686,
+      "step": 8240
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.807201010668164e-05,
+      "loss": 0.0761,
+      "step": 8242
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.807154220475388e-05,
+      "loss": 0.0635,
+      "step": 8244
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.807107430282613e-05,
+      "loss": 0.0915,
+      "step": 8246
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.807060640089837e-05,
+      "loss": 0.0766,
+      "step": 8248
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.807013849897062e-05,
+      "loss": 0.0665,
+      "step": 8250
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.806967059704286e-05,
+      "loss": 0.0415,
+      "step": 8252
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.8069202695115104e-05,
+      "loss": 0.0925,
+      "step": 8254
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.806873479318735e-05,
+      "loss": 0.0664,
+      "step": 8256
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.8068266891259596e-05,
+      "loss": 0.0654,
+      "step": 8258
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.8067798989331835e-05,
+      "loss": 0.0634,
+      "step": 8260
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.806733108740408e-05,
+      "loss": 0.0652,
+      "step": 8262
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.806686318547633e-05,
+      "loss": 0.0618,
+      "step": 8264
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.806639528354857e-05,
+      "loss": 0.0875,
+      "step": 8266
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.806592738162081e-05,
+      "loss": 0.063,
+      "step": 8268
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.806545947969306e-05,
+      "loss": 0.0749,
+      "step": 8270
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.8064991577765304e-05,
+      "loss": 0.0755,
+      "step": 8272
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.806452367583755e-05,
+      "loss": 0.0548,
+      "step": 8274
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.806405577390979e-05,
+      "loss": 0.0591,
+      "step": 8276
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.8063587871982035e-05,
+      "loss": 0.0816,
+      "step": 8278
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.8063119970054274e-05,
+      "loss": 0.0765,
+      "step": 8280
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.8062652068126526e-05,
+      "loss": 0.0692,
+      "step": 8282
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.8062184166198766e-05,
+      "loss": 0.0758,
+      "step": 8284
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.806171626427101e-05,
+      "loss": 0.0773,
+      "step": 8286
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.806124836234325e-05,
+      "loss": 0.0591,
+      "step": 8288
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.80607804604155e-05,
+      "loss": 0.0695,
+      "step": 8290
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.806031255848774e-05,
+      "loss": 0.0794,
+      "step": 8292
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.805984465655999e-05,
+      "loss": 0.0875,
+      "step": 8294
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.805937675463223e-05,
+      "loss": 0.0798,
+      "step": 8296
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.805890885270447e-05,
+      "loss": 0.0538,
+      "step": 8298
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.805844095077672e-05,
+      "loss": 0.0534,
+      "step": 8300
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.8057973048848965e-05,
+      "loss": 0.064,
+      "step": 8302
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.8057505146921204e-05,
+      "loss": 0.0612,
+      "step": 8304
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.805703724499345e-05,
+      "loss": 0.0589,
+      "step": 8306
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.8056569343065696e-05,
+      "loss": 0.0721,
+      "step": 8308
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.805610144113794e-05,
+      "loss": 0.0675,
+      "step": 8310
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.805563353921018e-05,
+      "loss": 0.0714,
+      "step": 8312
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.805516563728243e-05,
+      "loss": 0.069,
+      "step": 8314
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.805469773535467e-05,
+      "loss": 0.09,
+      "step": 8316
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.805422983342692e-05,
+      "loss": 0.0664,
+      "step": 8318
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.805376193149916e-05,
+      "loss": 0.0713,
+      "step": 8320
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.8053294029571404e-05,
+      "loss": 0.0632,
+      "step": 8322
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.805282612764364e-05,
+      "loss": 0.0608,
+      "step": 8324
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.8052358225715896e-05,
+      "loss": 0.0491,
+      "step": 8326
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.8051890323788135e-05,
+      "loss": 0.0648,
+      "step": 8328
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.805142242186038e-05,
+      "loss": 0.0589,
+      "step": 8330
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.805095451993262e-05,
+      "loss": 0.0694,
+      "step": 8332
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.805048661800487e-05,
+      "loss": 0.0937,
+      "step": 8334
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.805001871607711e-05,
+      "loss": 0.0812,
+      "step": 8336
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.804955081414936e-05,
+      "loss": 0.0723,
+      "step": 8338
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.80490829122216e-05,
+      "loss": 0.0675,
+      "step": 8340
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.804861501029384e-05,
+      "loss": 0.0632,
+      "step": 8342
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.804814710836609e-05,
+      "loss": 0.0716,
+      "step": 8344
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.8047679206438335e-05,
+      "loss": 0.0603,
+      "step": 8346
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.8047211304510574e-05,
+      "loss": 0.0534,
+      "step": 8348
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.804674340258282e-05,
+      "loss": 0.0696,
+      "step": 8350
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.8046275500655066e-05,
+      "loss": 0.058,
+      "step": 8352
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.804580759872731e-05,
+      "loss": 0.0739,
+      "step": 8354
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.804533969679955e-05,
+      "loss": 0.0697,
+      "step": 8356
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.8044871794871796e-05,
+      "loss": 0.0582,
+      "step": 8358
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.804440389294404e-05,
+      "loss": 0.0533,
+      "step": 8360
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.804393599101629e-05,
+      "loss": 0.086,
+      "step": 8362
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.804346808908853e-05,
+      "loss": 0.0734,
+      "step": 8364
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.804300018716077e-05,
+      "loss": 0.0668,
+      "step": 8366
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.804253228523302e-05,
+      "loss": 0.0731,
+      "step": 8368
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.8042064383305265e-05,
+      "loss": 0.0651,
+      "step": 8370
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.8041596481377504e-05,
+      "loss": 0.0637,
+      "step": 8372
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.804112857944975e-05,
+      "loss": 0.0726,
+      "step": 8374
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.804066067752199e-05,
+      "loss": 0.0624,
+      "step": 8376
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.804019277559424e-05,
+      "loss": 0.0543,
+      "step": 8378
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.803972487366648e-05,
+      "loss": 0.0877,
+      "step": 8380
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.803925697173873e-05,
+      "loss": 0.0859,
+      "step": 8382
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.8038789069810966e-05,
+      "loss": 0.0646,
+      "step": 8384
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.803832116788322e-05,
+      "loss": 0.0769,
+      "step": 8386
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.803785326595546e-05,
+      "loss": 0.0606,
+      "step": 8388
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.8037385364027704e-05,
+      "loss": 0.0608,
+      "step": 8390
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.803691746209994e-05,
+      "loss": 0.0569,
+      "step": 8392
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.803644956017219e-05,
+      "loss": 0.0498,
+      "step": 8394
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.8035981658244435e-05,
+      "loss": 0.0683,
+      "step": 8396
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.803551375631668e-05,
+      "loss": 0.0722,
+      "step": 8398
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.803504585438892e-05,
+      "loss": 0.0843,
+      "step": 8400
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.8034577952461166e-05,
+      "loss": 0.096,
+      "step": 8402
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.803411005053341e-05,
+      "loss": 0.072,
+      "step": 8404
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.803364214860566e-05,
+      "loss": 0.0681,
+      "step": 8406
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.80331742466779e-05,
+      "loss": 0.0835,
+      "step": 8408
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.803270634475014e-05,
+      "loss": 0.0779,
+      "step": 8410
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.803223844282239e-05,
+      "loss": 0.087,
+      "step": 8412
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.8031770540894635e-05,
+      "loss": 0.0525,
+      "step": 8414
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.8031302638966874e-05,
+      "loss": 0.0683,
+      "step": 8416
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.803083473703912e-05,
+      "loss": 0.0614,
+      "step": 8418
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.8030366835111366e-05,
+      "loss": 0.0683,
+      "step": 8420
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.802989893318361e-05,
+      "loss": 0.0632,
+      "step": 8422
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.802943103125585e-05,
+      "loss": 0.0556,
+      "step": 8424
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.8028963129328097e-05,
+      "loss": 0.0691,
+      "step": 8426
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.8028495227400336e-05,
+      "loss": 0.0858,
+      "step": 8428
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.802802732547259e-05,
+      "loss": 0.0655,
+      "step": 8430
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.802755942354483e-05,
+      "loss": 0.0626,
+      "step": 8432
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.802709152161707e-05,
+      "loss": 0.0795,
+      "step": 8434
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.802662361968931e-05,
+      "loss": 0.0672,
+      "step": 8436
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.802615571776156e-05,
+      "loss": 0.0803,
+      "step": 8438
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.8025687815833804e-05,
+      "loss": 0.0785,
+      "step": 8440
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.802521991390605e-05,
+      "loss": 0.0689,
+      "step": 8442
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.802475201197829e-05,
+      "loss": 0.0563,
+      "step": 8444
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.8024284110050535e-05,
+      "loss": 0.0754,
+      "step": 8446
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.802381620812278e-05,
+      "loss": 0.0804,
+      "step": 8448
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.802334830619503e-05,
+      "loss": 0.0577,
+      "step": 8450
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.8022880404267266e-05,
+      "loss": 0.0665,
+      "step": 8452
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.802241250233951e-05,
+      "loss": 0.062,
+      "step": 8454
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.802194460041176e-05,
+      "loss": 0.0794,
+      "step": 8456
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.8021476698484004e-05,
+      "loss": 0.0532,
+      "step": 8458
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.802100879655624e-05,
+      "loss": 0.0724,
+      "step": 8460
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.802054089462849e-05,
+      "loss": 0.0757,
+      "step": 8462
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.8020072992700735e-05,
+      "loss": 0.0747,
+      "step": 8464
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.801960509077298e-05,
+      "loss": 0.0794,
+      "step": 8466
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.801913718884522e-05,
+      "loss": 0.0628,
+      "step": 8468
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.8018669286917466e-05,
+      "loss": 0.0595,
+      "step": 8470
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.8018201384989705e-05,
+      "loss": 0.0721,
+      "step": 8472
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.801773348306196e-05,
+      "loss": 0.0644,
+      "step": 8474
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.80172655811342e-05,
+      "loss": 0.0698,
+      "step": 8476
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.801679767920644e-05,
+      "loss": 0.0772,
+      "step": 8478
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.801632977727868e-05,
+      "loss": 0.0635,
+      "step": 8480
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.8015861875350935e-05,
+      "loss": 0.0813,
+      "step": 8482
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.8015393973423174e-05,
+      "loss": 0.0753,
+      "step": 8484
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.801492607149541e-05,
+      "loss": 0.0805,
+      "step": 8486
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.801445816956766e-05,
+      "loss": 0.0754,
+      "step": 8488
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.8013990267639905e-05,
+      "loss": 0.0578,
+      "step": 8490
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.801352236571215e-05,
+      "loss": 0.0782,
+      "step": 8492
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.801305446378439e-05,
+      "loss": 0.0611,
+      "step": 8494
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.8012586561856636e-05,
+      "loss": 0.0837,
+      "step": 8496
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.801211865992888e-05,
+      "loss": 0.0929,
+      "step": 8498
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.801165075800113e-05,
+      "loss": 0.0574,
+      "step": 8500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.801118285607337e-05,
+      "loss": 0.0785,
+      "step": 8502
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.801071495414561e-05,
+      "loss": 0.0776,
+      "step": 8504
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.801024705221785e-05,
+      "loss": 0.0673,
+      "step": 8506
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.8009779150290104e-05,
+      "loss": 0.0857,
+      "step": 8508
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.8009311248362343e-05,
+      "loss": 0.0825,
+      "step": 8510
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.800884334643459e-05,
+      "loss": 0.0576,
+      "step": 8512
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.800837544450683e-05,
+      "loss": 0.0612,
+      "step": 8514
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.800790754257908e-05,
+      "loss": 0.0561,
+      "step": 8516
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.800743964065132e-05,
+      "loss": 0.0654,
+      "step": 8518
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.8006971738723566e-05,
+      "loss": 0.0605,
+      "step": 8520
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.8006503836795805e-05,
+      "loss": 0.0577,
+      "step": 8522
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.800603593486805e-05,
+      "loss": 0.0814,
+      "step": 8524
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.80055680329403e-05,
+      "loss": 0.0724,
+      "step": 8526
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.800510013101254e-05,
+      "loss": 0.0654,
+      "step": 8528
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.800463222908478e-05,
+      "loss": 0.0666,
+      "step": 8530
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.800416432715703e-05,
+      "loss": 0.0558,
+      "step": 8532
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.8003696425229274e-05,
+      "loss": 0.0732,
+      "step": 8534
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.800322852330152e-05,
+      "loss": 0.0567,
+      "step": 8536
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.800276062137376e-05,
+      "loss": 0.0474,
+      "step": 8538
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.8002292719446005e-05,
+      "loss": 0.0519,
+      "step": 8540
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.800182481751825e-05,
+      "loss": 0.0582,
+      "step": 8542
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.80013569155905e-05,
+      "loss": 0.0762,
+      "step": 8544
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.8000889013662736e-05,
+      "loss": 0.0652,
+      "step": 8546
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.800042111173498e-05,
+      "loss": 0.0581,
+      "step": 8548
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.799995320980723e-05,
+      "loss": 0.0561,
+      "step": 8550
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.7999485307879474e-05,
+      "loss": 0.0744,
+      "step": 8552
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.799901740595171e-05,
+      "loss": 0.0756,
+      "step": 8554
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.799854950402396e-05,
+      "loss": 0.0574,
+      "step": 8556
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.79980816020962e-05,
+      "loss": 0.0673,
+      "step": 8558
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.799761370016845e-05,
+      "loss": 0.112,
+      "step": 8560
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.799714579824069e-05,
+      "loss": 0.0745,
+      "step": 8562
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.7996677896312936e-05,
+      "loss": 0.0649,
+      "step": 8564
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.7996209994385175e-05,
+      "loss": 0.0651,
+      "step": 8566
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.799574209245743e-05,
+      "loss": 0.0765,
+      "step": 8568
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.799527419052967e-05,
+      "loss": 0.0595,
+      "step": 8570
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.799480628860191e-05,
+      "loss": 0.0711,
+      "step": 8572
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.799433838667415e-05,
+      "loss": 0.0622,
+      "step": 8574
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.79938704847464e-05,
+      "loss": 0.0687,
+      "step": 8576
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.7993402582818644e-05,
+      "loss": 0.0675,
+      "step": 8578
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.799293468089089e-05,
+      "loss": 0.0708,
+      "step": 8580
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.799246677896313e-05,
+      "loss": 0.0816,
+      "step": 8582
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.7991998877035374e-05,
+      "loss": 0.0767,
+      "step": 8584
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.799153097510762e-05,
+      "loss": 0.0521,
+      "step": 8586
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7991063073179866e-05,
+      "loss": 0.0607,
+      "step": 8588
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7990595171252105e-05,
+      "loss": 0.0895,
+      "step": 8590
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.799012726932435e-05,
+      "loss": 0.0655,
+      "step": 8592
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.79896593673966e-05,
+      "loss": 0.068,
+      "step": 8594
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.798919146546884e-05,
+      "loss": 0.0752,
+      "step": 8596
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.798872356354108e-05,
+      "loss": 0.0816,
+      "step": 8598
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.798825566161333e-05,
+      "loss": 0.0827,
+      "step": 8600
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.798778775968557e-05,
+      "loss": 0.0787,
+      "step": 8602
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.798731985775782e-05,
+      "loss": 0.0548,
+      "step": 8604
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.798685195583006e-05,
+      "loss": 0.0545,
+      "step": 8606
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7986384053902305e-05,
+      "loss": 0.073,
+      "step": 8608
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7985916151974544e-05,
+      "loss": 0.0662,
+      "step": 8610
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.79854482500468e-05,
+      "loss": 0.1088,
+      "step": 8612
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7984980348119036e-05,
+      "loss": 0.0798,
+      "step": 8614
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.798451244619128e-05,
+      "loss": 0.0667,
+      "step": 8616
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.798404454426352e-05,
+      "loss": 0.0683,
+      "step": 8618
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.798357664233577e-05,
+      "loss": 0.0882,
+      "step": 8620
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.798310874040801e-05,
+      "loss": 0.0718,
+      "step": 8622
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.798264083848026e-05,
+      "loss": 0.0746,
+      "step": 8624
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.79821729365525e-05,
+      "loss": 0.0832,
+      "step": 8626
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7981705034624744e-05,
+      "loss": 0.0532,
+      "step": 8628
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.798123713269699e-05,
+      "loss": 0.0666,
+      "step": 8630
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7980769230769236e-05,
+      "loss": 0.0945,
+      "step": 8632
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7980301328841475e-05,
+      "loss": 0.0786,
+      "step": 8634
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.797983342691372e-05,
+      "loss": 0.0674,
+      "step": 8636
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.797936552498597e-05,
+      "loss": 0.0609,
+      "step": 8638
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.797889762305821e-05,
+      "loss": 0.072,
+      "step": 8640
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.797842972113045e-05,
+      "loss": 0.0711,
+      "step": 8642
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.79779618192027e-05,
+      "loss": 0.0527,
+      "step": 8644
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7977493917274944e-05,
+      "loss": 0.0668,
+      "step": 8646
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.797702601534719e-05,
+      "loss": 0.0771,
+      "step": 8648
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.797655811341943e-05,
+      "loss": 0.0657,
+      "step": 8650
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7976090211491674e-05,
+      "loss": 0.0763,
+      "step": 8652
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.7975622309563914e-05,
+      "loss": 0.0674,
+      "step": 8654
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7975154407636166e-05,
+      "loss": 0.0775,
+      "step": 8656
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7974686505708405e-05,
+      "loss": 0.0635,
+      "step": 8658
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.797421860378065e-05,
+      "loss": 0.0639,
+      "step": 8660
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.797375070185289e-05,
+      "loss": 0.0586,
+      "step": 8662
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.797328279992514e-05,
+      "loss": 0.0707,
+      "step": 8664
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.797281489799738e-05,
+      "loss": 0.0516,
+      "step": 8666
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.797234699606963e-05,
+      "loss": 0.0704,
+      "step": 8668
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.797187909414187e-05,
+      "loss": 0.0894,
+      "step": 8670
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.797141119221411e-05,
+      "loss": 0.0601,
+      "step": 8672
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.797094329028636e-05,
+      "loss": 0.0699,
+      "step": 8674
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7970475388358605e-05,
+      "loss": 0.0617,
+      "step": 8676
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7970007486430844e-05,
+      "loss": 0.0697,
+      "step": 8678
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.796953958450309e-05,
+      "loss": 0.0829,
+      "step": 8680
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7969071682575336e-05,
+      "loss": 0.0682,
+      "step": 8682
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.796860378064758e-05,
+      "loss": 0.0778,
+      "step": 8684
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.796813587871982e-05,
+      "loss": 0.0773,
+      "step": 8686
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.796766797679207e-05,
+      "loss": 0.0738,
+      "step": 8688
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.796720007486431e-05,
+      "loss": 0.0564,
+      "step": 8690
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.796673217293656e-05,
+      "loss": 0.0794,
+      "step": 8692
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.79662642710088e-05,
+      "loss": 0.0731,
+      "step": 8694
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7965796369081044e-05,
+      "loss": 0.0756,
+      "step": 8696
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.796532846715329e-05,
+      "loss": 0.0585,
+      "step": 8698
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7964860565225536e-05,
+      "loss": 0.065,
+      "step": 8700
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7964392663297775e-05,
+      "loss": 0.0476,
+      "step": 8702
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.796392476137002e-05,
+      "loss": 0.0625,
+      "step": 8704
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.796345685944226e-05,
+      "loss": 0.0746,
+      "step": 8706
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.796298895751451e-05,
+      "loss": 0.0774,
+      "step": 8708
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.796252105558675e-05,
+      "loss": 0.0473,
+      "step": 8710
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7962053153659e-05,
+      "loss": 0.0834,
+      "step": 8712
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.796158525173124e-05,
+      "loss": 0.0597,
+      "step": 8714
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.796111734980348e-05,
+      "loss": 0.0669,
+      "step": 8716
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.796064944787573e-05,
+      "loss": 0.0567,
+      "step": 8718
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7960181545947974e-05,
+      "loss": 0.0691,
+      "step": 8720
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7959713644020214e-05,
+      "loss": 0.0713,
+      "step": 8722
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.795924574209246e-05,
+      "loss": 0.0685,
+      "step": 8724
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.7958777840164705e-05,
+      "loss": 0.0747,
+      "step": 8726
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.795830993823695e-05,
+      "loss": 0.0593,
+      "step": 8728
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.795784203630919e-05,
+      "loss": 0.0783,
+      "step": 8730
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7957374134381436e-05,
+      "loss": 0.0683,
+      "step": 8732
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.795690623245368e-05,
+      "loss": 0.0755,
+      "step": 8734
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.795643833052593e-05,
+      "loss": 0.0863,
+      "step": 8736
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.795597042859817e-05,
+      "loss": 0.0483,
+      "step": 8738
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7955502526670407e-05,
+      "loss": 0.0703,
+      "step": 8740
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.795503462474266e-05,
+      "loss": 0.0563,
+      "step": 8742
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.79545667228149e-05,
+      "loss": 0.0479,
+      "step": 8744
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7954098820887144e-05,
+      "loss": 0.0641,
+      "step": 8746
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.795363091895938e-05,
+      "loss": 0.0904,
+      "step": 8748
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.795316301703163e-05,
+      "loss": 0.089,
+      "step": 8750
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7952695115103875e-05,
+      "loss": 0.0574,
+      "step": 8752
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.795222721317612e-05,
+      "loss": 0.0807,
+      "step": 8754
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.795175931124836e-05,
+      "loss": 0.0497,
+      "step": 8756
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7951291409320606e-05,
+      "loss": 0.1108,
+      "step": 8758
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.795082350739285e-05,
+      "loss": 0.0649,
+      "step": 8760
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.79503556054651e-05,
+      "loss": 0.0619,
+      "step": 8762
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.794988770353734e-05,
+      "loss": 0.0562,
+      "step": 8764
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.794941980160958e-05,
+      "loss": 0.0662,
+      "step": 8766
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.794895189968183e-05,
+      "loss": 0.0562,
+      "step": 8768
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7948483997754075e-05,
+      "loss": 0.08,
+      "step": 8770
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7948016095826314e-05,
+      "loss": 0.0883,
+      "step": 8772
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.794754819389856e-05,
+      "loss": 0.0709,
+      "step": 8774
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7947080291970806e-05,
+      "loss": 0.0658,
+      "step": 8776
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.794661239004305e-05,
+      "loss": 0.0704,
+      "step": 8778
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.794614448811529e-05,
+      "loss": 0.0966,
+      "step": 8780
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.794567658618754e-05,
+      "loss": 0.081,
+      "step": 8782
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7945208684259776e-05,
+      "loss": 0.0678,
+      "step": 8784
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.794474078233203e-05,
+      "loss": 0.074,
+      "step": 8786
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.794427288040427e-05,
+      "loss": 0.0727,
+      "step": 8788
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7943804978476514e-05,
+      "loss": 0.0591,
+      "step": 8790
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.794333707654875e-05,
+      "loss": 0.095,
+      "step": 8792
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7942869174621005e-05,
+      "loss": 0.064,
+      "step": 8794
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.7942401272693245e-05,
+      "loss": 0.0724,
+      "step": 8796
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.794193337076549e-05,
+      "loss": 0.0756,
+      "step": 8798
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.794146546883773e-05,
+      "loss": 0.0755,
+      "step": 8800
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7940997566909976e-05,
+      "loss": 0.08,
+      "step": 8802
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.794052966498222e-05,
+      "loss": 0.074,
+      "step": 8804
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.794006176305447e-05,
+      "loss": 0.0783,
+      "step": 8806
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7939593861126707e-05,
+      "loss": 0.06,
+      "step": 8808
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.793912595919895e-05,
+      "loss": 0.0711,
+      "step": 8810
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.79386580572712e-05,
+      "loss": 0.0625,
+      "step": 8812
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7938190155343444e-05,
+      "loss": 0.0793,
+      "step": 8814
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7937722253415683e-05,
+      "loss": 0.0832,
+      "step": 8816
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.793725435148793e-05,
+      "loss": 0.0637,
+      "step": 8818
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7936786449560175e-05,
+      "loss": 0.0923,
+      "step": 8820
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.793631854763242e-05,
+      "loss": 0.0939,
+      "step": 8822
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.793585064570466e-05,
+      "loss": 0.0704,
+      "step": 8824
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7935382743776906e-05,
+      "loss": 0.0715,
+      "step": 8826
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.793491484184915e-05,
+      "loss": 0.0708,
+      "step": 8828
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.79344469399214e-05,
+      "loss": 0.0725,
+      "step": 8830
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.793397903799364e-05,
+      "loss": 0.0567,
+      "step": 8832
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.793351113606588e-05,
+      "loss": 0.087,
+      "step": 8834
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.793304323413812e-05,
+      "loss": 0.074,
+      "step": 8836
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7932575332210375e-05,
+      "loss": 0.0677,
+      "step": 8838
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7932107430282614e-05,
+      "loss": 0.0722,
+      "step": 8840
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.793163952835486e-05,
+      "loss": 0.0724,
+      "step": 8842
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.79311716264271e-05,
+      "loss": 0.0814,
+      "step": 8844
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.793070372449935e-05,
+      "loss": 0.0862,
+      "step": 8846
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.793023582257159e-05,
+      "loss": 0.0586,
+      "step": 8848
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.792976792064384e-05,
+      "loss": 0.0689,
+      "step": 8850
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7929300018716076e-05,
+      "loss": 0.0714,
+      "step": 8852
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.792883211678832e-05,
+      "loss": 0.0742,
+      "step": 8854
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.792836421486057e-05,
+      "loss": 0.0689,
+      "step": 8856
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7927896312932814e-05,
+      "loss": 0.0753,
+      "step": 8858
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.792742841100505e-05,
+      "loss": 0.0691,
+      "step": 8860
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.79269605090773e-05,
+      "loss": 0.065,
+      "step": 8862
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.7926492607149545e-05,
+      "loss": 0.0636,
+      "step": 8864
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.792602470522179e-05,
+      "loss": 0.0698,
+      "step": 8866
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.792555680329403e-05,
+      "loss": 0.0668,
+      "step": 8868
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7925088901366276e-05,
+      "loss": 0.0519,
+      "step": 8870
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.792462099943852e-05,
+      "loss": 0.0669,
+      "step": 8872
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.792415309751077e-05,
+      "loss": 0.0521,
+      "step": 8874
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7923685195583007e-05,
+      "loss": 0.0697,
+      "step": 8876
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.792321729365525e-05,
+      "loss": 0.0792,
+      "step": 8878
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.79227493917275e-05,
+      "loss": 0.1011,
+      "step": 8880
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7922281489799744e-05,
+      "loss": 0.078,
+      "step": 8882
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7921813587871983e-05,
+      "loss": 0.0777,
+      "step": 8884
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.792134568594423e-05,
+      "loss": 0.0759,
+      "step": 8886
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.792087778401647e-05,
+      "loss": 0.0692,
+      "step": 8888
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.792040988208872e-05,
+      "loss": 0.1084,
+      "step": 8890
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.791994198016096e-05,
+      "loss": 0.07,
+      "step": 8892
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7919474078233206e-05,
+      "loss": 0.0706,
+      "step": 8894
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7919006176305445e-05,
+      "loss": 0.0777,
+      "step": 8896
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.791853827437769e-05,
+      "loss": 0.0781,
+      "step": 8898
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.791807037244994e-05,
+      "loss": 0.0655,
+      "step": 8900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.791760247052218e-05,
+      "loss": 0.0752,
+      "step": 8902
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.791713456859442e-05,
+      "loss": 0.0952,
+      "step": 8904
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.791666666666667e-05,
+      "loss": 0.0797,
+      "step": 8906
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7916198764738914e-05,
+      "loss": 0.0725,
+      "step": 8908
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.791573086281116e-05,
+      "loss": 0.0733,
+      "step": 8910
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.79152629608834e-05,
+      "loss": 0.0604,
+      "step": 8912
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7914795058955645e-05,
+      "loss": 0.0751,
+      "step": 8914
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.791432715702789e-05,
+      "loss": 0.0783,
+      "step": 8916
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.791385925510014e-05,
+      "loss": 0.0816,
+      "step": 8918
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7913391353172376e-05,
+      "loss": 0.063,
+      "step": 8920
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.791292345124462e-05,
+      "loss": 0.0801,
+      "step": 8922
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.791245554931687e-05,
+      "loss": 0.0831,
+      "step": 8924
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7911987647389114e-05,
+      "loss": 0.085,
+      "step": 8926
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.791151974546135e-05,
+      "loss": 0.0568,
+      "step": 8928
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.79110518435336e-05,
+      "loss": 0.0749,
+      "step": 8930
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.791058394160584e-05,
+      "loss": 0.0779,
+      "step": 8932
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.791011603967809e-05,
+      "loss": 0.0623,
+      "step": 8934
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.790964813775033e-05,
+      "loss": 0.0537,
+      "step": 8936
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7909180235822576e-05,
+      "loss": 0.0671,
+      "step": 8938
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.7908712333894815e-05,
+      "loss": 0.0853,
+      "step": 8940
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.790824443196707e-05,
+      "loss": 0.0844,
+      "step": 8942
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7907776530039307e-05,
+      "loss": 0.0767,
+      "step": 8944
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.790730862811155e-05,
+      "loss": 0.0777,
+      "step": 8946
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.790684072618379e-05,
+      "loss": 0.0656,
+      "step": 8948
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.790637282425604e-05,
+      "loss": 0.0875,
+      "step": 8950
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7905904922328283e-05,
+      "loss": 0.0703,
+      "step": 8952
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.790543702040053e-05,
+      "loss": 0.0772,
+      "step": 8954
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.790496911847277e-05,
+      "loss": 0.082,
+      "step": 8956
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7904501216545014e-05,
+      "loss": 0.0669,
+      "step": 8958
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.790403331461726e-05,
+      "loss": 0.0686,
+      "step": 8960
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7903565412689506e-05,
+      "loss": 0.061,
+      "step": 8962
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7903097510761745e-05,
+      "loss": 0.0885,
+      "step": 8964
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.790262960883399e-05,
+      "loss": 0.0773,
+      "step": 8966
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.790216170690624e-05,
+      "loss": 0.0892,
+      "step": 8968
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.790169380497848e-05,
+      "loss": 0.083,
+      "step": 8970
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.790122590305072e-05,
+      "loss": 0.0689,
+      "step": 8972
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.790075800112297e-05,
+      "loss": 0.0794,
+      "step": 8974
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7900290099195214e-05,
+      "loss": 0.0811,
+      "step": 8976
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.789982219726746e-05,
+      "loss": 0.0799,
+      "step": 8978
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.78993542953397e-05,
+      "loss": 0.0745,
+      "step": 8980
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7898886393411945e-05,
+      "loss": 0.0651,
+      "step": 8982
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7898418491484184e-05,
+      "loss": 0.0807,
+      "step": 8984
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.789795058955644e-05,
+      "loss": 0.0698,
+      "step": 8986
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7897482687628676e-05,
+      "loss": 0.0688,
+      "step": 8988
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.789701478570092e-05,
+      "loss": 0.0598,
+      "step": 8990
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.789654688377316e-05,
+      "loss": 0.0674,
+      "step": 8992
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.789607898184541e-05,
+      "loss": 0.0773,
+      "step": 8994
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.789561107991765e-05,
+      "loss": 0.0767,
+      "step": 8996
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.789514317798989e-05,
+      "loss": 0.0646,
+      "step": 8998
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.789467527606214e-05,
+      "loss": 0.0806,
+      "step": 9000
+    },
+    {
+      "epoch": 1.26,
+      "eval_gen_len": 30.1724,
+      "eval_loss": 1.0504204034805298,
+      "eval_meteor": 0.0494,
+      "eval_runtime": 14.954,
+      "eval_samples_per_second": 3.879,
+      "eval_steps_per_second": 0.535,
+      "step": 9000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7894207374134384e-05,
+      "loss": 0.0568,
+      "step": 9002
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.789373947220663e-05,
+      "loss": 0.0617,
+      "step": 9004
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.789327157027887e-05,
+      "loss": 0.0897,
+      "step": 9006
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.7892803668351115e-05,
+      "loss": 0.0402,
+      "step": 9008
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.789233576642336e-05,
+      "loss": 0.0854,
+      "step": 9010
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7891867864495607e-05,
+      "loss": 0.0905,
+      "step": 9012
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7891399962567846e-05,
+      "loss": 0.0734,
+      "step": 9014
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.789093206064009e-05,
+      "loss": 0.0816,
+      "step": 9016
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.789046415871233e-05,
+      "loss": 0.0621,
+      "step": 9018
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7889996256784583e-05,
+      "loss": 0.0782,
+      "step": 9020
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.788952835485682e-05,
+      "loss": 0.0572,
+      "step": 9022
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.788906045292907e-05,
+      "loss": 0.0701,
+      "step": 9024
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.788859255100131e-05,
+      "loss": 0.0729,
+      "step": 9026
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7888124649073554e-05,
+      "loss": 0.0794,
+      "step": 9028
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.78876567471458e-05,
+      "loss": 0.0732,
+      "step": 9030
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7887188845218045e-05,
+      "loss": 0.1045,
+      "step": 9032
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7886720943290285e-05,
+      "loss": 0.0823,
+      "step": 9034
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.788625304136253e-05,
+      "loss": 0.0742,
+      "step": 9036
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7885785139434776e-05,
+      "loss": 0.0572,
+      "step": 9038
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.788531723750702e-05,
+      "loss": 0.0746,
+      "step": 9040
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.788484933557926e-05,
+      "loss": 0.0704,
+      "step": 9042
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.788438143365151e-05,
+      "loss": 0.1047,
+      "step": 9044
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.788391353172375e-05,
+      "loss": 0.0751,
+      "step": 9046
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7883445629796e-05,
+      "loss": 0.0787,
+      "step": 9048
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.788297772786824e-05,
+      "loss": 0.0724,
+      "step": 9050
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7882509825940484e-05,
+      "loss": 0.0762,
+      "step": 9052
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.788204192401273e-05,
+      "loss": 0.074,
+      "step": 9054
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7881574022084976e-05,
+      "loss": 0.0829,
+      "step": 9056
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7881106120157215e-05,
+      "loss": 0.0561,
+      "step": 9058
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.788063821822946e-05,
+      "loss": 0.0567,
+      "step": 9060
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.78801703163017e-05,
+      "loss": 0.0665,
+      "step": 9062
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.787970241437395e-05,
+      "loss": 0.0696,
+      "step": 9064
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.787923451244619e-05,
+      "loss": 0.0614,
+      "step": 9066
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.787876661051844e-05,
+      "loss": 0.0622,
+      "step": 9068
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.787829870859068e-05,
+      "loss": 0.0754,
+      "step": 9070
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.787783080666293e-05,
+      "loss": 0.0658,
+      "step": 9072
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.787736290473517e-05,
+      "loss": 0.0959,
+      "step": 9074
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7876895002807415e-05,
+      "loss": 0.0837,
+      "step": 9076
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7876427100879654e-05,
+      "loss": 0.0675,
+      "step": 9078
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.78759591989519e-05,
+      "loss": 0.0754,
+      "step": 9080
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.7875491297024146e-05,
+      "loss": 0.0684,
+      "step": 9082
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.787502339509639e-05,
+      "loss": 0.1071,
+      "step": 9084
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.787455549316863e-05,
+      "loss": 0.0761,
+      "step": 9086
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.787408759124088e-05,
+      "loss": 0.0935,
+      "step": 9088
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.787361968931312e-05,
+      "loss": 0.0645,
+      "step": 9090
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.787315178738537e-05,
+      "loss": 0.0616,
+      "step": 9092
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.787268388545761e-05,
+      "loss": 0.0762,
+      "step": 9094
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7872215983529854e-05,
+      "loss": 0.0685,
+      "step": 9096
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.78717480816021e-05,
+      "loss": 0.0776,
+      "step": 9098
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7871280179674345e-05,
+      "loss": 0.0725,
+      "step": 9100
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7870812277746585e-05,
+      "loss": 0.0693,
+      "step": 9102
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.787034437581883e-05,
+      "loss": 0.0788,
+      "step": 9104
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7869876473891076e-05,
+      "loss": 0.0564,
+      "step": 9106
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.786940857196332e-05,
+      "loss": 0.0688,
+      "step": 9108
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.786894067003556e-05,
+      "loss": 0.0705,
+      "step": 9110
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.786847276810781e-05,
+      "loss": 0.0418,
+      "step": 9112
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7868004866180046e-05,
+      "loss": 0.0612,
+      "step": 9114
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.78675369642523e-05,
+      "loss": 0.0784,
+      "step": 9116
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.786706906232454e-05,
+      "loss": 0.0913,
+      "step": 9118
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7866601160396784e-05,
+      "loss": 0.0921,
+      "step": 9120
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.786613325846902e-05,
+      "loss": 0.089,
+      "step": 9122
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7865665356541276e-05,
+      "loss": 0.0623,
+      "step": 9124
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7865197454613515e-05,
+      "loss": 0.0994,
+      "step": 9126
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.786472955268576e-05,
+      "loss": 0.0848,
+      "step": 9128
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7864261650758e-05,
+      "loss": 0.0723,
+      "step": 9130
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7863793748830246e-05,
+      "loss": 0.0634,
+      "step": 9132
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.786332584690249e-05,
+      "loss": 0.0634,
+      "step": 9134
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.786285794497474e-05,
+      "loss": 0.0679,
+      "step": 9136
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.786239004304698e-05,
+      "loss": 0.0653,
+      "step": 9138
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.786192214111922e-05,
+      "loss": 0.0844,
+      "step": 9140
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.786145423919147e-05,
+      "loss": 0.0672,
+      "step": 9142
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7860986337263715e-05,
+      "loss": 0.0579,
+      "step": 9144
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7860518435335954e-05,
+      "loss": 0.0592,
+      "step": 9146
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.78600505334082e-05,
+      "loss": 0.0571,
+      "step": 9148
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.7859582631480446e-05,
+      "loss": 0.069,
+      "step": 9150
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.785911472955269e-05,
+      "loss": 0.0774,
+      "step": 9152
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.785864682762493e-05,
+      "loss": 0.1049,
+      "step": 9154
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.785817892569718e-05,
+      "loss": 0.0677,
+      "step": 9156
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.785771102376942e-05,
+      "loss": 0.0734,
+      "step": 9158
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.785724312184167e-05,
+      "loss": 0.1023,
+      "step": 9160
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.785677521991391e-05,
+      "loss": 0.0664,
+      "step": 9162
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7856307317986154e-05,
+      "loss": 0.0645,
+      "step": 9164
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.785583941605839e-05,
+      "loss": 0.0904,
+      "step": 9166
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7855371514130645e-05,
+      "loss": 0.0822,
+      "step": 9168
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7854903612202885e-05,
+      "loss": 0.0716,
+      "step": 9170
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.785443571027513e-05,
+      "loss": 0.0632,
+      "step": 9172
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.785396780834737e-05,
+      "loss": 0.0616,
+      "step": 9174
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7853499906419615e-05,
+      "loss": 0.0643,
+      "step": 9176
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.785303200449186e-05,
+      "loss": 0.0805,
+      "step": 9178
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.785256410256411e-05,
+      "loss": 0.0551,
+      "step": 9180
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7852096200636346e-05,
+      "loss": 0.0611,
+      "step": 9182
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.785162829870859e-05,
+      "loss": 0.0653,
+      "step": 9184
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.785116039678084e-05,
+      "loss": 0.0741,
+      "step": 9186
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7850692494853084e-05,
+      "loss": 0.062,
+      "step": 9188
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.785022459292532e-05,
+      "loss": 0.0667,
+      "step": 9190
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.784975669099757e-05,
+      "loss": 0.0588,
+      "step": 9192
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7849288789069815e-05,
+      "loss": 0.0786,
+      "step": 9194
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.784882088714206e-05,
+      "loss": 0.1006,
+      "step": 9196
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.78483529852143e-05,
+      "loss": 0.0753,
+      "step": 9198
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7847885083286546e-05,
+      "loss": 0.0867,
+      "step": 9200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.784741718135879e-05,
+      "loss": 0.0679,
+      "step": 9202
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.784694927943104e-05,
+      "loss": 0.0838,
+      "step": 9204
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.784648137750328e-05,
+      "loss": 0.0921,
+      "step": 9206
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.784601347557552e-05,
+      "loss": 0.0794,
+      "step": 9208
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.784554557364776e-05,
+      "loss": 0.0599,
+      "step": 9210
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7845077671720015e-05,
+      "loss": 0.0683,
+      "step": 9212
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7844609769792254e-05,
+      "loss": 0.0855,
+      "step": 9214
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.78441418678645e-05,
+      "loss": 0.0653,
+      "step": 9216
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.784367396593674e-05,
+      "loss": 0.0585,
+      "step": 9218
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.784320606400899e-05,
+      "loss": 0.081,
+      "step": 9220
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.784273816208123e-05,
+      "loss": 0.0627,
+      "step": 9222
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.784227026015348e-05,
+      "loss": 0.0702,
+      "step": 9224
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.7841802358225716e-05,
+      "loss": 0.0584,
+      "step": 9226
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.784133445629796e-05,
+      "loss": 0.1109,
+      "step": 9228
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.784086655437021e-05,
+      "loss": 0.0746,
+      "step": 9230
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.7840398652442454e-05,
+      "loss": 0.078,
+      "step": 9232
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.783993075051469e-05,
+      "loss": 0.0719,
+      "step": 9234
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.783946284858694e-05,
+      "loss": 0.059,
+      "step": 9236
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.7838994946659185e-05,
+      "loss": 0.0641,
+      "step": 9238
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.783852704473143e-05,
+      "loss": 0.0597,
+      "step": 9240
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.783805914280367e-05,
+      "loss": 0.0759,
+      "step": 9242
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.783759124087591e-05,
+      "loss": 0.0763,
+      "step": 9244
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.783712333894816e-05,
+      "loss": 0.0799,
+      "step": 9246
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.78366554370204e-05,
+      "loss": 0.0796,
+      "step": 9248
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.7836187535092646e-05,
+      "loss": 0.0719,
+      "step": 9250
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.7835719633164886e-05,
+      "loss": 0.0486,
+      "step": 9252
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.783525173123714e-05,
+      "loss": 0.0624,
+      "step": 9254
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.783478382930938e-05,
+      "loss": 0.073,
+      "step": 9256
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.783431592738162e-05,
+      "loss": 0.057,
+      "step": 9258
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.783384802545386e-05,
+      "loss": 0.073,
+      "step": 9260
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.783338012352611e-05,
+      "loss": 0.0743,
+      "step": 9262
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.7832912221598354e-05,
+      "loss": 0.0986,
+      "step": 9264
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.78324443196706e-05,
+      "loss": 0.0742,
+      "step": 9266
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.783197641774284e-05,
+      "loss": 0.0651,
+      "step": 9268
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.7831508515815085e-05,
+      "loss": 0.08,
+      "step": 9270
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.783104061388733e-05,
+      "loss": 0.0776,
+      "step": 9272
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.783057271195958e-05,
+      "loss": 0.0654,
+      "step": 9274
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.7830104810031816e-05,
+      "loss": 0.0775,
+      "step": 9276
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.782963690810406e-05,
+      "loss": 0.0777,
+      "step": 9278
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.782916900617631e-05,
+      "loss": 0.0784,
+      "step": 9280
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.7828701104248554e-05,
+      "loss": 0.0645,
+      "step": 9282
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.782823320232079e-05,
+      "loss": 0.0714,
+      "step": 9284
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.782776530039304e-05,
+      "loss": 0.0904,
+      "step": 9286
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.7827297398465285e-05,
+      "loss": 0.0768,
+      "step": 9288
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.782682949653753e-05,
+      "loss": 0.0812,
+      "step": 9290
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.782636159460977e-05,
+      "loss": 0.093,
+      "step": 9292
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.7825893692682016e-05,
+      "loss": 0.1154,
+      "step": 9294
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.7825425790754255e-05,
+      "loss": 0.0621,
+      "step": 9296
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.782495788882651e-05,
+      "loss": 0.0908,
+      "step": 9298
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.782448998689875e-05,
+      "loss": 0.0811,
+      "step": 9300
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.782402208497099e-05,
+      "loss": 0.092,
+      "step": 9302
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.782355418304323e-05,
+      "loss": 0.0684,
+      "step": 9304
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.782308628111548e-05,
+      "loss": 0.0721,
+      "step": 9306
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.7822618379187724e-05,
+      "loss": 0.0758,
+      "step": 9308
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.782215047725997e-05,
+      "loss": 0.0707,
+      "step": 9310
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.782168257533221e-05,
+      "loss": 0.0724,
+      "step": 9312
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.7821214673404455e-05,
+      "loss": 0.0972,
+      "step": 9314
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.78207467714767e-05,
+      "loss": 0.0708,
+      "step": 9316
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.7820278869548946e-05,
+      "loss": 0.076,
+      "step": 9318
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.7819810967621186e-05,
+      "loss": 0.0801,
+      "step": 9320
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.781934306569343e-05,
+      "loss": 0.0643,
+      "step": 9322
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.781887516376568e-05,
+      "loss": 0.0598,
+      "step": 9324
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.781840726183792e-05,
+      "loss": 0.0849,
+      "step": 9326
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.781793935991016e-05,
+      "loss": 0.0593,
+      "step": 9328
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.781747145798241e-05,
+      "loss": 0.0835,
+      "step": 9330
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.7817003556054654e-05,
+      "loss": 0.0569,
+      "step": 9332
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.78165356541269e-05,
+      "loss": 0.0801,
+      "step": 9334
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.781606775219914e-05,
+      "loss": 0.0816,
+      "step": 9336
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.7815599850271385e-05,
+      "loss": 0.0702,
+      "step": 9338
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.7815131948343624e-05,
+      "loss": 0.0764,
+      "step": 9340
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.781466404641588e-05,
+      "loss": 0.0907,
+      "step": 9342
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.7814196144488116e-05,
+      "loss": 0.0728,
+      "step": 9344
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.781372824256036e-05,
+      "loss": 0.0776,
+      "step": 9346
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.78132603406326e-05,
+      "loss": 0.0794,
+      "step": 9348
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.7812792438704854e-05,
+      "loss": 0.0685,
+      "step": 9350
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.781232453677709e-05,
+      "loss": 0.0847,
+      "step": 9352
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.781185663484934e-05,
+      "loss": 0.0935,
+      "step": 9354
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.781138873292158e-05,
+      "loss": 0.0531,
+      "step": 9356
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.7810920830993824e-05,
+      "loss": 0.0529,
+      "step": 9358
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.781045292906607e-05,
+      "loss": 0.0577,
+      "step": 9360
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.7809985027138316e-05,
+      "loss": 0.0856,
+      "step": 9362
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.7809517125210555e-05,
+      "loss": 0.0855,
+      "step": 9364
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.78090492232828e-05,
+      "loss": 0.0776,
+      "step": 9366
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.780858132135505e-05,
+      "loss": 0.0795,
+      "step": 9368
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.780811341942729e-05,
+      "loss": 0.0794,
+      "step": 9370
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.780764551749953e-05,
+      "loss": 0.0785,
+      "step": 9372
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.780717761557178e-05,
+      "loss": 0.0669,
+      "step": 9374
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.7806709713644024e-05,
+      "loss": 0.0855,
+      "step": 9376
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.780624181171627e-05,
+      "loss": 0.0793,
+      "step": 9378
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.780577390978851e-05,
+      "loss": 0.0725,
+      "step": 9380
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.7805306007860755e-05,
+      "loss": 0.0715,
+      "step": 9382
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.7804838105933e-05,
+      "loss": 0.0715,
+      "step": 9384
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.7804370204005247e-05,
+      "loss": 0.0826,
+      "step": 9386
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.7803902302077486e-05,
+      "loss": 0.0786,
+      "step": 9388
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.780343440014973e-05,
+      "loss": 0.0659,
+      "step": 9390
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.780296649822197e-05,
+      "loss": 0.0815,
+      "step": 9392
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.780249859629422e-05,
+      "loss": 0.0662,
+      "step": 9394
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.780203069436646e-05,
+      "loss": 0.0941,
+      "step": 9396
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.780156279243871e-05,
+      "loss": 0.0717,
+      "step": 9398
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.780109489051095e-05,
+      "loss": 0.0826,
+      "step": 9400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.78006269885832e-05,
+      "loss": 0.0819,
+      "step": 9402
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.780015908665544e-05,
+      "loss": 0.0614,
+      "step": 9404
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.7799691184727685e-05,
+      "loss": 0.0575,
+      "step": 9406
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.7799223282799924e-05,
+      "loss": 0.0707,
+      "step": 9408
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.779875538087217e-05,
+      "loss": 0.0745,
+      "step": 9410
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.7798287478944416e-05,
+      "loss": 0.1053,
+      "step": 9412
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.779781957701666e-05,
+      "loss": 0.0779,
+      "step": 9414
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.77973516750889e-05,
+      "loss": 0.0562,
+      "step": 9416
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.779688377316115e-05,
+      "loss": 0.0793,
+      "step": 9418
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.779641587123339e-05,
+      "loss": 0.0771,
+      "step": 9420
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.779594796930564e-05,
+      "loss": 0.0901,
+      "step": 9422
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.779548006737788e-05,
+      "loss": 0.0701,
+      "step": 9424
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.7795012165450124e-05,
+      "loss": 0.0783,
+      "step": 9426
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.779454426352237e-05,
+      "loss": 0.0646,
+      "step": 9428
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.7794076361594616e-05,
+      "loss": 0.0641,
+      "step": 9430
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.7793608459666855e-05,
+      "loss": 0.0943,
+      "step": 9432
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.77931405577391e-05,
+      "loss": 0.0689,
+      "step": 9434
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.779267265581135e-05,
+      "loss": 0.0821,
+      "step": 9436
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.779220475388359e-05,
+      "loss": 0.0845,
+      "step": 9438
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.779173685195583e-05,
+      "loss": 0.059,
+      "step": 9440
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.779126895002808e-05,
+      "loss": 0.0603,
+      "step": 9442
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.779080104810032e-05,
+      "loss": 0.0869,
+      "step": 9444
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.779033314617257e-05,
+      "loss": 0.1109,
+      "step": 9446
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.778986524424481e-05,
+      "loss": 0.0692,
+      "step": 9448
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7789397342317055e-05,
+      "loss": 0.0742,
+      "step": 9450
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7788929440389294e-05,
+      "loss": 0.0545,
+      "step": 9452
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.778846153846154e-05,
+      "loss": 0.0703,
+      "step": 9454
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7787993636533786e-05,
+      "loss": 0.0618,
+      "step": 9456
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.778752573460603e-05,
+      "loss": 0.0836,
+      "step": 9458
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.778705783267827e-05,
+      "loss": 0.0541,
+      "step": 9460
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.778658993075052e-05,
+      "loss": 0.0667,
+      "step": 9462
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.778612202882276e-05,
+      "loss": 0.0661,
+      "step": 9464
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.778565412689501e-05,
+      "loss": 0.0839,
+      "step": 9466
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.778518622496725e-05,
+      "loss": 0.0702,
+      "step": 9468
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7784718323039493e-05,
+      "loss": 0.0738,
+      "step": 9470
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.778425042111174e-05,
+      "loss": 0.0702,
+      "step": 9472
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7783782519183985e-05,
+      "loss": 0.0947,
+      "step": 9474
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7783314617256224e-05,
+      "loss": 0.089,
+      "step": 9476
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.778284671532847e-05,
+      "loss": 0.0641,
+      "step": 9478
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7782378813400716e-05,
+      "loss": 0.0864,
+      "step": 9480
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.778191091147296e-05,
+      "loss": 0.0784,
+      "step": 9482
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.77814430095452e-05,
+      "loss": 0.0937,
+      "step": 9484
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.778097510761745e-05,
+      "loss": 0.0642,
+      "step": 9486
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7780507205689686e-05,
+      "loss": 0.0696,
+      "step": 9488
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.778003930376194e-05,
+      "loss": 0.0641,
+      "step": 9490
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.777957140183418e-05,
+      "loss": 0.0722,
+      "step": 9492
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7779103499906424e-05,
+      "loss": 0.0727,
+      "step": 9494
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.777863559797866e-05,
+      "loss": 0.0869,
+      "step": 9496
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.777816769605091e-05,
+      "loss": 0.0742,
+      "step": 9498
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7777699794123155e-05,
+      "loss": 0.0686,
+      "step": 9500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7777231892195394e-05,
+      "loss": 0.0714,
+      "step": 9502
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.777676399026764e-05,
+      "loss": 0.0534,
+      "step": 9504
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.7776296088339886e-05,
+      "loss": 0.0602,
+      "step": 9506
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.777582818641213e-05,
+      "loss": 0.0547,
+      "step": 9508
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.777536028448437e-05,
+      "loss": 0.0715,
+      "step": 9510
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.777489238255662e-05,
+      "loss": 0.0819,
+      "step": 9512
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.777442448062886e-05,
+      "loss": 0.0779,
+      "step": 9514
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.777395657870111e-05,
+      "loss": 0.0715,
+      "step": 9516
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.777348867677335e-05,
+      "loss": 0.0691,
+      "step": 9518
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.7773020774845594e-05,
+      "loss": 0.0671,
+      "step": 9520
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.777255287291783e-05,
+      "loss": 0.0839,
+      "step": 9522
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.7772084970990086e-05,
+      "loss": 0.0706,
+      "step": 9524
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.7771617069062325e-05,
+      "loss": 0.1025,
+      "step": 9526
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.777114916713457e-05,
+      "loss": 0.0597,
+      "step": 9528
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.777068126520681e-05,
+      "loss": 0.0777,
+      "step": 9530
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.777021336327906e-05,
+      "loss": 0.0769,
+      "step": 9532
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.77697454613513e-05,
+      "loss": 0.0754,
+      "step": 9534
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.776927755942355e-05,
+      "loss": 0.0726,
+      "step": 9536
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.776880965749579e-05,
+      "loss": 0.0891,
+      "step": 9538
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.776834175556803e-05,
+      "loss": 0.0779,
+      "step": 9540
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.776787385364028e-05,
+      "loss": 0.0538,
+      "step": 9542
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.7767405951712524e-05,
+      "loss": 0.0767,
+      "step": 9544
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.7766938049784764e-05,
+      "loss": 0.0645,
+      "step": 9546
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.776647014785701e-05,
+      "loss": 0.0704,
+      "step": 9548
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.7766002245929255e-05,
+      "loss": 0.0526,
+      "step": 9550
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.77655343440015e-05,
+      "loss": 0.0731,
+      "step": 9552
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.776506644207374e-05,
+      "loss": 0.0616,
+      "step": 9554
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.7764598540145986e-05,
+      "loss": 0.0945,
+      "step": 9556
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.776413063821823e-05,
+      "loss": 0.0987,
+      "step": 9558
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.776366273629048e-05,
+      "loss": 0.0715,
+      "step": 9560
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.776319483436272e-05,
+      "loss": 0.0913,
+      "step": 9562
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.776272693243496e-05,
+      "loss": 0.0794,
+      "step": 9564
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.776225903050721e-05,
+      "loss": 0.0623,
+      "step": 9566
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.7761791128579455e-05,
+      "loss": 0.05,
+      "step": 9568
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.7761323226651694e-05,
+      "loss": 0.0791,
+      "step": 9570
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.776085532472394e-05,
+      "loss": 0.0791,
+      "step": 9572
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.776038742279618e-05,
+      "loss": 0.0999,
+      "step": 9574
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.775991952086843e-05,
+      "loss": 0.0884,
+      "step": 9576
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.775945161894067e-05,
+      "loss": 0.0693,
+      "step": 9578
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.775898371701292e-05,
+      "loss": 0.0907,
+      "step": 9580
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7758515815085156e-05,
+      "loss": 0.0747,
+      "step": 9582
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.775804791315741e-05,
+      "loss": 0.0725,
+      "step": 9584
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.775758001122965e-05,
+      "loss": 0.0684,
+      "step": 9586
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7757112109301894e-05,
+      "loss": 0.065,
+      "step": 9588
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.775664420737413e-05,
+      "loss": 0.0754,
+      "step": 9590
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.775617630544638e-05,
+      "loss": 0.0679,
+      "step": 9592
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7755708403518625e-05,
+      "loss": 0.0682,
+      "step": 9594
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.775524050159087e-05,
+      "loss": 0.0815,
+      "step": 9596
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.775477259966311e-05,
+      "loss": 0.0898,
+      "step": 9598
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7754304697735356e-05,
+      "loss": 0.0839,
+      "step": 9600
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.77538367958076e-05,
+      "loss": 0.083,
+      "step": 9602
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.775336889387985e-05,
+      "loss": 0.0856,
+      "step": 9604
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.775290099195209e-05,
+      "loss": 0.0677,
+      "step": 9606
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.775243309002433e-05,
+      "loss": 0.0781,
+      "step": 9608
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.775196518809658e-05,
+      "loss": 0.0991,
+      "step": 9610
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7751497286168824e-05,
+      "loss": 0.0913,
+      "step": 9612
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7751029384241064e-05,
+      "loss": 0.0608,
+      "step": 9614
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.775056148231331e-05,
+      "loss": 0.0653,
+      "step": 9616
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.775009358038555e-05,
+      "loss": 0.0609,
+      "step": 9618
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.77496256784578e-05,
+      "loss": 0.0785,
+      "step": 9620
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.774915777653004e-05,
+      "loss": 0.0784,
+      "step": 9622
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7748689874602286e-05,
+      "loss": 0.0833,
+      "step": 9624
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7748221972674526e-05,
+      "loss": 0.0717,
+      "step": 9626
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.774775407074678e-05,
+      "loss": 0.0682,
+      "step": 9628
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.774728616881902e-05,
+      "loss": 0.0794,
+      "step": 9630
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.774681826689126e-05,
+      "loss": 0.0642,
+      "step": 9632
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.77463503649635e-05,
+      "loss": 0.0813,
+      "step": 9634
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.774588246303575e-05,
+      "loss": 0.063,
+      "step": 9636
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7745414561107994e-05,
+      "loss": 0.072,
+      "step": 9638
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.774494665918024e-05,
+      "loss": 0.072,
+      "step": 9640
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.774447875725248e-05,
+      "loss": 0.0679,
+      "step": 9642
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7744010855324725e-05,
+      "loss": 0.0886,
+      "step": 9644
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.774354295339697e-05,
+      "loss": 0.0753,
+      "step": 9646
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.774307505146922e-05,
+      "loss": 0.073,
+      "step": 9648
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7742607149541456e-05,
+      "loss": 0.0789,
+      "step": 9650
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.77421392476137e-05,
+      "loss": 0.0915,
+      "step": 9652
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.774167134568595e-05,
+      "loss": 0.0621,
+      "step": 9654
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7741203443758194e-05,
+      "loss": 0.0681,
+      "step": 9656
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.774073554183043e-05,
+      "loss": 0.1235,
+      "step": 9658
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.774026763990268e-05,
+      "loss": 0.0724,
+      "step": 9660
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7739799737974925e-05,
+      "loss": 0.0849,
+      "step": 9662
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.773933183604717e-05,
+      "loss": 0.1005,
+      "step": 9664
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.773886393411941e-05,
+      "loss": 0.0999,
+      "step": 9666
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7738396032191656e-05,
+      "loss": 0.0581,
+      "step": 9668
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7737928130263895e-05,
+      "loss": 0.0877,
+      "step": 9670
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.773746022833615e-05,
+      "loss": 0.0776,
+      "step": 9672
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.773699232640839e-05,
+      "loss": 0.0849,
+      "step": 9674
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.773652442448063e-05,
+      "loss": 0.0882,
+      "step": 9676
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.773605652255287e-05,
+      "loss": 0.074,
+      "step": 9678
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7735588620625125e-05,
+      "loss": 0.0653,
+      "step": 9680
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7735120718697364e-05,
+      "loss": 0.0706,
+      "step": 9682
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.773465281676961e-05,
+      "loss": 0.0886,
+      "step": 9684
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.773418491484185e-05,
+      "loss": 0.0674,
+      "step": 9686
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7733717012914095e-05,
+      "loss": 0.0827,
+      "step": 9688
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.773324911098634e-05,
+      "loss": 0.0712,
+      "step": 9690
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7732781209058586e-05,
+      "loss": 0.064,
+      "step": 9692
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7732313307130826e-05,
+      "loss": 0.0863,
+      "step": 9694
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.773184540520307e-05,
+      "loss": 0.0711,
+      "step": 9696
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.773137750327532e-05,
+      "loss": 0.0988,
+      "step": 9698
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.773090960134756e-05,
+      "loss": 0.0715,
+      "step": 9700
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.77304416994198e-05,
+      "loss": 0.069,
+      "step": 9702
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.772997379749205e-05,
+      "loss": 0.0843,
+      "step": 9704
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7729505895564294e-05,
+      "loss": 0.0505,
+      "step": 9706
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.772903799363654e-05,
+      "loss": 0.0832,
+      "step": 9708
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.772857009170878e-05,
+      "loss": 0.0674,
+      "step": 9710
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7728102189781025e-05,
+      "loss": 0.0918,
+      "step": 9712
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.772763428785327e-05,
+      "loss": 0.062,
+      "step": 9714
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.772716638592552e-05,
+      "loss": 0.0787,
+      "step": 9716
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7726698483997756e-05,
+      "loss": 0.1003,
+      "step": 9718
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.772623058207e-05,
+      "loss": 0.0892,
+      "step": 9720
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.772576268014224e-05,
+      "loss": 0.0901,
+      "step": 9722
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7725294778214494e-05,
+      "loss": 0.0626,
+      "step": 9724
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.772482687628673e-05,
+      "loss": 0.0543,
+      "step": 9726
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.772435897435898e-05,
+      "loss": 0.0589,
+      "step": 9728
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.772389107243122e-05,
+      "loss": 0.0743,
+      "step": 9730
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7723423170503464e-05,
+      "loss": 0.0698,
+      "step": 9732
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.772295526857571e-05,
+      "loss": 0.0566,
+      "step": 9734
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7722487366647956e-05,
+      "loss": 0.0756,
+      "step": 9736
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7722019464720195e-05,
+      "loss": 0.0818,
+      "step": 9738
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.772155156279244e-05,
+      "loss": 0.08,
+      "step": 9740
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.772108366086469e-05,
+      "loss": 0.0814,
+      "step": 9742
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.772061575893693e-05,
+      "loss": 0.0714,
+      "step": 9744
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.772014785700917e-05,
+      "loss": 0.0875,
+      "step": 9746
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.771967995508142e-05,
+      "loss": 0.0761,
+      "step": 9748
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7719212053153664e-05,
+      "loss": 0.104,
+      "step": 9750
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.77187441512259e-05,
+      "loss": 0.0747,
+      "step": 9752
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.771827624929815e-05,
+      "loss": 0.0648,
+      "step": 9754
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.771780834737039e-05,
+      "loss": 0.0821,
+      "step": 9756
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.771734044544264e-05,
+      "loss": 0.075,
+      "step": 9758
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.771687254351488e-05,
+      "loss": 0.0703,
+      "step": 9760
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7716404641587126e-05,
+      "loss": 0.0719,
+      "step": 9762
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7715936739659365e-05,
+      "loss": 0.0866,
+      "step": 9764
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.771546883773161e-05,
+      "loss": 0.0786,
+      "step": 9766
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7715000935803857e-05,
+      "loss": 0.0608,
+      "step": 9768
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.77145330338761e-05,
+      "loss": 0.0894,
+      "step": 9770
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.771406513194834e-05,
+      "loss": 0.0729,
+      "step": 9772
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.771359723002059e-05,
+      "loss": 0.0702,
+      "step": 9774
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7713129328092833e-05,
+      "loss": 0.0739,
+      "step": 9776
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.771266142616508e-05,
+      "loss": 0.0684,
+      "step": 9778
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.771219352423732e-05,
+      "loss": 0.0607,
+      "step": 9780
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7711725622309564e-05,
+      "loss": 0.0645,
+      "step": 9782
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.771125772038181e-05,
+      "loss": 0.0874,
+      "step": 9784
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7710789818454056e-05,
+      "loss": 0.0736,
+      "step": 9786
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7710321916526295e-05,
+      "loss": 0.0786,
+      "step": 9788
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.770985401459854e-05,
+      "loss": 0.0883,
+      "step": 9790
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.770938611267079e-05,
+      "loss": 0.066,
+      "step": 9792
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.770891821074303e-05,
+      "loss": 0.0771,
+      "step": 9794
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.770845030881527e-05,
+      "loss": 0.0729,
+      "step": 9796
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.770798240688752e-05,
+      "loss": 0.0639,
+      "step": 9798
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.770751450495976e-05,
+      "loss": 0.0947,
+      "step": 9800
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.770704660303201e-05,
+      "loss": 0.0579,
+      "step": 9802
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.770657870110425e-05,
+      "loss": 0.0829,
+      "step": 9804
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7706110799176495e-05,
+      "loss": 0.0607,
+      "step": 9806
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7705642897248734e-05,
+      "loss": 0.086,
+      "step": 9808
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.770517499532099e-05,
+      "loss": 0.0956,
+      "step": 9810
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7704707093393226e-05,
+      "loss": 0.0797,
+      "step": 9812
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.770423919146547e-05,
+      "loss": 0.0675,
+      "step": 9814
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.770377128953771e-05,
+      "loss": 0.0789,
+      "step": 9816
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.770330338760996e-05,
+      "loss": 0.0852,
+      "step": 9818
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.77028354856822e-05,
+      "loss": 0.0734,
+      "step": 9820
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.770236758375445e-05,
+      "loss": 0.0697,
+      "step": 9822
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.770189968182669e-05,
+      "loss": 0.0735,
+      "step": 9824
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7701431779898934e-05,
+      "loss": 0.075,
+      "step": 9826
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.770096387797118e-05,
+      "loss": 0.0738,
+      "step": 9828
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7700495976043426e-05,
+      "loss": 0.0763,
+      "step": 9830
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7700028074115665e-05,
+      "loss": 0.0636,
+      "step": 9832
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.769956017218791e-05,
+      "loss": 0.0753,
+      "step": 9834
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7699092270260157e-05,
+      "loss": 0.0515,
+      "step": 9836
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.76986243683324e-05,
+      "loss": 0.0864,
+      "step": 9838
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.769815646640464e-05,
+      "loss": 0.082,
+      "step": 9840
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.769768856447689e-05,
+      "loss": 0.103,
+      "step": 9842
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7697220662549133e-05,
+      "loss": 0.0795,
+      "step": 9844
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.769675276062138e-05,
+      "loss": 0.0762,
+      "step": 9846
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.769628485869362e-05,
+      "loss": 0.0741,
+      "step": 9848
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7695816956765864e-05,
+      "loss": 0.0859,
+      "step": 9850
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7695349054838104e-05,
+      "loss": 0.0911,
+      "step": 9852
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7694881152910356e-05,
+      "loss": 0.0538,
+      "step": 9854
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7694413250982595e-05,
+      "loss": 0.0864,
+      "step": 9856
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.769394534905484e-05,
+      "loss": 0.0714,
+      "step": 9858
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.769347744712708e-05,
+      "loss": 0.076,
+      "step": 9860
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.769300954519933e-05,
+      "loss": 0.0867,
+      "step": 9862
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.769254164327157e-05,
+      "loss": 0.0533,
+      "step": 9864
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.769207374134382e-05,
+      "loss": 0.0657,
+      "step": 9866
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.769160583941606e-05,
+      "loss": 0.0729,
+      "step": 9868
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.76911379374883e-05,
+      "loss": 0.0832,
+      "step": 9870
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.769067003556055e-05,
+      "loss": 0.0678,
+      "step": 9872
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.7690202133632795e-05,
+      "loss": 0.0666,
+      "step": 9874
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.7689734231705034e-05,
+      "loss": 0.0718,
+      "step": 9876
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.768926632977728e-05,
+      "loss": 0.112,
+      "step": 9878
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.7688798427849526e-05,
+      "loss": 0.0844,
+      "step": 9880
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.768833052592177e-05,
+      "loss": 0.0779,
+      "step": 9882
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.768786262399401e-05,
+      "loss": 0.0621,
+      "step": 9884
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.768739472206626e-05,
+      "loss": 0.0749,
+      "step": 9886
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.76869268201385e-05,
+      "loss": 0.0599,
+      "step": 9888
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.768645891821075e-05,
+      "loss": 0.0632,
+      "step": 9890
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.768599101628299e-05,
+      "loss": 0.0714,
+      "step": 9892
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.7685523114355234e-05,
+      "loss": 0.0759,
+      "step": 9894
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.768505521242748e-05,
+      "loss": 0.0895,
+      "step": 9896
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.7684587310499726e-05,
+      "loss": 0.0576,
+      "step": 9898
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.7684119408571965e-05,
+      "loss": 0.0895,
+      "step": 9900
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.768365150664421e-05,
+      "loss": 0.1024,
+      "step": 9902
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.768318360471645e-05,
+      "loss": 0.0746,
+      "step": 9904
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.76827157027887e-05,
+      "loss": 0.064,
+      "step": 9906
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.768224780086094e-05,
+      "loss": 0.0898,
+      "step": 9908
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.768177989893319e-05,
+      "loss": 0.0732,
+      "step": 9910
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.768131199700543e-05,
+      "loss": 0.0689,
+      "step": 9912
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.768084409507767e-05,
+      "loss": 0.0677,
+      "step": 9914
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.768037619314992e-05,
+      "loss": 0.0953,
+      "step": 9916
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.7679908291222164e-05,
+      "loss": 0.0761,
+      "step": 9918
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.7679440389294404e-05,
+      "loss": 0.0828,
+      "step": 9920
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.767897248736665e-05,
+      "loss": 0.0798,
+      "step": 9922
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.7678504585438895e-05,
+      "loss": 0.0787,
+      "step": 9924
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.767803668351114e-05,
+      "loss": 0.0637,
+      "step": 9926
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.767756878158338e-05,
+      "loss": 0.0898,
+      "step": 9928
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.7677100879655626e-05,
+      "loss": 0.0719,
+      "step": 9930
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.767663297772787e-05,
+      "loss": 0.0693,
+      "step": 9932
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.767616507580012e-05,
+      "loss": 0.0773,
+      "step": 9934
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.767569717387236e-05,
+      "loss": 0.0691,
+      "step": 9936
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.76752292719446e-05,
+      "loss": 0.0797,
+      "step": 9938
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.767476137001685e-05,
+      "loss": 0.0809,
+      "step": 9940
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.7674293468089095e-05,
+      "loss": 0.0595,
+      "step": 9942
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.7673825566161334e-05,
+      "loss": 0.0704,
+      "step": 9944
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.767335766423358e-05,
+      "loss": 0.0709,
+      "step": 9946
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.767288976230582e-05,
+      "loss": 0.0863,
+      "step": 9948
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.767242186037807e-05,
+      "loss": 0.0692,
+      "step": 9950
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.767195395845031e-05,
+      "loss": 0.0904,
+      "step": 9952
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.767148605652256e-05,
+      "loss": 0.0807,
+      "step": 9954
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.7671018154594796e-05,
+      "loss": 0.0644,
+      "step": 9956
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.767055025266705e-05,
+      "loss": 0.0687,
+      "step": 9958
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.767008235073929e-05,
+      "loss": 0.0609,
+      "step": 9960
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.7669614448811534e-05,
+      "loss": 0.0771,
+      "step": 9962
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.766914654688377e-05,
+      "loss": 0.0836,
+      "step": 9964
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.766867864495602e-05,
+      "loss": 0.08,
+      "step": 9966
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.7668210743028265e-05,
+      "loss": 0.0753,
+      "step": 9968
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.766774284110051e-05,
+      "loss": 0.0781,
+      "step": 9970
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.766727493917275e-05,
+      "loss": 0.0704,
+      "step": 9972
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.7666807037244996e-05,
+      "loss": 0.0707,
+      "step": 9974
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.766633913531724e-05,
+      "loss": 0.0661,
+      "step": 9976
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.766587123338949e-05,
+      "loss": 0.0683,
+      "step": 9978
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.766540333146173e-05,
+      "loss": 0.0688,
+      "step": 9980
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.766493542953397e-05,
+      "loss": 0.0902,
+      "step": 9982
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.766446752760622e-05,
+      "loss": 0.0783,
+      "step": 9984
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.7663999625678464e-05,
+      "loss": 0.0778,
+      "step": 9986
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.7663531723750704e-05,
+      "loss": 0.0921,
+      "step": 9988
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.766306382182295e-05,
+      "loss": 0.066,
+      "step": 9990
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.7662595919895195e-05,
+      "loss": 0.0598,
+      "step": 9992
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.766212801796744e-05,
+      "loss": 0.0784,
+      "step": 9994
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.766166011603968e-05,
+      "loss": 0.0627,
+      "step": 9996
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.7661192214111926e-05,
+      "loss": 0.0974,
+      "step": 9998
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.7660724312184165e-05,
+      "loss": 0.09,
+      "step": 10000
+    },
+    {
+      "epoch": 1.4,
+      "eval_gen_len": 28.5517,
+      "eval_loss": 1.054301142692566,
+      "eval_meteor": 0.0451,
+      "eval_runtime": 13.332,
+      "eval_samples_per_second": 4.35,
+      "eval_steps_per_second": 0.6,
+      "step": 10000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.766025641025642e-05,
+      "loss": 0.0818,
+      "step": 10002
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.765978850832866e-05,
+      "loss": 0.0792,
+      "step": 10004
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.7659320606400896e-05,
+      "loss": 0.0772,
+      "step": 10006
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.765885270447314e-05,
+      "loss": 0.0575,
+      "step": 10008
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.765838480254539e-05,
+      "loss": 0.0836,
+      "step": 10010
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.7657916900617634e-05,
+      "loss": 0.0748,
+      "step": 10012
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.765744899868987e-05,
+      "loss": 0.0925,
+      "step": 10014
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.765698109676212e-05,
+      "loss": 0.0833,
+      "step": 10016
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.7656513194834365e-05,
+      "loss": 0.0727,
+      "step": 10018
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.765604529290661e-05,
+      "loss": 0.0711,
+      "step": 10020
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.765557739097885e-05,
+      "loss": 0.075,
+      "step": 10022
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.7655109489051096e-05,
+      "loss": 0.0952,
+      "step": 10024
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.765464158712334e-05,
+      "loss": 0.0643,
+      "step": 10026
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.765417368519559e-05,
+      "loss": 0.0896,
+      "step": 10028
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.765370578326783e-05,
+      "loss": 0.0653,
+      "step": 10030
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.765323788134007e-05,
+      "loss": 0.0816,
+      "step": 10032
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.765276997941231e-05,
+      "loss": 0.0766,
+      "step": 10034
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.7652302077484565e-05,
+      "loss": 0.0978,
+      "step": 10036
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.7651834175556804e-05,
+      "loss": 0.0816,
+      "step": 10038
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.765136627362905e-05,
+      "loss": 0.0778,
+      "step": 10040
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.765089837170129e-05,
+      "loss": 0.0833,
+      "step": 10042
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.7650430469773535e-05,
+      "loss": 0.0851,
+      "step": 10044
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.764996256784578e-05,
+      "loss": 0.063,
+      "step": 10046
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.764949466591803e-05,
+      "loss": 0.0694,
+      "step": 10048
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.7649026763990266e-05,
+      "loss": 0.0671,
+      "step": 10050
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.764855886206251e-05,
+      "loss": 0.0687,
+      "step": 10052
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.764809096013476e-05,
+      "loss": 0.0728,
+      "step": 10054
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.7647623058207004e-05,
+      "loss": 0.0609,
+      "step": 10056
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.764715515627924e-05,
+      "loss": 0.0868,
+      "step": 10058
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.764668725435149e-05,
+      "loss": 0.0629,
+      "step": 10060
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.7646219352423735e-05,
+      "loss": 0.0795,
+      "step": 10062
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.764575145049598e-05,
+      "loss": 0.0821,
+      "step": 10064
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.764528354856822e-05,
+      "loss": 0.0911,
+      "step": 10066
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.7644815646640465e-05,
+      "loss": 0.0839,
+      "step": 10068
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.764434774471271e-05,
+      "loss": 0.0989,
+      "step": 10070
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.764387984278496e-05,
+      "loss": 0.0642,
+      "step": 10072
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.7643411940857196e-05,
+      "loss": 0.074,
+      "step": 10074
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.764294403892944e-05,
+      "loss": 0.0802,
+      "step": 10076
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.764247613700168e-05,
+      "loss": 0.066,
+      "step": 10078
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.7642008235073934e-05,
+      "loss": 0.063,
+      "step": 10080
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.764154033314617e-05,
+      "loss": 0.0642,
+      "step": 10082
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.764107243121842e-05,
+      "loss": 0.0953,
+      "step": 10084
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.764060452929066e-05,
+      "loss": 0.0785,
+      "step": 10086
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.764013662736291e-05,
+      "loss": 0.0715,
+      "step": 10088
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.763966872543515e-05,
+      "loss": 0.0878,
+      "step": 10090
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.7639200823507396e-05,
+      "loss": 0.0906,
+      "step": 10092
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.7638732921579635e-05,
+      "loss": 0.0669,
+      "step": 10094
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.763826501965188e-05,
+      "loss": 0.0847,
+      "step": 10096
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.763779711772413e-05,
+      "loss": 0.1002,
+      "step": 10098
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.763732921579637e-05,
+      "loss": 0.0795,
+      "step": 10100
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.763686131386861e-05,
+      "loss": 0.0743,
+      "step": 10102
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.763639341194086e-05,
+      "loss": 0.0795,
+      "step": 10104
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.7635925510013104e-05,
+      "loss": 0.0786,
+      "step": 10106
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.763545760808535e-05,
+      "loss": 0.0795,
+      "step": 10108
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.763498970615759e-05,
+      "loss": 0.0643,
+      "step": 10110
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.7634521804229835e-05,
+      "loss": 0.0751,
+      "step": 10112
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.763405390230208e-05,
+      "loss": 0.0759,
+      "step": 10114
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.763358600037433e-05,
+      "loss": 0.0764,
+      "step": 10116
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.7633118098446566e-05,
+      "loss": 0.0866,
+      "step": 10118
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.763265019651881e-05,
+      "loss": 0.0631,
+      "step": 10120
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.763218229459106e-05,
+      "loss": 0.0834,
+      "step": 10122
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.7631714392663304e-05,
+      "loss": 0.0667,
+      "step": 10124
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.763124649073554e-05,
+      "loss": 0.0964,
+      "step": 10126
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.763077858880779e-05,
+      "loss": 0.0615,
+      "step": 10128
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.763031068688003e-05,
+      "loss": 0.0611,
+      "step": 10130
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.762984278495228e-05,
+      "loss": 0.0695,
+      "step": 10132
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.762937488302452e-05,
+      "loss": 0.0709,
+      "step": 10134
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.7628906981096766e-05,
+      "loss": 0.0801,
+      "step": 10136
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.7628439079169005e-05,
+      "loss": 0.0692,
+      "step": 10138
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.762797117724126e-05,
+      "loss": 0.0708,
+      "step": 10140
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.7627503275313496e-05,
+      "loss": 0.0976,
+      "step": 10142
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.762703537338574e-05,
+      "loss": 0.0699,
+      "step": 10144
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.762656747145798e-05,
+      "loss": 0.0636,
+      "step": 10146
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.762609956953023e-05,
+      "loss": 0.0588,
+      "step": 10148
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.762563166760247e-05,
+      "loss": 0.059,
+      "step": 10150
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.762516376567472e-05,
+      "loss": 0.0797,
+      "step": 10152
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.762469586374696e-05,
+      "loss": 0.0753,
+      "step": 10154
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.7624227961819204e-05,
+      "loss": 0.0878,
+      "step": 10156
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.762376005989145e-05,
+      "loss": 0.0826,
+      "step": 10158
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.7623292157963696e-05,
+      "loss": 0.0902,
+      "step": 10160
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.7622824256035935e-05,
+      "loss": 0.06,
+      "step": 10162
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.762235635410818e-05,
+      "loss": 0.0776,
+      "step": 10164
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.762188845218043e-05,
+      "loss": 0.0643,
+      "step": 10166
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.762142055025267e-05,
+      "loss": 0.0829,
+      "step": 10168
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.762095264832491e-05,
+      "loss": 0.0756,
+      "step": 10170
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.762048474639716e-05,
+      "loss": 0.0951,
+      "step": 10172
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.7620016844469404e-05,
+      "loss": 0.076,
+      "step": 10174
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.761954894254165e-05,
+      "loss": 0.0738,
+      "step": 10176
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.761908104061389e-05,
+      "loss": 0.0772,
+      "step": 10178
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.7618613138686135e-05,
+      "loss": 0.0677,
+      "step": 10180
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.7618145236758374e-05,
+      "loss": 0.0833,
+      "step": 10182
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.761767733483063e-05,
+      "loss": 0.0759,
+      "step": 10184
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.7617209432902866e-05,
+      "loss": 0.0751,
+      "step": 10186
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.761674153097511e-05,
+      "loss": 0.0665,
+      "step": 10188
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.761627362904735e-05,
+      "loss": 0.0852,
+      "step": 10190
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.76158057271196e-05,
+      "loss": 0.0953,
+      "step": 10192
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.761533782519184e-05,
+      "loss": 0.0823,
+      "step": 10194
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.761486992326409e-05,
+      "loss": 0.0812,
+      "step": 10196
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.761440202133633e-05,
+      "loss": 0.0773,
+      "step": 10198
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.7613934119408574e-05,
+      "loss": 0.0887,
+      "step": 10200
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.761346621748082e-05,
+      "loss": 0.0602,
+      "step": 10202
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.7612998315553066e-05,
+      "loss": 0.0776,
+      "step": 10204
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.7612530413625305e-05,
+      "loss": 0.1012,
+      "step": 10206
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.761206251169755e-05,
+      "loss": 0.098,
+      "step": 10208
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.7611594609769796e-05,
+      "loss": 0.0564,
+      "step": 10210
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.761112670784204e-05,
+      "loss": 0.115,
+      "step": 10212
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.761065880591428e-05,
+      "loss": 0.0707,
+      "step": 10214
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.761019090398653e-05,
+      "loss": 0.0586,
+      "step": 10216
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.760972300205877e-05,
+      "loss": 0.0837,
+      "step": 10218
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.760925510013102e-05,
+      "loss": 0.057,
+      "step": 10220
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.760878719820326e-05,
+      "loss": 0.0619,
+      "step": 10222
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.7608319296275504e-05,
+      "loss": 0.0778,
+      "step": 10224
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.7607851394347743e-05,
+      "loss": 0.0855,
+      "step": 10226
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.7607383492419996e-05,
+      "loss": 0.0645,
+      "step": 10228
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.7606915590492235e-05,
+      "loss": 0.0837,
+      "step": 10230
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.760644768856448e-05,
+      "loss": 0.0745,
+      "step": 10232
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.760597978663672e-05,
+      "loss": 0.0639,
+      "step": 10234
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.760551188470897e-05,
+      "loss": 0.0886,
+      "step": 10236
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.760504398278121e-05,
+      "loss": 0.0946,
+      "step": 10238
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.760457608085346e-05,
+      "loss": 0.0819,
+      "step": 10240
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.76041081789257e-05,
+      "loss": 0.0925,
+      "step": 10242
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.760364027699794e-05,
+      "loss": 0.0561,
+      "step": 10244
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.760317237507019e-05,
+      "loss": 0.0709,
+      "step": 10246
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.7602704473142435e-05,
+      "loss": 0.064,
+      "step": 10248
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.7602236571214674e-05,
+      "loss": 0.047,
+      "step": 10250
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.760176866928692e-05,
+      "loss": 0.0794,
+      "step": 10252
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.7601300767359166e-05,
+      "loss": 0.0786,
+      "step": 10254
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.760083286543141e-05,
+      "loss": 0.0785,
+      "step": 10256
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.760036496350365e-05,
+      "loss": 0.0695,
+      "step": 10258
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.759989706157589e-05,
+      "loss": 0.0832,
+      "step": 10260
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.759942915964814e-05,
+      "loss": 0.0776,
+      "step": 10262
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.759896125772038e-05,
+      "loss": 0.0844,
+      "step": 10264
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.759849335579263e-05,
+      "loss": 0.0635,
+      "step": 10266
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.759802545386487e-05,
+      "loss": 0.0904,
+      "step": 10268
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.759755755193712e-05,
+      "loss": 0.0815,
+      "step": 10270
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.759708965000936e-05,
+      "loss": 0.0622,
+      "step": 10272
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.7596621748081605e-05,
+      "loss": 0.0852,
+      "step": 10274
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.7596153846153844e-05,
+      "loss": 0.0782,
+      "step": 10276
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.759568594422609e-05,
+      "loss": 0.0733,
+      "step": 10278
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.7595218042298336e-05,
+      "loss": 0.0831,
+      "step": 10280
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.759475014037058e-05,
+      "loss": 0.069,
+      "step": 10282
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.759428223844282e-05,
+      "loss": 0.0582,
+      "step": 10284
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.7593814336515067e-05,
+      "loss": 0.0581,
+      "step": 10286
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.759334643458731e-05,
+      "loss": 0.0904,
+      "step": 10288
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.759287853265956e-05,
+      "loss": 0.0674,
+      "step": 10290
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.75924106307318e-05,
+      "loss": 0.068,
+      "step": 10292
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.7591942728804043e-05,
+      "loss": 0.0824,
+      "step": 10294
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.759147482687629e-05,
+      "loss": 0.0638,
+      "step": 10296
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7591006924948535e-05,
+      "loss": 0.0851,
+      "step": 10298
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7590539023020774e-05,
+      "loss": 0.0628,
+      "step": 10300
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.759007112109302e-05,
+      "loss": 0.0814,
+      "step": 10302
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7589603219165266e-05,
+      "loss": 0.0811,
+      "step": 10304
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.758913531723751e-05,
+      "loss": 0.0766,
+      "step": 10306
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.758866741530975e-05,
+      "loss": 0.0863,
+      "step": 10308
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7588199513382e-05,
+      "loss": 0.072,
+      "step": 10310
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7587731611454236e-05,
+      "loss": 0.065,
+      "step": 10312
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.758726370952649e-05,
+      "loss": 0.0783,
+      "step": 10314
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.758679580759873e-05,
+      "loss": 0.0976,
+      "step": 10316
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7586327905670974e-05,
+      "loss": 0.0812,
+      "step": 10318
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.758586000374321e-05,
+      "loss": 0.0647,
+      "step": 10320
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.758539210181546e-05,
+      "loss": 0.0655,
+      "step": 10322
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7584924199887705e-05,
+      "loss": 0.0794,
+      "step": 10324
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.758445629795995e-05,
+      "loss": 0.0911,
+      "step": 10326
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.758398839603219e-05,
+      "loss": 0.0695,
+      "step": 10328
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7583520494104436e-05,
+      "loss": 0.0937,
+      "step": 10330
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.758305259217668e-05,
+      "loss": 0.0858,
+      "step": 10332
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.758258469024893e-05,
+      "loss": 0.086,
+      "step": 10334
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.758211678832117e-05,
+      "loss": 0.0736,
+      "step": 10336
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.758164888639341e-05,
+      "loss": 0.0772,
+      "step": 10338
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.758118098446566e-05,
+      "loss": 0.069,
+      "step": 10340
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7580713082537905e-05,
+      "loss": 0.0619,
+      "step": 10342
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7580245180610144e-05,
+      "loss": 0.0741,
+      "step": 10344
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.757977727868239e-05,
+      "loss": 0.067,
+      "step": 10346
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7579309376754636e-05,
+      "loss": 0.1078,
+      "step": 10348
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.757884147482688e-05,
+      "loss": 0.0737,
+      "step": 10350
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.757837357289912e-05,
+      "loss": 0.0616,
+      "step": 10352
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7577905670971367e-05,
+      "loss": 0.0721,
+      "step": 10354
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7577437769043606e-05,
+      "loss": 0.0865,
+      "step": 10356
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.757696986711586e-05,
+      "loss": 0.0837,
+      "step": 10358
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.75765019651881e-05,
+      "loss": 0.0855,
+      "step": 10360
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.7576034063260343e-05,
+      "loss": 0.0708,
+      "step": 10362
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.757556616133258e-05,
+      "loss": 0.087,
+      "step": 10364
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.7575098259404835e-05,
+      "loss": 0.0723,
+      "step": 10366
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.7574630357477074e-05,
+      "loss": 0.0713,
+      "step": 10368
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.757416245554932e-05,
+      "loss": 0.0699,
+      "step": 10370
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.757369455362156e-05,
+      "loss": 0.0911,
+      "step": 10372
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.7573226651693805e-05,
+      "loss": 0.0783,
+      "step": 10374
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.757275874976605e-05,
+      "loss": 0.0519,
+      "step": 10376
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.75722908478383e-05,
+      "loss": 0.0662,
+      "step": 10378
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.7571822945910536e-05,
+      "loss": 0.0599,
+      "step": 10380
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.757135504398278e-05,
+      "loss": 0.067,
+      "step": 10382
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.757088714205503e-05,
+      "loss": 0.0655,
+      "step": 10384
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.7570419240127274e-05,
+      "loss": 0.1293,
+      "step": 10386
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.756995133819951e-05,
+      "loss": 0.0776,
+      "step": 10388
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.756948343627176e-05,
+      "loss": 0.0816,
+      "step": 10390
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.7569015534344005e-05,
+      "loss": 0.0835,
+      "step": 10392
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.756854763241625e-05,
+      "loss": 0.0476,
+      "step": 10394
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.756807973048849e-05,
+      "loss": 0.0549,
+      "step": 10396
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.7567611828560736e-05,
+      "loss": 0.1,
+      "step": 10398
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.756714392663298e-05,
+      "loss": 0.0943,
+      "step": 10400
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.756667602470523e-05,
+      "loss": 0.0853,
+      "step": 10402
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.756620812277747e-05,
+      "loss": 0.0702,
+      "step": 10404
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.756574022084971e-05,
+      "loss": 0.0667,
+      "step": 10406
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.756527231892195e-05,
+      "loss": 0.0767,
+      "step": 10408
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.7564804416994205e-05,
+      "loss": 0.0769,
+      "step": 10410
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.7564336515066444e-05,
+      "loss": 0.0684,
+      "step": 10412
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.756386861313869e-05,
+      "loss": 0.0733,
+      "step": 10414
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.756340071121093e-05,
+      "loss": 0.0682,
+      "step": 10416
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.756293280928318e-05,
+      "loss": 0.0649,
+      "step": 10418
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.756246490735542e-05,
+      "loss": 0.0797,
+      "step": 10420
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.756199700542767e-05,
+      "loss": 0.0649,
+      "step": 10422
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.7561529103499906e-05,
+      "loss": 0.0745,
+      "step": 10424
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.756106120157215e-05,
+      "loss": 0.0737,
+      "step": 10426
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.75605932996444e-05,
+      "loss": 0.0702,
+      "step": 10428
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.7560125397716643e-05,
+      "loss": 0.0813,
+      "step": 10430
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.755965749578888e-05,
+      "loss": 0.071,
+      "step": 10432
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.755918959386113e-05,
+      "loss": 0.0971,
+      "step": 10434
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.7558721691933374e-05,
+      "loss": 0.0881,
+      "step": 10436
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.755825379000562e-05,
+      "loss": 0.0977,
+      "step": 10438
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.755778588807786e-05,
+      "loss": 0.0615,
+      "step": 10440
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.7557317986150105e-05,
+      "loss": 0.0871,
+      "step": 10442
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.755685008422235e-05,
+      "loss": 0.094,
+      "step": 10444
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.75563821822946e-05,
+      "loss": 0.0777,
+      "step": 10446
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.7555914280366836e-05,
+      "loss": 0.0763,
+      "step": 10448
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.755544637843908e-05,
+      "loss": 0.0814,
+      "step": 10450
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.755497847651133e-05,
+      "loss": 0.1032,
+      "step": 10452
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.7554510574583574e-05,
+      "loss": 0.0849,
+      "step": 10454
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.755404267265581e-05,
+      "loss": 0.0768,
+      "step": 10456
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.755357477072806e-05,
+      "loss": 0.08,
+      "step": 10458
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.75531068688003e-05,
+      "loss": 0.0926,
+      "step": 10460
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.755263896687255e-05,
+      "loss": 0.0698,
+      "step": 10462
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.755217106494479e-05,
+      "loss": 0.077,
+      "step": 10464
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.7551703163017036e-05,
+      "loss": 0.0722,
+      "step": 10466
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.7551235261089275e-05,
+      "loss": 0.0538,
+      "step": 10468
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.755076735916152e-05,
+      "loss": 0.1016,
+      "step": 10470
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.755029945723377e-05,
+      "loss": 0.0763,
+      "step": 10472
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.754983155530601e-05,
+      "loss": 0.0851,
+      "step": 10474
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.754936365337825e-05,
+      "loss": 0.081,
+      "step": 10476
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.75488957514505e-05,
+      "loss": 0.0741,
+      "step": 10478
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.7548427849522744e-05,
+      "loss": 0.0784,
+      "step": 10480
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.754795994759499e-05,
+      "loss": 0.0804,
+      "step": 10482
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.754749204566723e-05,
+      "loss": 0.0797,
+      "step": 10484
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.7547024143739475e-05,
+      "loss": 0.0931,
+      "step": 10486
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.754655624181172e-05,
+      "loss": 0.0706,
+      "step": 10488
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.754608833988397e-05,
+      "loss": 0.072,
+      "step": 10490
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.7545620437956206e-05,
+      "loss": 0.0801,
+      "step": 10492
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.754515253602845e-05,
+      "loss": 0.074,
+      "step": 10494
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.75446846341007e-05,
+      "loss": 0.0835,
+      "step": 10496
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.7544216732172944e-05,
+      "loss": 0.0612,
+      "step": 10498
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.754374883024518e-05,
+      "loss": 0.0785,
+      "step": 10500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.754328092831743e-05,
+      "loss": 0.0584,
+      "step": 10502
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.754281302638967e-05,
+      "loss": 0.0711,
+      "step": 10504
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.754234512446192e-05,
+      "loss": 0.079,
+      "step": 10506
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.754187722253416e-05,
+      "loss": 0.0791,
+      "step": 10508
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.75414093206064e-05,
+      "loss": 0.0831,
+      "step": 10510
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.7540941418678645e-05,
+      "loss": 0.0661,
+      "step": 10512
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.754047351675089e-05,
+      "loss": 0.0787,
+      "step": 10514
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.7540005614823136e-05,
+      "loss": 0.0951,
+      "step": 10516
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.7539537712895376e-05,
+      "loss": 0.0659,
+      "step": 10518
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.753906981096762e-05,
+      "loss": 0.0759,
+      "step": 10520
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.753860190903987e-05,
+      "loss": 0.073,
+      "step": 10522
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.753813400711211e-05,
+      "loss": 0.0576,
+      "step": 10524
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.753766610518435e-05,
+      "loss": 0.0819,
+      "step": 10526
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.75371982032566e-05,
+      "loss": 0.0805,
+      "step": 10528
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.7536730301328844e-05,
+      "loss": 0.0822,
+      "step": 10530
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.753626239940109e-05,
+      "loss": 0.0725,
+      "step": 10532
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.753579449747333e-05,
+      "loss": 0.0829,
+      "step": 10534
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.7535326595545575e-05,
+      "loss": 0.0851,
+      "step": 10536
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.7534858693617814e-05,
+      "loss": 0.0814,
+      "step": 10538
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.753439079169007e-05,
+      "loss": 0.0964,
+      "step": 10540
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.7533922889762306e-05,
+      "loss": 0.0887,
+      "step": 10542
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.753345498783455e-05,
+      "loss": 0.0728,
+      "step": 10544
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.753298708590679e-05,
+      "loss": 0.0643,
+      "step": 10546
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.7532519183979044e-05,
+      "loss": 0.0664,
+      "step": 10548
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.753205128205128e-05,
+      "loss": 0.0892,
+      "step": 10550
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.753158338012353e-05,
+      "loss": 0.0759,
+      "step": 10552
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.753111547819577e-05,
+      "loss": 0.0717,
+      "step": 10554
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.7530647576268014e-05,
+      "loss": 0.0964,
+      "step": 10556
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.753017967434026e-05,
+      "loss": 0.0866,
+      "step": 10558
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.7529711772412506e-05,
+      "loss": 0.076,
+      "step": 10560
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.7529243870484745e-05,
+      "loss": 0.0838,
+      "step": 10562
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.752877596855699e-05,
+      "loss": 0.0642,
+      "step": 10564
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.752830806662924e-05,
+      "loss": 0.0777,
+      "step": 10566
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.752784016470148e-05,
+      "loss": 0.0674,
+      "step": 10568
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.752737226277372e-05,
+      "loss": 0.0672,
+      "step": 10570
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.752690436084597e-05,
+      "loss": 0.0698,
+      "step": 10572
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.7526436458918214e-05,
+      "loss": 0.0635,
+      "step": 10574
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.752596855699046e-05,
+      "loss": 0.0889,
+      "step": 10576
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.75255006550627e-05,
+      "loss": 0.0894,
+      "step": 10578
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7525032753134945e-05,
+      "loss": 0.0573,
+      "step": 10580
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.752456485120719e-05,
+      "loss": 0.0707,
+      "step": 10582
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7524096949279436e-05,
+      "loss": 0.0701,
+      "step": 10584
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7523629047351676e-05,
+      "loss": 0.066,
+      "step": 10586
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.752316114542392e-05,
+      "loss": 0.0738,
+      "step": 10588
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.752269324349616e-05,
+      "loss": 0.073,
+      "step": 10590
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.752222534156841e-05,
+      "loss": 0.0956,
+      "step": 10592
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.752175743964065e-05,
+      "loss": 0.0718,
+      "step": 10594
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.75212895377129e-05,
+      "loss": 0.0701,
+      "step": 10596
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.752082163578514e-05,
+      "loss": 0.0723,
+      "step": 10598
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.752035373385738e-05,
+      "loss": 0.0728,
+      "step": 10600
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.751988583192963e-05,
+      "loss": 0.0656,
+      "step": 10602
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7519417930001875e-05,
+      "loss": 0.0809,
+      "step": 10604
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7518950028074114e-05,
+      "loss": 0.0736,
+      "step": 10606
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.751848212614636e-05,
+      "loss": 0.0861,
+      "step": 10608
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7518014224218606e-05,
+      "loss": 0.0679,
+      "step": 10610
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.751754632229085e-05,
+      "loss": 0.0859,
+      "step": 10612
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.751707842036309e-05,
+      "loss": 0.0857,
+      "step": 10614
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.751661051843534e-05,
+      "loss": 0.0977,
+      "step": 10616
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.751614261650758e-05,
+      "loss": 0.0763,
+      "step": 10618
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.751567471457983e-05,
+      "loss": 0.0703,
+      "step": 10620
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.751520681265207e-05,
+      "loss": 0.0713,
+      "step": 10622
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7514738910724314e-05,
+      "loss": 0.0648,
+      "step": 10624
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.751427100879656e-05,
+      "loss": 0.0707,
+      "step": 10626
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7513803106868806e-05,
+      "loss": 0.077,
+      "step": 10628
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7513335204941045e-05,
+      "loss": 0.102,
+      "step": 10630
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.751286730301329e-05,
+      "loss": 0.067,
+      "step": 10632
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.751239940108553e-05,
+      "loss": 0.0642,
+      "step": 10634
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.751193149915778e-05,
+      "loss": 0.0691,
+      "step": 10636
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.751146359723002e-05,
+      "loss": 0.0809,
+      "step": 10638
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.751099569530227e-05,
+      "loss": 0.0847,
+      "step": 10640
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.751052779337451e-05,
+      "loss": 0.069,
+      "step": 10642
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.751005989144676e-05,
+      "loss": 0.0711,
+      "step": 10644
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7509591989519e-05,
+      "loss": 0.0763,
+      "step": 10646
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7509124087591245e-05,
+      "loss": 0.0765,
+      "step": 10648
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7508656185663484e-05,
+      "loss": 0.1077,
+      "step": 10650
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.750818828373573e-05,
+      "loss": 0.0587,
+      "step": 10652
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.7507720381807976e-05,
+      "loss": 0.0733,
+      "step": 10654
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.750725247988022e-05,
+      "loss": 0.0625,
+      "step": 10656
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.750678457795246e-05,
+      "loss": 0.0723,
+      "step": 10658
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.7506316676024707e-05,
+      "loss": 0.0725,
+      "step": 10660
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.750584877409695e-05,
+      "loss": 0.0743,
+      "step": 10662
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.75053808721692e-05,
+      "loss": 0.0622,
+      "step": 10664
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.750491297024144e-05,
+      "loss": 0.1071,
+      "step": 10666
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.7504445068313683e-05,
+      "loss": 0.0714,
+      "step": 10668
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.750397716638593e-05,
+      "loss": 0.05,
+      "step": 10670
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.7503509264458175e-05,
+      "loss": 0.0744,
+      "step": 10672
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.7503041362530414e-05,
+      "loss": 0.0863,
+      "step": 10674
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.750257346060266e-05,
+      "loss": 0.1183,
+      "step": 10676
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.7502105558674906e-05,
+      "loss": 0.1107,
+      "step": 10678
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.750163765674715e-05,
+      "loss": 0.0695,
+      "step": 10680
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.750116975481939e-05,
+      "loss": 0.0695,
+      "step": 10682
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.750070185289164e-05,
+      "loss": 0.0654,
+      "step": 10684
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.7500233950963876e-05,
+      "loss": 0.0953,
+      "step": 10686
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.749976604903613e-05,
+      "loss": 0.0743,
+      "step": 10688
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.749929814710837e-05,
+      "loss": 0.073,
+      "step": 10690
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.7498830245180614e-05,
+      "loss": 0.0664,
+      "step": 10692
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.749836234325285e-05,
+      "loss": 0.0779,
+      "step": 10694
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.7497894441325106e-05,
+      "loss": 0.0899,
+      "step": 10696
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.7497426539397345e-05,
+      "loss": 0.0873,
+      "step": 10698
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.749695863746959e-05,
+      "loss": 0.0767,
+      "step": 10700
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.749649073554183e-05,
+      "loss": 0.0691,
+      "step": 10702
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.7496022833614076e-05,
+      "loss": 0.0822,
+      "step": 10704
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.749555493168632e-05,
+      "loss": 0.0739,
+      "step": 10706
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.749508702975857e-05,
+      "loss": 0.0483,
+      "step": 10708
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.749461912783081e-05,
+      "loss": 0.0882,
+      "step": 10710
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.749415122590305e-05,
+      "loss": 0.0757,
+      "step": 10712
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.74936833239753e-05,
+      "loss": 0.0726,
+      "step": 10714
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.7493215422047545e-05,
+      "loss": 0.0577,
+      "step": 10716
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.7492747520119784e-05,
+      "loss": 0.0701,
+      "step": 10718
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.749227961819203e-05,
+      "loss": 0.1074,
+      "step": 10720
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.7491811716264276e-05,
+      "loss": 0.0624,
+      "step": 10722
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.749134381433652e-05,
+      "loss": 0.0882,
+      "step": 10724
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.749087591240876e-05,
+      "loss": 0.0645,
+      "step": 10726
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.7490408010481007e-05,
+      "loss": 0.0711,
+      "step": 10728
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.748994010855325e-05,
+      "loss": 0.0796,
+      "step": 10730
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.74894722066255e-05,
+      "loss": 0.0767,
+      "step": 10732
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.748900430469774e-05,
+      "loss": 0.0687,
+      "step": 10734
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.7488536402769983e-05,
+      "loss": 0.084,
+      "step": 10736
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.748806850084222e-05,
+      "loss": 0.067,
+      "step": 10738
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.7487600598914475e-05,
+      "loss": 0.0952,
+      "step": 10740
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.7487132696986714e-05,
+      "loss": 0.0873,
+      "step": 10742
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.748666479505896e-05,
+      "loss": 0.0961,
+      "step": 10744
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.74861968931312e-05,
+      "loss": 0.0729,
+      "step": 10746
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.7485728991203445e-05,
+      "loss": 0.0772,
+      "step": 10748
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.748526108927569e-05,
+      "loss": 0.0884,
+      "step": 10750
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.748479318734794e-05,
+      "loss": 0.0868,
+      "step": 10752
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.7484325285420176e-05,
+      "loss": 0.0791,
+      "step": 10754
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.748385738349242e-05,
+      "loss": 0.0874,
+      "step": 10756
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.748338948156467e-05,
+      "loss": 0.0987,
+      "step": 10758
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.7482921579636914e-05,
+      "loss": 0.0765,
+      "step": 10760
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.748245367770915e-05,
+      "loss": 0.0673,
+      "step": 10762
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.74819857757814e-05,
+      "loss": 0.0719,
+      "step": 10764
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.7481517873853645e-05,
+      "loss": 0.0771,
+      "step": 10766
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.7481049971925884e-05,
+      "loss": 0.0686,
+      "step": 10768
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.748058206999813e-05,
+      "loss": 0.0705,
+      "step": 10770
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.748011416807037e-05,
+      "loss": 0.0792,
+      "step": 10772
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.747964626614262e-05,
+      "loss": 0.072,
+      "step": 10774
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.747917836421486e-05,
+      "loss": 0.06,
+      "step": 10776
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.747871046228711e-05,
+      "loss": 0.1011,
+      "step": 10778
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.7478242560359346e-05,
+      "loss": 0.0713,
+      "step": 10780
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.747777465843159e-05,
+      "loss": 0.0661,
+      "step": 10782
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.747730675650384e-05,
+      "loss": 0.0876,
+      "step": 10784
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.7476838854576084e-05,
+      "loss": 0.0917,
+      "step": 10786
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.747637095264832e-05,
+      "loss": 0.1038,
+      "step": 10788
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.747590305072057e-05,
+      "loss": 0.0869,
+      "step": 10790
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.7475435148792815e-05,
+      "loss": 0.0682,
+      "step": 10792
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.747496724686506e-05,
+      "loss": 0.1075,
+      "step": 10794
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.74744993449373e-05,
+      "loss": 0.0597,
+      "step": 10796
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.7474031443009546e-05,
+      "loss": 0.0841,
+      "step": 10798
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.747356354108179e-05,
+      "loss": 0.0933,
+      "step": 10800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.747309563915404e-05,
+      "loss": 0.0673,
+      "step": 10802
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.747262773722628e-05,
+      "loss": 0.0675,
+      "step": 10804
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.747215983529852e-05,
+      "loss": 0.0756,
+      "step": 10806
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.747169193337077e-05,
+      "loss": 0.07,
+      "step": 10808
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.7471224031443014e-05,
+      "loss": 0.0655,
+      "step": 10810
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.7470756129515254e-05,
+      "loss": 0.074,
+      "step": 10812
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.74702882275875e-05,
+      "loss": 0.0651,
+      "step": 10814
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.746982032565974e-05,
+      "loss": 0.0483,
+      "step": 10816
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.746935242373199e-05,
+      "loss": 0.0795,
+      "step": 10818
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.746888452180423e-05,
+      "loss": 0.0669,
+      "step": 10820
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.7468416619876476e-05,
+      "loss": 0.0668,
+      "step": 10822
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.7467948717948715e-05,
+      "loss": 0.0602,
+      "step": 10824
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.746748081602097e-05,
+      "loss": 0.0764,
+      "step": 10826
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.746701291409321e-05,
+      "loss": 0.0816,
+      "step": 10828
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.746654501216545e-05,
+      "loss": 0.1023,
+      "step": 10830
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.746607711023769e-05,
+      "loss": 0.0804,
+      "step": 10832
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.746560920830994e-05,
+      "loss": 0.1046,
+      "step": 10834
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.7465141306382184e-05,
+      "loss": 0.0669,
+      "step": 10836
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.746467340445443e-05,
+      "loss": 0.0675,
+      "step": 10838
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.746420550252667e-05,
+      "loss": 0.1118,
+      "step": 10840
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.7463737600598915e-05,
+      "loss": 0.0717,
+      "step": 10842
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.746326969867116e-05,
+      "loss": 0.0678,
+      "step": 10844
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.746280179674341e-05,
+      "loss": 0.0864,
+      "step": 10846
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.7462333894815646e-05,
+      "loss": 0.0957,
+      "step": 10848
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.746186599288789e-05,
+      "loss": 0.0651,
+      "step": 10850
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.746139809096014e-05,
+      "loss": 0.0623,
+      "step": 10852
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.7460930189032384e-05,
+      "loss": 0.0774,
+      "step": 10854
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.746046228710462e-05,
+      "loss": 0.0969,
+      "step": 10856
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.745999438517687e-05,
+      "loss": 0.0752,
+      "step": 10858
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.7459526483249115e-05,
+      "loss": 0.0795,
+      "step": 10860
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.745905858132136e-05,
+      "loss": 0.0802,
+      "step": 10862
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.74585906793936e-05,
+      "loss": 0.0815,
+      "step": 10864
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7458122777465846e-05,
+      "loss": 0.0612,
+      "step": 10866
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7457654875538085e-05,
+      "loss": 0.0575,
+      "step": 10868
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.745718697361034e-05,
+      "loss": 0.0899,
+      "step": 10870
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.745671907168258e-05,
+      "loss": 0.082,
+      "step": 10872
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.745625116975482e-05,
+      "loss": 0.084,
+      "step": 10874
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.745578326782706e-05,
+      "loss": 0.0872,
+      "step": 10876
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7455315365899314e-05,
+      "loss": 0.0752,
+      "step": 10878
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7454847463971554e-05,
+      "loss": 0.0753,
+      "step": 10880
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.74543795620438e-05,
+      "loss": 0.0819,
+      "step": 10882
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.745391166011604e-05,
+      "loss": 0.0843,
+      "step": 10884
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7453443758188284e-05,
+      "loss": 0.098,
+      "step": 10886
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.745297585626053e-05,
+      "loss": 0.0839,
+      "step": 10888
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7452507954332776e-05,
+      "loss": 0.0642,
+      "step": 10890
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7452040052405015e-05,
+      "loss": 0.0771,
+      "step": 10892
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.745157215047726e-05,
+      "loss": 0.08,
+      "step": 10894
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.745110424854951e-05,
+      "loss": 0.0784,
+      "step": 10896
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.745063634662175e-05,
+      "loss": 0.0784,
+      "step": 10898
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.745016844469399e-05,
+      "loss": 0.0807,
+      "step": 10900
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.744970054276624e-05,
+      "loss": 0.0517,
+      "step": 10902
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7449232640838484e-05,
+      "loss": 0.0802,
+      "step": 10904
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.744876473891073e-05,
+      "loss": 0.0746,
+      "step": 10906
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.744829683698297e-05,
+      "loss": 0.0678,
+      "step": 10908
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7447828935055215e-05,
+      "loss": 0.0581,
+      "step": 10910
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7447361033127454e-05,
+      "loss": 0.0775,
+      "step": 10912
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.744689313119971e-05,
+      "loss": 0.0733,
+      "step": 10914
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7446425229271946e-05,
+      "loss": 0.0898,
+      "step": 10916
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.744595732734419e-05,
+      "loss": 0.1295,
+      "step": 10918
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.744548942541643e-05,
+      "loss": 0.0819,
+      "step": 10920
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7445021523488684e-05,
+      "loss": 0.0753,
+      "step": 10922
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.744455362156092e-05,
+      "loss": 0.0874,
+      "step": 10924
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.744408571963317e-05,
+      "loss": 0.0692,
+      "step": 10926
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.744361781770541e-05,
+      "loss": 0.1035,
+      "step": 10928
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7443149915777654e-05,
+      "loss": 0.0888,
+      "step": 10930
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.74426820138499e-05,
+      "loss": 0.0741,
+      "step": 10932
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7442214111922146e-05,
+      "loss": 0.0764,
+      "step": 10934
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.7441746209994385e-05,
+      "loss": 0.0679,
+      "step": 10936
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.744127830806663e-05,
+      "loss": 0.0854,
+      "step": 10938
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.744081040613888e-05,
+      "loss": 0.0817,
+      "step": 10940
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.744034250421112e-05,
+      "loss": 0.0852,
+      "step": 10942
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.743987460228336e-05,
+      "loss": 0.0856,
+      "step": 10944
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.743940670035561e-05,
+      "loss": 0.0787,
+      "step": 10946
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.7438938798427854e-05,
+      "loss": 0.0854,
+      "step": 10948
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.74384708965001e-05,
+      "loss": 0.0775,
+      "step": 10950
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.743800299457234e-05,
+      "loss": 0.0794,
+      "step": 10952
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.7437535092644585e-05,
+      "loss": 0.0756,
+      "step": 10954
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.743706719071683e-05,
+      "loss": 0.085,
+      "step": 10956
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.7436599288789076e-05,
+      "loss": 0.0747,
+      "step": 10958
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.7436131386861315e-05,
+      "loss": 0.0997,
+      "step": 10960
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.743566348493356e-05,
+      "loss": 0.0884,
+      "step": 10962
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.74351955830058e-05,
+      "loss": 0.1031,
+      "step": 10964
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.743472768107805e-05,
+      "loss": 0.0583,
+      "step": 10966
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.743425977915029e-05,
+      "loss": 0.0806,
+      "step": 10968
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.743379187722254e-05,
+      "loss": 0.0795,
+      "step": 10970
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.743332397529478e-05,
+      "loss": 0.0618,
+      "step": 10972
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.743285607336703e-05,
+      "loss": 0.0768,
+      "step": 10974
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.743238817143927e-05,
+      "loss": 0.0965,
+      "step": 10976
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.7431920269511515e-05,
+      "loss": 0.0713,
+      "step": 10978
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.7431452367583754e-05,
+      "loss": 0.072,
+      "step": 10980
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.7430984465656e-05,
+      "loss": 0.0749,
+      "step": 10982
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.7430516563728246e-05,
+      "loss": 0.0773,
+      "step": 10984
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.743004866180049e-05,
+      "loss": 0.0777,
+      "step": 10986
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.742958075987273e-05,
+      "loss": 0.0919,
+      "step": 10988
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.742911285794498e-05,
+      "loss": 0.0824,
+      "step": 10990
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.742864495601722e-05,
+      "loss": 0.0676,
+      "step": 10992
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.742817705408947e-05,
+      "loss": 0.0821,
+      "step": 10994
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.742770915216171e-05,
+      "loss": 0.0648,
+      "step": 10996
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.7427241250233954e-05,
+      "loss": 0.0847,
+      "step": 10998
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.74267733483062e-05,
+      "loss": 0.0878,
+      "step": 11000
+    },
+    {
+      "epoch": 1.54,
+      "eval_gen_len": 29.569,
+      "eval_loss": 1.0419230461120605,
+      "eval_meteor": 0.0426,
+      "eval_runtime": 14.397,
+      "eval_samples_per_second": 4.029,
+      "eval_steps_per_second": 0.556,
+      "step": 11000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.7426305446378446e-05,
+      "loss": 0.0881,
+      "step": 11002
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.7425837544450685e-05,
+      "loss": 0.0838,
+      "step": 11004
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.742536964252293e-05,
+      "loss": 0.0721,
+      "step": 11006
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.742490174059518e-05,
+      "loss": 0.1109,
+      "step": 11008
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.742443383866742e-05,
+      "loss": 0.1078,
+      "step": 11010
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.742396593673966e-05,
+      "loss": 0.0749,
+      "step": 11012
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.742349803481191e-05,
+      "loss": 0.0613,
+      "step": 11014
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.742303013288415e-05,
+      "loss": 0.0906,
+      "step": 11016
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.742256223095639e-05,
+      "loss": 0.0931,
+      "step": 11018
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.742209432902864e-05,
+      "loss": 0.0662,
+      "step": 11020
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.742162642710088e-05,
+      "loss": 0.0639,
+      "step": 11022
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7421158525173124e-05,
+      "loss": 0.0959,
+      "step": 11024
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.742069062324537e-05,
+      "loss": 0.096,
+      "step": 11026
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7420222721317615e-05,
+      "loss": 0.0694,
+      "step": 11028
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7419754819389855e-05,
+      "loss": 0.0707,
+      "step": 11030
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.74192869174621e-05,
+      "loss": 0.068,
+      "step": 11032
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7418819015534346e-05,
+      "loss": 0.075,
+      "step": 11034
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.741835111360659e-05,
+      "loss": 0.0659,
+      "step": 11036
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.741788321167883e-05,
+      "loss": 0.0663,
+      "step": 11038
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.741741530975108e-05,
+      "loss": 0.1099,
+      "step": 11040
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.741694740782332e-05,
+      "loss": 0.0676,
+      "step": 11042
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.741647950589557e-05,
+      "loss": 0.0628,
+      "step": 11044
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.741601160396781e-05,
+      "loss": 0.0762,
+      "step": 11046
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7415543702040054e-05,
+      "loss": 0.1015,
+      "step": 11048
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7415075800112293e-05,
+      "loss": 0.0709,
+      "step": 11050
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7414607898184546e-05,
+      "loss": 0.0794,
+      "step": 11052
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7414139996256785e-05,
+      "loss": 0.097,
+      "step": 11054
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.741367209432903e-05,
+      "loss": 0.0749,
+      "step": 11056
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.741320419240127e-05,
+      "loss": 0.0902,
+      "step": 11058
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7412736290473516e-05,
+      "loss": 0.0622,
+      "step": 11060
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.741226838854576e-05,
+      "loss": 0.0613,
+      "step": 11062
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.741180048661801e-05,
+      "loss": 0.0799,
+      "step": 11064
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.741133258469025e-05,
+      "loss": 0.0979,
+      "step": 11066
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.741086468276249e-05,
+      "loss": 0.0748,
+      "step": 11068
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.741039678083474e-05,
+      "loss": 0.0949,
+      "step": 11070
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7409928878906985e-05,
+      "loss": 0.0605,
+      "step": 11072
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.7409460976979224e-05,
+      "loss": 0.0771,
+      "step": 11074
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.740899307505147e-05,
+      "loss": 0.0686,
+      "step": 11076
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.7408525173123716e-05,
+      "loss": 0.0721,
+      "step": 11078
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.740805727119596e-05,
+      "loss": 0.0767,
+      "step": 11080
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.74075893692682e-05,
+      "loss": 0.0871,
+      "step": 11082
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.740712146734045e-05,
+      "loss": 0.0645,
+      "step": 11084
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.740665356541269e-05,
+      "loss": 0.0693,
+      "step": 11086
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.740618566348494e-05,
+      "loss": 0.0795,
+      "step": 11088
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.740571776155718e-05,
+      "loss": 0.0766,
+      "step": 11090
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.7405249859629424e-05,
+      "loss": 0.0742,
+      "step": 11092
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.740478195770166e-05,
+      "loss": 0.0835,
+      "step": 11094
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.7404314055773916e-05,
+      "loss": 0.1059,
+      "step": 11096
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.7403846153846155e-05,
+      "loss": 0.0674,
+      "step": 11098
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.74033782519184e-05,
+      "loss": 0.1083,
+      "step": 11100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.740291034999064e-05,
+      "loss": 0.0774,
+      "step": 11102
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.740244244806289e-05,
+      "loss": 0.0647,
+      "step": 11104
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.740197454613513e-05,
+      "loss": 0.0657,
+      "step": 11106
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.740150664420738e-05,
+      "loss": 0.0769,
+      "step": 11108
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.7401038742279617e-05,
+      "loss": 0.0954,
+      "step": 11110
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.740057084035186e-05,
+      "loss": 0.0713,
+      "step": 11112
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.740010293842411e-05,
+      "loss": 0.1011,
+      "step": 11114
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.7399635036496354e-05,
+      "loss": 0.0694,
+      "step": 11116
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.7399167134568593e-05,
+      "loss": 0.0856,
+      "step": 11118
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.739869923264084e-05,
+      "loss": 0.0795,
+      "step": 11120
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.7398231330713085e-05,
+      "loss": 0.075,
+      "step": 11122
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.739776342878533e-05,
+      "loss": 0.0653,
+      "step": 11124
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.739729552685757e-05,
+      "loss": 0.079,
+      "step": 11126
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.7396827624929816e-05,
+      "loss": 0.0973,
+      "step": 11128
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.739635972300206e-05,
+      "loss": 0.0547,
+      "step": 11130
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.739589182107431e-05,
+      "loss": 0.1045,
+      "step": 11132
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.739542391914655e-05,
+      "loss": 0.0797,
+      "step": 11134
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.739495601721879e-05,
+      "loss": 0.0819,
+      "step": 11136
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.739448811529104e-05,
+      "loss": 0.0838,
+      "step": 11138
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.7394020213363285e-05,
+      "loss": 0.0873,
+      "step": 11140
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.7393552311435524e-05,
+      "loss": 0.0934,
+      "step": 11142
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.739308440950777e-05,
+      "loss": 0.071,
+      "step": 11144
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.739261650758001e-05,
+      "loss": 0.0773,
+      "step": 11146
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.739214860565226e-05,
+      "loss": 0.0948,
+      "step": 11148
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.73916807037245e-05,
+      "loss": 0.0941,
+      "step": 11150
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.739121280179675e-05,
+      "loss": 0.082,
+      "step": 11152
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7390744899868986e-05,
+      "loss": 0.0725,
+      "step": 11154
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.739027699794124e-05,
+      "loss": 0.0982,
+      "step": 11156
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.738980909601348e-05,
+      "loss": 0.0918,
+      "step": 11158
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7389341194085724e-05,
+      "loss": 0.0903,
+      "step": 11160
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.738887329215796e-05,
+      "loss": 0.0882,
+      "step": 11162
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.738840539023021e-05,
+      "loss": 0.0734,
+      "step": 11164
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7387937488302455e-05,
+      "loss": 0.0719,
+      "step": 11166
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.73874695863747e-05,
+      "loss": 0.088,
+      "step": 11168
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.738700168444694e-05,
+      "loss": 0.073,
+      "step": 11170
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7386533782519186e-05,
+      "loss": 0.0931,
+      "step": 11172
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.738606588059143e-05,
+      "loss": 0.1425,
+      "step": 11174
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.738559797866368e-05,
+      "loss": 0.064,
+      "step": 11176
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7385130076735917e-05,
+      "loss": 0.0759,
+      "step": 11178
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.738466217480816e-05,
+      "loss": 0.1033,
+      "step": 11180
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.738419427288041e-05,
+      "loss": 0.0938,
+      "step": 11182
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7383726370952654e-05,
+      "loss": 0.0601,
+      "step": 11184
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7383258469024893e-05,
+      "loss": 0.0816,
+      "step": 11186
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.738279056709714e-05,
+      "loss": 0.0874,
+      "step": 11188
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7382322665169385e-05,
+      "loss": 0.0709,
+      "step": 11190
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.738185476324163e-05,
+      "loss": 0.0896,
+      "step": 11192
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.738138686131387e-05,
+      "loss": 0.0777,
+      "step": 11194
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7380918959386116e-05,
+      "loss": 0.0628,
+      "step": 11196
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7380451057458355e-05,
+      "loss": 0.1021,
+      "step": 11198
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.737998315553061e-05,
+      "loss": 0.073,
+      "step": 11200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.737951525360285e-05,
+      "loss": 0.0818,
+      "step": 11202
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.737904735167509e-05,
+      "loss": 0.0749,
+      "step": 11204
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.737857944974733e-05,
+      "loss": 0.0763,
+      "step": 11206
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.737811154781958e-05,
+      "loss": 0.069,
+      "step": 11208
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7377643645891824e-05,
+      "loss": 0.0722,
+      "step": 11210
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.737717574396407e-05,
+      "loss": 0.0952,
+      "step": 11212
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.737670784203631e-05,
+      "loss": 0.0932,
+      "step": 11214
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7376239940108555e-05,
+      "loss": 0.0806,
+      "step": 11216
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.73757720381808e-05,
+      "loss": 0.0695,
+      "step": 11218
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.737530413625305e-05,
+      "loss": 0.0598,
+      "step": 11220
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.7374836234325286e-05,
+      "loss": 0.1067,
+      "step": 11222
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.737436833239753e-05,
+      "loss": 0.0689,
+      "step": 11224
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.737390043046978e-05,
+      "loss": 0.1037,
+      "step": 11226
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.7373432528542024e-05,
+      "loss": 0.0744,
+      "step": 11228
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.737296462661426e-05,
+      "loss": 0.0758,
+      "step": 11230
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.737249672468651e-05,
+      "loss": 0.0667,
+      "step": 11232
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.7372028822758755e-05,
+      "loss": 0.0683,
+      "step": 11234
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.7371560920831e-05,
+      "loss": 0.0885,
+      "step": 11236
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.737109301890324e-05,
+      "loss": 0.0687,
+      "step": 11238
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.7370625116975486e-05,
+      "loss": 0.0814,
+      "step": 11240
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.7370157215047725e-05,
+      "loss": 0.0836,
+      "step": 11242
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.736968931311998e-05,
+      "loss": 0.0732,
+      "step": 11244
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.7369221411192217e-05,
+      "loss": 0.0917,
+      "step": 11246
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.736875350926446e-05,
+      "loss": 0.0698,
+      "step": 11248
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.73682856073367e-05,
+      "loss": 0.0857,
+      "step": 11250
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.7367817705408954e-05,
+      "loss": 0.0953,
+      "step": 11252
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.7367349803481193e-05,
+      "loss": 0.0474,
+      "step": 11254
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.736688190155344e-05,
+      "loss": 0.0712,
+      "step": 11256
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.736641399962568e-05,
+      "loss": 0.0865,
+      "step": 11258
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.7365946097697924e-05,
+      "loss": 0.0683,
+      "step": 11260
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.736547819577017e-05,
+      "loss": 0.0531,
+      "step": 11262
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.7365010293842416e-05,
+      "loss": 0.078,
+      "step": 11264
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.7364542391914655e-05,
+      "loss": 0.1029,
+      "step": 11266
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.73640744899869e-05,
+      "loss": 0.0774,
+      "step": 11268
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.736360658805915e-05,
+      "loss": 0.0738,
+      "step": 11270
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.7363138686131386e-05,
+      "loss": 0.084,
+      "step": 11272
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.736267078420363e-05,
+      "loss": 0.0607,
+      "step": 11274
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.736220288227587e-05,
+      "loss": 0.0743,
+      "step": 11276
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.7361734980348124e-05,
+      "loss": 0.0675,
+      "step": 11278
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.736126707842036e-05,
+      "loss": 0.0873,
+      "step": 11280
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.736079917649261e-05,
+      "loss": 0.0808,
+      "step": 11282
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.736033127456485e-05,
+      "loss": 0.0797,
+      "step": 11284
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.73598633726371e-05,
+      "loss": 0.0756,
+      "step": 11286
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.735939547070934e-05,
+      "loss": 0.0828,
+      "step": 11288
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.7358927568781586e-05,
+      "loss": 0.0776,
+      "step": 11290
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.7358459666853825e-05,
+      "loss": 0.1002,
+      "step": 11292
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.735799176492607e-05,
+      "loss": 0.0764,
+      "step": 11294
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.735752386299832e-05,
+      "loss": 0.0853,
+      "step": 11296
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.735705596107056e-05,
+      "loss": 0.0628,
+      "step": 11298
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.73565880591428e-05,
+      "loss": 0.0834,
+      "step": 11300
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.735612015721505e-05,
+      "loss": 0.0747,
+      "step": 11302
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.7355652255287294e-05,
+      "loss": 0.0772,
+      "step": 11304
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.735518435335954e-05,
+      "loss": 0.0777,
+      "step": 11306
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.735471645143178e-05,
+      "loss": 0.0768,
+      "step": 11308
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.7354248549504025e-05,
+      "loss": 0.0784,
+      "step": 11310
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.735378064757627e-05,
+      "loss": 0.0723,
+      "step": 11312
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.735331274564852e-05,
+      "loss": 0.0832,
+      "step": 11314
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.7352844843720756e-05,
+      "loss": 0.086,
+      "step": 11316
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.7352376941793e-05,
+      "loss": 0.087,
+      "step": 11318
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.735190903986525e-05,
+      "loss": 0.0731,
+      "step": 11320
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.7351441137937493e-05,
+      "loss": 0.0677,
+      "step": 11322
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.735097323600973e-05,
+      "loss": 0.0751,
+      "step": 11324
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.735050533408198e-05,
+      "loss": 0.071,
+      "step": 11326
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.735003743215422e-05,
+      "loss": 0.0663,
+      "step": 11328
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.734956953022647e-05,
+      "loss": 0.0683,
+      "step": 11330
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.734910162829871e-05,
+      "loss": 0.0889,
+      "step": 11332
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.7348633726370955e-05,
+      "loss": 0.0515,
+      "step": 11334
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.7348165824443195e-05,
+      "loss": 0.0791,
+      "step": 11336
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.734769792251544e-05,
+      "loss": 0.0804,
+      "step": 11338
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.7347230020587686e-05,
+      "loss": 0.0884,
+      "step": 11340
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.734676211865993e-05,
+      "loss": 0.0555,
+      "step": 11342
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.734629421673217e-05,
+      "loss": 0.0682,
+      "step": 11344
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.734582631480442e-05,
+      "loss": 0.0737,
+      "step": 11346
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.734535841287666e-05,
+      "loss": 0.0816,
+      "step": 11348
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.734489051094891e-05,
+      "loss": 0.1033,
+      "step": 11350
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.734442260902115e-05,
+      "loss": 0.0915,
+      "step": 11352
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.7343954707093394e-05,
+      "loss": 0.0765,
+      "step": 11354
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.734348680516564e-05,
+      "loss": 0.0822,
+      "step": 11356
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.7343018903237886e-05,
+      "loss": 0.0814,
+      "step": 11358
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.7342551001310125e-05,
+      "loss": 0.0828,
+      "step": 11360
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.734208309938237e-05,
+      "loss": 0.0481,
+      "step": 11362
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.734161519745462e-05,
+      "loss": 0.0683,
+      "step": 11364
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.734114729552686e-05,
+      "loss": 0.0599,
+      "step": 11366
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.73406793935991e-05,
+      "loss": 0.0598,
+      "step": 11368
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.734021149167135e-05,
+      "loss": 0.0633,
+      "step": 11370
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.733974358974359e-05,
+      "loss": 0.0766,
+      "step": 11372
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.733927568781584e-05,
+      "loss": 0.0828,
+      "step": 11374
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.733880778588808e-05,
+      "loss": 0.0787,
+      "step": 11376
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.7338339883960325e-05,
+      "loss": 0.074,
+      "step": 11378
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.7337871982032564e-05,
+      "loss": 0.0945,
+      "step": 11380
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.733740408010482e-05,
+      "loss": 0.0493,
+      "step": 11382
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.7336936178177056e-05,
+      "loss": 0.0778,
+      "step": 11384
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.73364682762493e-05,
+      "loss": 0.0715,
+      "step": 11386
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.733600037432154e-05,
+      "loss": 0.0661,
+      "step": 11388
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.733553247239379e-05,
+      "loss": 0.0892,
+      "step": 11390
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.733506457046603e-05,
+      "loss": 0.0931,
+      "step": 11392
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.733459666853828e-05,
+      "loss": 0.0753,
+      "step": 11394
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.733412876661052e-05,
+      "loss": 0.0888,
+      "step": 11396
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.7333660864682764e-05,
+      "loss": 0.0854,
+      "step": 11398
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.733319296275501e-05,
+      "loss": 0.0715,
+      "step": 11400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.7332725060827255e-05,
+      "loss": 0.089,
+      "step": 11402
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.7332257158899495e-05,
+      "loss": 0.0884,
+      "step": 11404
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.733178925697174e-05,
+      "loss": 0.0781,
+      "step": 11406
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.7331321355043986e-05,
+      "loss": 0.0712,
+      "step": 11408
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.733085345311623e-05,
+      "loss": 0.064,
+      "step": 11410
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.733038555118847e-05,
+      "loss": 0.0692,
+      "step": 11412
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.732991764926072e-05,
+      "loss": 0.085,
+      "step": 11414
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.732944974733296e-05,
+      "loss": 0.0743,
+      "step": 11416
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.732898184540521e-05,
+      "loss": 0.0652,
+      "step": 11418
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.732851394347745e-05,
+      "loss": 0.0898,
+      "step": 11420
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.7328046041549694e-05,
+      "loss": 0.1076,
+      "step": 11422
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.732757813962193e-05,
+      "loss": 0.073,
+      "step": 11424
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.7327110237694186e-05,
+      "loss": 0.0987,
+      "step": 11426
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.7326642335766425e-05,
+      "loss": 0.0615,
+      "step": 11428
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.732617443383867e-05,
+      "loss": 0.0756,
+      "step": 11430
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.732570653191091e-05,
+      "loss": 0.0616,
+      "step": 11432
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.732523862998316e-05,
+      "loss": 0.0834,
+      "step": 11434
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.73247707280554e-05,
+      "loss": 0.0648,
+      "step": 11436
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.732430282612765e-05,
+      "loss": 0.079,
+      "step": 11438
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.732383492419989e-05,
+      "loss": 0.0773,
+      "step": 11440
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.732336702227213e-05,
+      "loss": 0.0984,
+      "step": 11442
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.732289912034438e-05,
+      "loss": 0.0939,
+      "step": 11444
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.7322431218416625e-05,
+      "loss": 0.0867,
+      "step": 11446
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.7321963316488864e-05,
+      "loss": 0.0737,
+      "step": 11448
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.732149541456111e-05,
+      "loss": 0.062,
+      "step": 11450
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.7321027512633356e-05,
+      "loss": 0.0977,
+      "step": 11452
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.73205596107056e-05,
+      "loss": 0.0768,
+      "step": 11454
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.732009170877784e-05,
+      "loss": 0.0847,
+      "step": 11456
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.731962380685009e-05,
+      "loss": 0.0885,
+      "step": 11458
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.731915590492233e-05,
+      "loss": 0.1158,
+      "step": 11460
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.731868800299458e-05,
+      "loss": 0.0986,
+      "step": 11462
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.731822010106682e-05,
+      "loss": 0.088,
+      "step": 11464
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.7317752199139064e-05,
+      "loss": 0.0747,
+      "step": 11466
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.731728429721131e-05,
+      "loss": 0.0672,
+      "step": 11468
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.7316816395283555e-05,
+      "loss": 0.0751,
+      "step": 11470
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.7316348493355795e-05,
+      "loss": 0.0903,
+      "step": 11472
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.731588059142804e-05,
+      "loss": 0.0736,
+      "step": 11474
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.731541268950028e-05,
+      "loss": 0.0677,
+      "step": 11476
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.731494478757253e-05,
+      "loss": 0.0791,
+      "step": 11478
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.731447688564477e-05,
+      "loss": 0.0899,
+      "step": 11480
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.731400898371702e-05,
+      "loss": 0.0824,
+      "step": 11482
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.7313541081789256e-05,
+      "loss": 0.0991,
+      "step": 11484
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.73130731798615e-05,
+      "loss": 0.0627,
+      "step": 11486
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.731260527793375e-05,
+      "loss": 0.1129,
+      "step": 11488
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.7312137376005994e-05,
+      "loss": 0.0945,
+      "step": 11490
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.731166947407823e-05,
+      "loss": 0.0911,
+      "step": 11492
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.731120157215048e-05,
+      "loss": 0.0818,
+      "step": 11494
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.7310733670222725e-05,
+      "loss": 0.0887,
+      "step": 11496
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.731026576829497e-05,
+      "loss": 0.0907,
+      "step": 11498
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.730979786636721e-05,
+      "loss": 0.0902,
+      "step": 11500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.7309329964439456e-05,
+      "loss": 0.081,
+      "step": 11502
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.73088620625117e-05,
+      "loss": 0.0923,
+      "step": 11504
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.730839416058395e-05,
+      "loss": 0.0881,
+      "step": 11506
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.730792625865619e-05,
+      "loss": 0.1113,
+      "step": 11508
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.730745835672843e-05,
+      "loss": 0.0689,
+      "step": 11510
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.730699045480068e-05,
+      "loss": 0.0657,
+      "step": 11512
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.7306522552872925e-05,
+      "loss": 0.0717,
+      "step": 11514
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.7306054650945164e-05,
+      "loss": 0.1176,
+      "step": 11516
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.730558674901741e-05,
+      "loss": 0.0772,
+      "step": 11518
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.730511884708965e-05,
+      "loss": 0.0606,
+      "step": 11520
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.73046509451619e-05,
+      "loss": 0.0775,
+      "step": 11522
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.730418304323414e-05,
+      "loss": 0.0849,
+      "step": 11524
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.730371514130638e-05,
+      "loss": 0.0774,
+      "step": 11526
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.7303247239378626e-05,
+      "loss": 0.0619,
+      "step": 11528
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.730277933745087e-05,
+      "loss": 0.0864,
+      "step": 11530
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.730231143552312e-05,
+      "loss": 0.0747,
+      "step": 11532
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.730184353359536e-05,
+      "loss": 0.0905,
+      "step": 11534
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.73013756316676e-05,
+      "loss": 0.0724,
+      "step": 11536
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.730090772973985e-05,
+      "loss": 0.0693,
+      "step": 11538
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.7300439827812095e-05,
+      "loss": 0.0869,
+      "step": 11540
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.7299971925884334e-05,
+      "loss": 0.069,
+      "step": 11542
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.729950402395658e-05,
+      "loss": 0.068,
+      "step": 11544
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.7299036122028826e-05,
+      "loss": 0.0752,
+      "step": 11546
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.729856822010107e-05,
+      "loss": 0.0795,
+      "step": 11548
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.729810031817331e-05,
+      "loss": 0.071,
+      "step": 11550
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.7297632416245557e-05,
+      "loss": 0.0766,
+      "step": 11552
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.7297164514317796e-05,
+      "loss": 0.0614,
+      "step": 11554
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.729669661239005e-05,
+      "loss": 0.0749,
+      "step": 11556
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.729622871046229e-05,
+      "loss": 0.0758,
+      "step": 11558
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.729576080853453e-05,
+      "loss": 0.0642,
+      "step": 11560
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.729529290660677e-05,
+      "loss": 0.0569,
+      "step": 11562
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.7294825004679025e-05,
+      "loss": 0.0919,
+      "step": 11564
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.7294357102751264e-05,
+      "loss": 0.0871,
+      "step": 11566
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.729388920082351e-05,
+      "loss": 0.0754,
+      "step": 11568
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.729342129889575e-05,
+      "loss": 0.091,
+      "step": 11570
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.7292953396967995e-05,
+      "loss": 0.0681,
+      "step": 11572
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.729248549504024e-05,
+      "loss": 0.055,
+      "step": 11574
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.729201759311249e-05,
+      "loss": 0.0793,
+      "step": 11576
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.7291549691184726e-05,
+      "loss": 0.0748,
+      "step": 11578
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.729108178925697e-05,
+      "loss": 0.0647,
+      "step": 11580
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.729061388732922e-05,
+      "loss": 0.1023,
+      "step": 11582
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.7290145985401464e-05,
+      "loss": 0.0651,
+      "step": 11584
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.72896780834737e-05,
+      "loss": 0.0554,
+      "step": 11586
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.728921018154595e-05,
+      "loss": 0.0867,
+      "step": 11588
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.7288742279618195e-05,
+      "loss": 0.0707,
+      "step": 11590
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.728827437769044e-05,
+      "loss": 0.0832,
+      "step": 11592
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.728780647576268e-05,
+      "loss": 0.0965,
+      "step": 11594
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.7287338573834926e-05,
+      "loss": 0.08,
+      "step": 11596
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.728687067190717e-05,
+      "loss": 0.0704,
+      "step": 11598
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.728640276997942e-05,
+      "loss": 0.0839,
+      "step": 11600
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.728593486805166e-05,
+      "loss": 0.0665,
+      "step": 11602
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.72854669661239e-05,
+      "loss": 0.0895,
+      "step": 11604
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.728499906419614e-05,
+      "loss": 0.0749,
+      "step": 11606
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.7284531162268395e-05,
+      "loss": 0.0692,
+      "step": 11608
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.7284063260340634e-05,
+      "loss": 0.0591,
+      "step": 11610
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.728359535841288e-05,
+      "loss": 0.0833,
+      "step": 11612
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.728312745648512e-05,
+      "loss": 0.0784,
+      "step": 11614
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.7282659554557365e-05,
+      "loss": 0.0692,
+      "step": 11616
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.728219165262961e-05,
+      "loss": 0.0848,
+      "step": 11618
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.7281723750701857e-05,
+      "loss": 0.0873,
+      "step": 11620
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.7281255848774096e-05,
+      "loss": 0.0795,
+      "step": 11622
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.728078794684634e-05,
+      "loss": 0.0778,
+      "step": 11624
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.728032004491859e-05,
+      "loss": 0.078,
+      "step": 11626
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.7279852142990833e-05,
+      "loss": 0.0792,
+      "step": 11628
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.727938424106307e-05,
+      "loss": 0.098,
+      "step": 11630
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.727891633913532e-05,
+      "loss": 0.0812,
+      "step": 11632
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.7278448437207564e-05,
+      "loss": 0.0792,
+      "step": 11634
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.727798053527981e-05,
+      "loss": 0.0957,
+      "step": 11636
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.727751263335205e-05,
+      "loss": 0.0758,
+      "step": 11638
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.7277044731424295e-05,
+      "loss": 0.0885,
+      "step": 11640
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.727657682949654e-05,
+      "loss": 0.0937,
+      "step": 11642
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.727610892756879e-05,
+      "loss": 0.0795,
+      "step": 11644
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.7275641025641026e-05,
+      "loss": 0.0655,
+      "step": 11646
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.727517312371327e-05,
+      "loss": 0.0676,
+      "step": 11648
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.727470522178551e-05,
+      "loss": 0.064,
+      "step": 11650
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7274237319857764e-05,
+      "loss": 0.0745,
+      "step": 11652
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.727376941793e-05,
+      "loss": 0.0913,
+      "step": 11654
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.727330151600225e-05,
+      "loss": 0.0831,
+      "step": 11656
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.727283361407449e-05,
+      "loss": 0.0773,
+      "step": 11658
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.727236571214674e-05,
+      "loss": 0.0705,
+      "step": 11660
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.727189781021898e-05,
+      "loss": 0.092,
+      "step": 11662
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7271429908291226e-05,
+      "loss": 0.0741,
+      "step": 11664
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7270962006363465e-05,
+      "loss": 0.0664,
+      "step": 11666
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.727049410443571e-05,
+      "loss": 0.0743,
+      "step": 11668
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.727002620250796e-05,
+      "loss": 0.0884,
+      "step": 11670
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.72695583005802e-05,
+      "loss": 0.0837,
+      "step": 11672
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.726909039865244e-05,
+      "loss": 0.0726,
+      "step": 11674
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.726862249672469e-05,
+      "loss": 0.0882,
+      "step": 11676
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7268154594796934e-05,
+      "loss": 0.0856,
+      "step": 11678
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.726768669286918e-05,
+      "loss": 0.0879,
+      "step": 11680
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.726721879094142e-05,
+      "loss": 0.0768,
+      "step": 11682
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7266750889013665e-05,
+      "loss": 0.0636,
+      "step": 11684
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.726628298708591e-05,
+      "loss": 0.0611,
+      "step": 11686
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7265815085158157e-05,
+      "loss": 0.0699,
+      "step": 11688
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7265347183230396e-05,
+      "loss": 0.0718,
+      "step": 11690
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.726487928130264e-05,
+      "loss": 0.0774,
+      "step": 11692
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.726441137937489e-05,
+      "loss": 0.0655,
+      "step": 11694
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7263943477447133e-05,
+      "loss": 0.0833,
+      "step": 11696
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.726347557551937e-05,
+      "loss": 0.0958,
+      "step": 11698
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.726300767359162e-05,
+      "loss": 0.0883,
+      "step": 11700
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.726253977166386e-05,
+      "loss": 0.0853,
+      "step": 11702
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.726207186973611e-05,
+      "loss": 0.0585,
+      "step": 11704
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.726160396780835e-05,
+      "loss": 0.0812,
+      "step": 11706
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7261136065880595e-05,
+      "loss": 0.082,
+      "step": 11708
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7260668163952834e-05,
+      "loss": 0.0982,
+      "step": 11710
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.726020026202509e-05,
+      "loss": 0.0671,
+      "step": 11712
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.7259732360097326e-05,
+      "loss": 0.0791,
+      "step": 11714
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.725926445816957e-05,
+      "loss": 0.1029,
+      "step": 11716
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.725879655624181e-05,
+      "loss": 0.0689,
+      "step": 11718
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.725832865431406e-05,
+      "loss": 0.0912,
+      "step": 11720
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.72578607523863e-05,
+      "loss": 0.072,
+      "step": 11722
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.725739285045855e-05,
+      "loss": 0.0864,
+      "step": 11724
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.725692494853079e-05,
+      "loss": 0.0821,
+      "step": 11726
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.7256457046603034e-05,
+      "loss": 0.0626,
+      "step": 11728
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.725598914467528e-05,
+      "loss": 0.099,
+      "step": 11730
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.7255521242747526e-05,
+      "loss": 0.0964,
+      "step": 11732
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.7255053340819765e-05,
+      "loss": 0.0935,
+      "step": 11734
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.725458543889201e-05,
+      "loss": 0.0822,
+      "step": 11736
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.725411753696426e-05,
+      "loss": 0.0706,
+      "step": 11738
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.72536496350365e-05,
+      "loss": 0.065,
+      "step": 11740
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.725318173310874e-05,
+      "loss": 0.1036,
+      "step": 11742
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.725271383118099e-05,
+      "loss": 0.0634,
+      "step": 11744
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.7252245929253234e-05,
+      "loss": 0.0652,
+      "step": 11746
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.725177802732548e-05,
+      "loss": 0.0994,
+      "step": 11748
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.725131012539772e-05,
+      "loss": 0.0934,
+      "step": 11750
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.7250842223469965e-05,
+      "loss": 0.0844,
+      "step": 11752
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.7250374321542204e-05,
+      "loss": 0.0772,
+      "step": 11754
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.7249906419614457e-05,
+      "loss": 0.1009,
+      "step": 11756
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.7249438517686696e-05,
+      "loss": 0.086,
+      "step": 11758
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.724897061575894e-05,
+      "loss": 0.0648,
+      "step": 11760
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.724850271383118e-05,
+      "loss": 0.0779,
+      "step": 11762
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.724803481190343e-05,
+      "loss": 0.0707,
+      "step": 11764
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.724756690997567e-05,
+      "loss": 0.076,
+      "step": 11766
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.724709900804792e-05,
+      "loss": 0.0667,
+      "step": 11768
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.724663110612016e-05,
+      "loss": 0.0736,
+      "step": 11770
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.7246163204192404e-05,
+      "loss": 0.0739,
+      "step": 11772
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.724569530226465e-05,
+      "loss": 0.0869,
+      "step": 11774
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.724522740033689e-05,
+      "loss": 0.0745,
+      "step": 11776
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.7244759498409134e-05,
+      "loss": 0.0821,
+      "step": 11778
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.724429159648138e-05,
+      "loss": 0.0757,
+      "step": 11780
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.7243823694553626e-05,
+      "loss": 0.0797,
+      "step": 11782
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.7243355792625865e-05,
+      "loss": 0.0739,
+      "step": 11784
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.724288789069811e-05,
+      "loss": 0.0636,
+      "step": 11786
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.724241998877035e-05,
+      "loss": 0.0643,
+      "step": 11788
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.72419520868426e-05,
+      "loss": 0.1214,
+      "step": 11790
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.724148418491484e-05,
+      "loss": 0.0793,
+      "step": 11792
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.724101628298709e-05,
+      "loss": 0.0764,
+      "step": 11794
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.724054838105933e-05,
+      "loss": 0.0554,
+      "step": 11796
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.724008047913157e-05,
+      "loss": 0.0658,
+      "step": 11798
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.723961257720382e-05,
+      "loss": 0.0616,
+      "step": 11800
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.7239144675276065e-05,
+      "loss": 0.0725,
+      "step": 11802
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.7238676773348304e-05,
+      "loss": 0.0815,
+      "step": 11804
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.723820887142055e-05,
+      "loss": 0.0898,
+      "step": 11806
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.7237740969492796e-05,
+      "loss": 0.0748,
+      "step": 11808
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.723727306756504e-05,
+      "loss": 0.0658,
+      "step": 11810
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.723680516563728e-05,
+      "loss": 0.0806,
+      "step": 11812
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.723633726370953e-05,
+      "loss": 0.0626,
+      "step": 11814
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.723586936178177e-05,
+      "loss": 0.0744,
+      "step": 11816
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.723540145985402e-05,
+      "loss": 0.0747,
+      "step": 11818
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.723493355792626e-05,
+      "loss": 0.0838,
+      "step": 11820
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.7234465655998504e-05,
+      "loss": 0.0588,
+      "step": 11822
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.723399775407075e-05,
+      "loss": 0.0908,
+      "step": 11824
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.7233529852142996e-05,
+      "loss": 0.1086,
+      "step": 11826
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.7233061950215235e-05,
+      "loss": 0.0839,
+      "step": 11828
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.723259404828748e-05,
+      "loss": 0.0733,
+      "step": 11830
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.723212614635972e-05,
+      "loss": 0.0764,
+      "step": 11832
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.723165824443197e-05,
+      "loss": 0.085,
+      "step": 11834
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.723119034250421e-05,
+      "loss": 0.1012,
+      "step": 11836
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.723072244057646e-05,
+      "loss": 0.0772,
+      "step": 11838
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.72302545386487e-05,
+      "loss": 0.0957,
+      "step": 11840
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.722978663672095e-05,
+      "loss": 0.0894,
+      "step": 11842
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.722931873479319e-05,
+      "loss": 0.0849,
+      "step": 11844
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.7228850832865435e-05,
+      "loss": 0.0983,
+      "step": 11846
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.7228382930937674e-05,
+      "loss": 0.0646,
+      "step": 11848
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.722791502900992e-05,
+      "loss": 0.0889,
+      "step": 11850
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.7227447127082165e-05,
+      "loss": 0.0802,
+      "step": 11852
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.722697922515441e-05,
+      "loss": 0.0982,
+      "step": 11854
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.722651132322665e-05,
+      "loss": 0.0758,
+      "step": 11856
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.7226043421298896e-05,
+      "loss": 0.0803,
+      "step": 11858
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.722557551937114e-05,
+      "loss": 0.0862,
+      "step": 11860
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.722510761744339e-05,
+      "loss": 0.0828,
+      "step": 11862
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.722463971551563e-05,
+      "loss": 0.0986,
+      "step": 11864
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.722417181358787e-05,
+      "loss": 0.0898,
+      "step": 11866
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.722370391166012e-05,
+      "loss": 0.0875,
+      "step": 11868
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.7223236009732365e-05,
+      "loss": 0.0923,
+      "step": 11870
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.7222768107804604e-05,
+      "loss": 0.0614,
+      "step": 11872
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.722230020587685e-05,
+      "loss": 0.0825,
+      "step": 11874
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.7221832303949096e-05,
+      "loss": 0.0857,
+      "step": 11876
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.722136440202134e-05,
+      "loss": 0.0788,
+      "step": 11878
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.722089650009358e-05,
+      "loss": 0.0658,
+      "step": 11880
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.722042859816583e-05,
+      "loss": 0.0642,
+      "step": 11882
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.7219960696238066e-05,
+      "loss": 0.0727,
+      "step": 11884
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.721949279431032e-05,
+      "loss": 0.071,
+      "step": 11886
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.721902489238256e-05,
+      "loss": 0.1086,
+      "step": 11888
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.7218556990454804e-05,
+      "loss": 0.0937,
+      "step": 11890
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.721808908852704e-05,
+      "loss": 0.0744,
+      "step": 11892
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.7217621186599296e-05,
+      "loss": 0.0981,
+      "step": 11894
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.7217153284671535e-05,
+      "loss": 0.0722,
+      "step": 11896
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.721668538274378e-05,
+      "loss": 0.091,
+      "step": 11898
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.721621748081602e-05,
+      "loss": 0.0834,
+      "step": 11900
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.7215749578888266e-05,
+      "loss": 0.0942,
+      "step": 11902
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.721528167696051e-05,
+      "loss": 0.0841,
+      "step": 11904
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.721481377503276e-05,
+      "loss": 0.0716,
+      "step": 11906
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.7214345873105e-05,
+      "loss": 0.0618,
+      "step": 11908
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.721387797117724e-05,
+      "loss": 0.0806,
+      "step": 11910
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.721341006924949e-05,
+      "loss": 0.0829,
+      "step": 11912
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.7212942167321735e-05,
+      "loss": 0.0685,
+      "step": 11914
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.7212474265393974e-05,
+      "loss": 0.067,
+      "step": 11916
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.721200636346622e-05,
+      "loss": 0.0713,
+      "step": 11918
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.7211538461538465e-05,
+      "loss": 0.0798,
+      "step": 11920
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.721107055961071e-05,
+      "loss": 0.0572,
+      "step": 11922
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.721060265768295e-05,
+      "loss": 0.0808,
+      "step": 11924
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.7210134755755196e-05,
+      "loss": 0.076,
+      "step": 11926
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.7209666853827436e-05,
+      "loss": 0.0811,
+      "step": 11928
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.720919895189969e-05,
+      "loss": 0.0944,
+      "step": 11930
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.720873104997193e-05,
+      "loss": 0.066,
+      "step": 11932
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.720826314804417e-05,
+      "loss": 0.0877,
+      "step": 11934
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.720779524611641e-05,
+      "loss": 0.0935,
+      "step": 11936
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7207327344188665e-05,
+      "loss": 0.078,
+      "step": 11938
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7206859442260904e-05,
+      "loss": 0.081,
+      "step": 11940
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.720639154033315e-05,
+      "loss": 0.0885,
+      "step": 11942
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.720592363840539e-05,
+      "loss": 0.0958,
+      "step": 11944
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7205455736477635e-05,
+      "loss": 0.0868,
+      "step": 11946
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.720498783454988e-05,
+      "loss": 0.0872,
+      "step": 11948
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.720451993262213e-05,
+      "loss": 0.0728,
+      "step": 11950
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7204052030694366e-05,
+      "loss": 0.0711,
+      "step": 11952
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.720358412876661e-05,
+      "loss": 0.0688,
+      "step": 11954
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.720311622683886e-05,
+      "loss": 0.0639,
+      "step": 11956
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7202648324911104e-05,
+      "loss": 0.0766,
+      "step": 11958
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.720218042298334e-05,
+      "loss": 0.0897,
+      "step": 11960
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.720171252105559e-05,
+      "loss": 0.1089,
+      "step": 11962
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7201244619127835e-05,
+      "loss": 0.0947,
+      "step": 11964
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.720077671720008e-05,
+      "loss": 0.0831,
+      "step": 11966
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.720030881527232e-05,
+      "loss": 0.0735,
+      "step": 11968
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7199840913344566e-05,
+      "loss": 0.1122,
+      "step": 11970
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.719937301141681e-05,
+      "loss": 0.0753,
+      "step": 11972
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.719890510948906e-05,
+      "loss": 0.0854,
+      "step": 11974
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.71984372075613e-05,
+      "loss": 0.0752,
+      "step": 11976
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.719796930563354e-05,
+      "loss": 0.0778,
+      "step": 11978
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.719750140370578e-05,
+      "loss": 0.0834,
+      "step": 11980
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7197033501778035e-05,
+      "loss": 0.0861,
+      "step": 11982
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7196565599850274e-05,
+      "loss": 0.0943,
+      "step": 11984
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.719609769792252e-05,
+      "loss": 0.0579,
+      "step": 11986
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.719562979599476e-05,
+      "loss": 0.0936,
+      "step": 11988
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.719516189406701e-05,
+      "loss": 0.0587,
+      "step": 11990
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.719469399213925e-05,
+      "loss": 0.085,
+      "step": 11992
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7194226090211496e-05,
+      "loss": 0.0986,
+      "step": 11994
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7193758188283736e-05,
+      "loss": 0.0844,
+      "step": 11996
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.719329028635598e-05,
+      "loss": 0.0631,
+      "step": 11998
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.719282238442823e-05,
+      "loss": 0.0657,
+      "step": 12000
+    },
+    {
+      "epoch": 1.68,
+      "eval_gen_len": 28.4483,
+      "eval_loss": 1.0475629568099976,
+      "eval_meteor": 0.0375,
+      "eval_runtime": 14.051,
+      "eval_samples_per_second": 4.128,
+      "eval_steps_per_second": 0.569,
+      "step": 12000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.719235448250047e-05,
+      "loss": 0.0606,
+      "step": 12002
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.719188658057271e-05,
+      "loss": 0.0709,
+      "step": 12004
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.719141867864496e-05,
+      "loss": 0.071,
+      "step": 12006
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7190950776717204e-05,
+      "loss": 0.0771,
+      "step": 12008
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.719048287478945e-05,
+      "loss": 0.0841,
+      "step": 12010
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.719001497286169e-05,
+      "loss": 0.0795,
+      "step": 12012
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7189547070933935e-05,
+      "loss": 0.0765,
+      "step": 12014
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.718907916900618e-05,
+      "loss": 0.0778,
+      "step": 12016
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.718861126707843e-05,
+      "loss": 0.0827,
+      "step": 12018
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7188143365150666e-05,
+      "loss": 0.0841,
+      "step": 12020
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.718767546322291e-05,
+      "loss": 0.0687,
+      "step": 12022
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.718720756129516e-05,
+      "loss": 0.0862,
+      "step": 12024
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7186739659367404e-05,
+      "loss": 0.0736,
+      "step": 12026
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.718627175743964e-05,
+      "loss": 0.0894,
+      "step": 12028
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.718580385551188e-05,
+      "loss": 0.0997,
+      "step": 12030
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.718533595358413e-05,
+      "loss": 0.0545,
+      "step": 12032
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7184868051656374e-05,
+      "loss": 0.0907,
+      "step": 12034
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.718440014972862e-05,
+      "loss": 0.098,
+      "step": 12036
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.718393224780086e-05,
+      "loss": 0.0891,
+      "step": 12038
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7183464345873105e-05,
+      "loss": 0.0955,
+      "step": 12040
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.718299644394535e-05,
+      "loss": 0.0897,
+      "step": 12042
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.71825285420176e-05,
+      "loss": 0.0927,
+      "step": 12044
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7182060640089836e-05,
+      "loss": 0.0754,
+      "step": 12046
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.718159273816208e-05,
+      "loss": 0.0868,
+      "step": 12048
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.718112483623433e-05,
+      "loss": 0.0834,
+      "step": 12050
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7180656934306574e-05,
+      "loss": 0.0663,
+      "step": 12052
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.718018903237881e-05,
+      "loss": 0.0725,
+      "step": 12054
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.717972113045106e-05,
+      "loss": 0.0634,
+      "step": 12056
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7179253228523305e-05,
+      "loss": 0.0943,
+      "step": 12058
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.717878532659555e-05,
+      "loss": 0.0955,
+      "step": 12060
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.717831742466779e-05,
+      "loss": 0.0678,
+      "step": 12062
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7177849522740036e-05,
+      "loss": 0.0826,
+      "step": 12064
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7177381620812275e-05,
+      "loss": 0.0751,
+      "step": 12066
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.717691371888453e-05,
+      "loss": 0.0981,
+      "step": 12068
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7176445816956767e-05,
+      "loss": 0.0855,
+      "step": 12070
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.717597791502901e-05,
+      "loss": 0.0867,
+      "step": 12072
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.717551001310125e-05,
+      "loss": 0.094,
+      "step": 12074
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.71750421111735e-05,
+      "loss": 0.0927,
+      "step": 12076
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7174574209245743e-05,
+      "loss": 0.086,
+      "step": 12078
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.717410630731799e-05,
+      "loss": 0.0841,
+      "step": 12080
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.717363840539023e-05,
+      "loss": 0.0949,
+      "step": 12082
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7173170503462474e-05,
+      "loss": 0.0747,
+      "step": 12084
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.717270260153472e-05,
+      "loss": 0.0669,
+      "step": 12086
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7172234699606966e-05,
+      "loss": 0.0851,
+      "step": 12088
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7171766797679205e-05,
+      "loss": 0.072,
+      "step": 12090
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.717129889575145e-05,
+      "loss": 0.0783,
+      "step": 12092
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.71708309938237e-05,
+      "loss": 0.0915,
+      "step": 12094
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.717036309189594e-05,
+      "loss": 0.0717,
+      "step": 12096
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.716989518996818e-05,
+      "loss": 0.1175,
+      "step": 12098
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.716942728804043e-05,
+      "loss": 0.101,
+      "step": 12100
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7168959386112674e-05,
+      "loss": 0.0834,
+      "step": 12102
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.716849148418492e-05,
+      "loss": 0.0789,
+      "step": 12104
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.716802358225716e-05,
+      "loss": 0.0702,
+      "step": 12106
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7167555680329405e-05,
+      "loss": 0.1083,
+      "step": 12108
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7167087778401644e-05,
+      "loss": 0.086,
+      "step": 12110
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.71666198764739e-05,
+      "loss": 0.0903,
+      "step": 12112
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7166151974546136e-05,
+      "loss": 0.0627,
+      "step": 12114
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.716568407261838e-05,
+      "loss": 0.0804,
+      "step": 12116
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.716521617069062e-05,
+      "loss": 0.0791,
+      "step": 12118
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7164748268762874e-05,
+      "loss": 0.0845,
+      "step": 12120
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.716428036683511e-05,
+      "loss": 0.0908,
+      "step": 12122
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.716381246490736e-05,
+      "loss": 0.0899,
+      "step": 12124
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.71633445629796e-05,
+      "loss": 0.0888,
+      "step": 12126
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7162876661051844e-05,
+      "loss": 0.0692,
+      "step": 12128
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.716240875912409e-05,
+      "loss": 0.0769,
+      "step": 12130
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7161940857196336e-05,
+      "loss": 0.1018,
+      "step": 12132
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7161472955268575e-05,
+      "loss": 0.0935,
+      "step": 12134
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.716100505334082e-05,
+      "loss": 0.0946,
+      "step": 12136
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7160537151413067e-05,
+      "loss": 0.0889,
+      "step": 12138
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.716006924948531e-05,
+      "loss": 0.107,
+      "step": 12140
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.715960134755755e-05,
+      "loss": 0.0577,
+      "step": 12142
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.71591334456298e-05,
+      "loss": 0.0901,
+      "step": 12144
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7158665543702043e-05,
+      "loss": 0.0711,
+      "step": 12146
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.715819764177429e-05,
+      "loss": 0.0928,
+      "step": 12148
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.715772973984653e-05,
+      "loss": 0.0766,
+      "step": 12150
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.7157261837918774e-05,
+      "loss": 0.0681,
+      "step": 12152
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.715679393599102e-05,
+      "loss": 0.0737,
+      "step": 12154
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.7156326034063266e-05,
+      "loss": 0.095,
+      "step": 12156
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.7155858132135505e-05,
+      "loss": 0.0668,
+      "step": 12158
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.715539023020775e-05,
+      "loss": 0.0609,
+      "step": 12160
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.715492232827999e-05,
+      "loss": 0.0812,
+      "step": 12162
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.715445442635224e-05,
+      "loss": 0.0752,
+      "step": 12164
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.715398652442448e-05,
+      "loss": 0.0787,
+      "step": 12166
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.715351862249673e-05,
+      "loss": 0.0929,
+      "step": 12168
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.715305072056897e-05,
+      "loss": 0.0748,
+      "step": 12170
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.715258281864122e-05,
+      "loss": 0.0749,
+      "step": 12172
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.715211491671346e-05,
+      "loss": 0.0839,
+      "step": 12174
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.7151647014785705e-05,
+      "loss": 0.0936,
+      "step": 12176
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.7151179112857944e-05,
+      "loss": 0.0704,
+      "step": 12178
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.715071121093019e-05,
+      "loss": 0.0778,
+      "step": 12180
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.7150243309002436e-05,
+      "loss": 0.109,
+      "step": 12182
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.714977540707468e-05,
+      "loss": 0.0578,
+      "step": 12184
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.714930750514692e-05,
+      "loss": 0.0732,
+      "step": 12186
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.714883960321917e-05,
+      "loss": 0.0868,
+      "step": 12188
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.714837170129141e-05,
+      "loss": 0.0722,
+      "step": 12190
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.714790379936366e-05,
+      "loss": 0.0947,
+      "step": 12192
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.71474358974359e-05,
+      "loss": 0.0965,
+      "step": 12194
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.7146967995508144e-05,
+      "loss": 0.0617,
+      "step": 12196
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.714650009358039e-05,
+      "loss": 0.0687,
+      "step": 12198
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.7146032191652636e-05,
+      "loss": 0.0615,
+      "step": 12200
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.7145564289724875e-05,
+      "loss": 0.0919,
+      "step": 12202
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.714509638779712e-05,
+      "loss": 0.0799,
+      "step": 12204
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.714462848586936e-05,
+      "loss": 0.0799,
+      "step": 12206
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.714416058394161e-05,
+      "loss": 0.1178,
+      "step": 12208
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.714369268201385e-05,
+      "loss": 0.0729,
+      "step": 12210
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.71432247800861e-05,
+      "loss": 0.0889,
+      "step": 12212
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.714275687815834e-05,
+      "loss": 0.0741,
+      "step": 12214
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.714228897623059e-05,
+      "loss": 0.0819,
+      "step": 12216
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.714182107430283e-05,
+      "loss": 0.0892,
+      "step": 12218
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7141353172375074e-05,
+      "loss": 0.0782,
+      "step": 12220
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7140885270447314e-05,
+      "loss": 0.1023,
+      "step": 12222
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.714041736851956e-05,
+      "loss": 0.0975,
+      "step": 12224
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7139949466591805e-05,
+      "loss": 0.0752,
+      "step": 12226
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.713948156466405e-05,
+      "loss": 0.0733,
+      "step": 12228
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.713901366273629e-05,
+      "loss": 0.0735,
+      "step": 12230
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7138545760808536e-05,
+      "loss": 0.0606,
+      "step": 12232
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.713807785888078e-05,
+      "loss": 0.0787,
+      "step": 12234
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.713760995695303e-05,
+      "loss": 0.0747,
+      "step": 12236
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.713714205502527e-05,
+      "loss": 0.077,
+      "step": 12238
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.713667415309751e-05,
+      "loss": 0.0584,
+      "step": 12240
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.713620625116976e-05,
+      "loss": 0.082,
+      "step": 12242
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7135738349242005e-05,
+      "loss": 0.0814,
+      "step": 12244
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7135270447314244e-05,
+      "loss": 0.0826,
+      "step": 12246
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.713480254538649e-05,
+      "loss": 0.0768,
+      "step": 12248
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7134334643458736e-05,
+      "loss": 0.0752,
+      "step": 12250
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.713386674153098e-05,
+      "loss": 0.0939,
+      "step": 12252
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.713339883960322e-05,
+      "loss": 0.0845,
+      "step": 12254
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.713293093767547e-05,
+      "loss": 0.1071,
+      "step": 12256
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7132463035747706e-05,
+      "loss": 0.0772,
+      "step": 12258
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.713199513381996e-05,
+      "loss": 0.0782,
+      "step": 12260
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.71315272318922e-05,
+      "loss": 0.0823,
+      "step": 12262
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7131059329964444e-05,
+      "loss": 0.0587,
+      "step": 12264
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.713059142803668e-05,
+      "loss": 0.0788,
+      "step": 12266
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7130123526108936e-05,
+      "loss": 0.0772,
+      "step": 12268
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7129655624181175e-05,
+      "loss": 0.0952,
+      "step": 12270
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.712918772225342e-05,
+      "loss": 0.0765,
+      "step": 12272
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.712871982032566e-05,
+      "loss": 0.0812,
+      "step": 12274
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7128251918397906e-05,
+      "loss": 0.088,
+      "step": 12276
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.712778401647015e-05,
+      "loss": 0.0745,
+      "step": 12278
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.71273161145424e-05,
+      "loss": 0.1049,
+      "step": 12280
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.712684821261464e-05,
+      "loss": 0.0813,
+      "step": 12282
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.712638031068688e-05,
+      "loss": 0.1017,
+      "step": 12284
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.712591240875913e-05,
+      "loss": 0.074,
+      "step": 12286
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.712544450683137e-05,
+      "loss": 0.085,
+      "step": 12288
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7124976604903614e-05,
+      "loss": 0.1091,
+      "step": 12290
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.712450870297585e-05,
+      "loss": 0.0696,
+      "step": 12292
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7124040801048105e-05,
+      "loss": 0.056,
+      "step": 12294
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7123572899120345e-05,
+      "loss": 0.0998,
+      "step": 12296
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.712310499719259e-05,
+      "loss": 0.0656,
+      "step": 12298
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.712263709526483e-05,
+      "loss": 0.0775,
+      "step": 12300
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.712216919333708e-05,
+      "loss": 0.0605,
+      "step": 12302
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.712170129140932e-05,
+      "loss": 0.1085,
+      "step": 12304
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.712123338948157e-05,
+      "loss": 0.0705,
+      "step": 12306
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7120765487553806e-05,
+      "loss": 0.0794,
+      "step": 12308
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.712029758562605e-05,
+      "loss": 0.0838,
+      "step": 12310
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.71198296836983e-05,
+      "loss": 0.0954,
+      "step": 12312
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7119361781770544e-05,
+      "loss": 0.0668,
+      "step": 12314
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.711889387984278e-05,
+      "loss": 0.0723,
+      "step": 12316
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.711842597791503e-05,
+      "loss": 0.0811,
+      "step": 12318
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7117958075987275e-05,
+      "loss": 0.0921,
+      "step": 12320
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.711749017405952e-05,
+      "loss": 0.1172,
+      "step": 12322
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.711702227213176e-05,
+      "loss": 0.0791,
+      "step": 12324
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7116554370204006e-05,
+      "loss": 0.0773,
+      "step": 12326
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.711608646827625e-05,
+      "loss": 0.0753,
+      "step": 12328
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.71156185663485e-05,
+      "loss": 0.0813,
+      "step": 12330
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.711515066442074e-05,
+      "loss": 0.0911,
+      "step": 12332
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.711468276249298e-05,
+      "loss": 0.1135,
+      "step": 12334
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.711421486056523e-05,
+      "loss": 0.0854,
+      "step": 12336
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7113746958637475e-05,
+      "loss": 0.0765,
+      "step": 12338
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7113279056709714e-05,
+      "loss": 0.0631,
+      "step": 12340
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.711281115478196e-05,
+      "loss": 0.0707,
+      "step": 12342
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.71123432528542e-05,
+      "loss": 0.0709,
+      "step": 12344
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.711187535092645e-05,
+      "loss": 0.1139,
+      "step": 12346
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.711140744899869e-05,
+      "loss": 0.0931,
+      "step": 12348
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.711093954707094e-05,
+      "loss": 0.0723,
+      "step": 12350
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7110471645143176e-05,
+      "loss": 0.082,
+      "step": 12352
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.711000374321542e-05,
+      "loss": 0.0655,
+      "step": 12354
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.710953584128767e-05,
+      "loss": 0.0797,
+      "step": 12356
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7109067939359914e-05,
+      "loss": 0.0938,
+      "step": 12358
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.710860003743215e-05,
+      "loss": 0.0988,
+      "step": 12360
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.71081321355044e-05,
+      "loss": 0.0772,
+      "step": 12362
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.7107664233576645e-05,
+      "loss": 0.0752,
+      "step": 12364
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.710719633164889e-05,
+      "loss": 0.0749,
+      "step": 12366
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.710672842972113e-05,
+      "loss": 0.087,
+      "step": 12368
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.7106260527793376e-05,
+      "loss": 0.0851,
+      "step": 12370
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.710579262586562e-05,
+      "loss": 0.1129,
+      "step": 12372
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.710532472393787e-05,
+      "loss": 0.0716,
+      "step": 12374
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.7104856822010106e-05,
+      "loss": 0.0682,
+      "step": 12376
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.710438892008235e-05,
+      "loss": 0.0663,
+      "step": 12378
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.71039210181546e-05,
+      "loss": 0.0867,
+      "step": 12380
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.7103453116226844e-05,
+      "loss": 0.0715,
+      "step": 12382
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.710298521429908e-05,
+      "loss": 0.0789,
+      "step": 12384
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.710251731237133e-05,
+      "loss": 0.0809,
+      "step": 12386
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.710204941044357e-05,
+      "loss": 0.0786,
+      "step": 12388
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.710158150851582e-05,
+      "loss": 0.0854,
+      "step": 12390
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.710111360658806e-05,
+      "loss": 0.1066,
+      "step": 12392
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.7100645704660306e-05,
+      "loss": 0.0754,
+      "step": 12394
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.7100177802732545e-05,
+      "loss": 0.0738,
+      "step": 12396
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.70997099008048e-05,
+      "loss": 0.0855,
+      "step": 12398
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.709924199887704e-05,
+      "loss": 0.0652,
+      "step": 12400
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.709877409694928e-05,
+      "loss": 0.0815,
+      "step": 12402
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.709830619502152e-05,
+      "loss": 0.0983,
+      "step": 12404
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.709783829309377e-05,
+      "loss": 0.0887,
+      "step": 12406
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.7097370391166014e-05,
+      "loss": 0.0949,
+      "step": 12408
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.709690248923826e-05,
+      "loss": 0.0763,
+      "step": 12410
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.70964345873105e-05,
+      "loss": 0.0612,
+      "step": 12412
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.7095966685382745e-05,
+      "loss": 0.0726,
+      "step": 12414
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.709549878345499e-05,
+      "loss": 0.0978,
+      "step": 12416
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.709503088152724e-05,
+      "loss": 0.0781,
+      "step": 12418
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.7094562979599476e-05,
+      "loss": 0.1019,
+      "step": 12420
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.709409507767172e-05,
+      "loss": 0.074,
+      "step": 12422
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.709362717574397e-05,
+      "loss": 0.0949,
+      "step": 12424
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.7093159273816214e-05,
+      "loss": 0.0798,
+      "step": 12426
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.709269137188845e-05,
+      "loss": 0.0883,
+      "step": 12428
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.70922234699607e-05,
+      "loss": 0.0895,
+      "step": 12430
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.7091755568032945e-05,
+      "loss": 0.0651,
+      "step": 12432
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.709128766610519e-05,
+      "loss": 0.0896,
+      "step": 12434
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.709081976417743e-05,
+      "loss": 0.0778,
+      "step": 12436
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.7090351862249676e-05,
+      "loss": 0.0979,
+      "step": 12438
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.7089883960321915e-05,
+      "loss": 0.0776,
+      "step": 12440
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.708941605839417e-05,
+      "loss": 0.0702,
+      "step": 12442
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.7088948156466407e-05,
+      "loss": 0.0927,
+      "step": 12444
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.708848025453865e-05,
+      "loss": 0.1335,
+      "step": 12446
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.708801235261089e-05,
+      "loss": 0.0757,
+      "step": 12448
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.7087544450683144e-05,
+      "loss": 0.0841,
+      "step": 12450
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.708707654875538e-05,
+      "loss": 0.1079,
+      "step": 12452
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.708660864682763e-05,
+      "loss": 0.0901,
+      "step": 12454
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.708614074489987e-05,
+      "loss": 0.0774,
+      "step": 12456
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.7085672842972114e-05,
+      "loss": 0.0933,
+      "step": 12458
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.708520494104436e-05,
+      "loss": 0.0999,
+      "step": 12460
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.7084737039116606e-05,
+      "loss": 0.0647,
+      "step": 12462
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.7084269137188845e-05,
+      "loss": 0.1042,
+      "step": 12464
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.708380123526109e-05,
+      "loss": 0.0931,
+      "step": 12466
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.708333333333334e-05,
+      "loss": 0.087,
+      "step": 12468
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.708286543140558e-05,
+      "loss": 0.0654,
+      "step": 12470
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.708239752947782e-05,
+      "loss": 0.1034,
+      "step": 12472
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.708192962755007e-05,
+      "loss": 0.0683,
+      "step": 12474
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.7081461725622314e-05,
+      "loss": 0.0623,
+      "step": 12476
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.708099382369456e-05,
+      "loss": 0.0858,
+      "step": 12478
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.70805259217668e-05,
+      "loss": 0.0856,
+      "step": 12480
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.7080058019839045e-05,
+      "loss": 0.0915,
+      "step": 12482
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.707959011791129e-05,
+      "loss": 0.0699,
+      "step": 12484
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.707912221598354e-05,
+      "loss": 0.0971,
+      "step": 12486
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.7078654314055776e-05,
+      "loss": 0.0924,
+      "step": 12488
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.707818641212802e-05,
+      "loss": 0.079,
+      "step": 12490
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.707771851020026e-05,
+      "loss": 0.1105,
+      "step": 12492
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.7077250608272514e-05,
+      "loss": 0.0796,
+      "step": 12494
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.707678270634475e-05,
+      "loss": 0.0816,
+      "step": 12496
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.7076314804417e-05,
+      "loss": 0.0782,
+      "step": 12498
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.707584690248924e-05,
+      "loss": 0.0678,
+      "step": 12500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.7075379000561484e-05,
+      "loss": 0.0701,
+      "step": 12502
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.707491109863373e-05,
+      "loss": 0.0743,
+      "step": 12504
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.7074443196705976e-05,
+      "loss": 0.0765,
+      "step": 12506
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.7073975294778215e-05,
+      "loss": 0.0664,
+      "step": 12508
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.707350739285046e-05,
+      "loss": 0.0884,
+      "step": 12510
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.7073039490922707e-05,
+      "loss": 0.0774,
+      "step": 12512
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.707257158899495e-05,
+      "loss": 0.0648,
+      "step": 12514
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.707210368706719e-05,
+      "loss": 0.0909,
+      "step": 12516
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.707163578513944e-05,
+      "loss": 0.1022,
+      "step": 12518
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.7071167883211683e-05,
+      "loss": 0.0714,
+      "step": 12520
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.707069998128393e-05,
+      "loss": 0.0633,
+      "step": 12522
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.707023207935617e-05,
+      "loss": 0.0935,
+      "step": 12524
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.7069764177428414e-05,
+      "loss": 0.0655,
+      "step": 12526
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.706929627550066e-05,
+      "loss": 0.0806,
+      "step": 12528
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.7068828373572906e-05,
+      "loss": 0.0893,
+      "step": 12530
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.7068360471645145e-05,
+      "loss": 0.079,
+      "step": 12532
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.706789256971739e-05,
+      "loss": 0.0798,
+      "step": 12534
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.706742466778963e-05,
+      "loss": 0.0623,
+      "step": 12536
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.7066956765861876e-05,
+      "loss": 0.0825,
+      "step": 12538
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.706648886393412e-05,
+      "loss": 0.0769,
+      "step": 12540
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.706602096200636e-05,
+      "loss": 0.0712,
+      "step": 12542
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.706555306007861e-05,
+      "loss": 0.0721,
+      "step": 12544
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.706508515815085e-05,
+      "loss": 0.0735,
+      "step": 12546
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.70646172562231e-05,
+      "loss": 0.0798,
+      "step": 12548
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.706414935429534e-05,
+      "loss": 0.11,
+      "step": 12550
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.7063681452367584e-05,
+      "loss": 0.085,
+      "step": 12552
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.706321355043983e-05,
+      "loss": 0.0641,
+      "step": 12554
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.7062745648512076e-05,
+      "loss": 0.0903,
+      "step": 12556
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.7062277746584315e-05,
+      "loss": 0.0814,
+      "step": 12558
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.706180984465656e-05,
+      "loss": 0.075,
+      "step": 12560
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.706134194272881e-05,
+      "loss": 0.0837,
+      "step": 12562
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.706087404080105e-05,
+      "loss": 0.0807,
+      "step": 12564
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.706040613887329e-05,
+      "loss": 0.102,
+      "step": 12566
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.705993823694554e-05,
+      "loss": 0.0906,
+      "step": 12568
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.705947033501778e-05,
+      "loss": 0.081,
+      "step": 12570
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.705900243309003e-05,
+      "loss": 0.0921,
+      "step": 12572
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.705853453116227e-05,
+      "loss": 0.093,
+      "step": 12574
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.7058066629234515e-05,
+      "loss": 0.0976,
+      "step": 12576
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.7057598727306754e-05,
+      "loss": 0.0803,
+      "step": 12578
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.7057130825379007e-05,
+      "loss": 0.0962,
+      "step": 12580
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.7056662923451246e-05,
+      "loss": 0.0884,
+      "step": 12582
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.705619502152349e-05,
+      "loss": 0.0813,
+      "step": 12584
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.705572711959573e-05,
+      "loss": 0.07,
+      "step": 12586
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.705525921766798e-05,
+      "loss": 0.0996,
+      "step": 12588
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.705479131574022e-05,
+      "loss": 0.0847,
+      "step": 12590
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.705432341381247e-05,
+      "loss": 0.0983,
+      "step": 12592
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.705385551188471e-05,
+      "loss": 0.0751,
+      "step": 12594
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.7053387609956954e-05,
+      "loss": 0.081,
+      "step": 12596
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.70529197080292e-05,
+      "loss": 0.0744,
+      "step": 12598
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.7052451806101445e-05,
+      "loss": 0.0707,
+      "step": 12600
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.7051983904173684e-05,
+      "loss": 0.0723,
+      "step": 12602
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.705151600224593e-05,
+      "loss": 0.0709,
+      "step": 12604
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.7051048100318176e-05,
+      "loss": 0.0801,
+      "step": 12606
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.705058019839042e-05,
+      "loss": 0.0836,
+      "step": 12608
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.705011229646266e-05,
+      "loss": 0.0852,
+      "step": 12610
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.704964439453491e-05,
+      "loss": 0.0622,
+      "step": 12612
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.704917649260715e-05,
+      "loss": 0.0797,
+      "step": 12614
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.70487085906794e-05,
+      "loss": 0.084,
+      "step": 12616
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.704824068875164e-05,
+      "loss": 0.0826,
+      "step": 12618
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.7047772786823884e-05,
+      "loss": 0.1171,
+      "step": 12620
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.704730488489612e-05,
+      "loss": 0.0766,
+      "step": 12622
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.7046836982968376e-05,
+      "loss": 0.0831,
+      "step": 12624
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.7046369081040615e-05,
+      "loss": 0.098,
+      "step": 12626
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.704590117911286e-05,
+      "loss": 0.0922,
+      "step": 12628
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.70454332771851e-05,
+      "loss": 0.0845,
+      "step": 12630
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.7044965375257346e-05,
+      "loss": 0.0798,
+      "step": 12632
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.704449747332959e-05,
+      "loss": 0.0753,
+      "step": 12634
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.704402957140184e-05,
+      "loss": 0.0997,
+      "step": 12636
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.704356166947408e-05,
+      "loss": 0.0875,
+      "step": 12638
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.704309376754632e-05,
+      "loss": 0.0841,
+      "step": 12640
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.704262586561857e-05,
+      "loss": 0.0661,
+      "step": 12642
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.7042157963690815e-05,
+      "loss": 0.1243,
+      "step": 12644
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.7041690061763054e-05,
+      "loss": 0.0647,
+      "step": 12646
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.70412221598353e-05,
+      "loss": 0.0703,
+      "step": 12648
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.7040754257907546e-05,
+      "loss": 0.0813,
+      "step": 12650
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.704028635597979e-05,
+      "loss": 0.0737,
+      "step": 12652
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.703981845405203e-05,
+      "loss": 0.0834,
+      "step": 12654
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.703935055212428e-05,
+      "loss": 0.0771,
+      "step": 12656
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.703888265019652e-05,
+      "loss": 0.0748,
+      "step": 12658
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.703841474826877e-05,
+      "loss": 0.0708,
+      "step": 12660
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.703794684634101e-05,
+      "loss": 0.0635,
+      "step": 12662
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.7037478944413254e-05,
+      "loss": 0.0806,
+      "step": 12664
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.703701104248549e-05,
+      "loss": 0.0738,
+      "step": 12666
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.7036543140557745e-05,
+      "loss": 0.0633,
+      "step": 12668
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.7036075238629984e-05,
+      "loss": 0.1057,
+      "step": 12670
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.703560733670223e-05,
+      "loss": 0.0974,
+      "step": 12672
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.703513943477447e-05,
+      "loss": 0.0805,
+      "step": 12674
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.703467153284672e-05,
+      "loss": 0.1314,
+      "step": 12676
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.703420363091896e-05,
+      "loss": 0.068,
+      "step": 12678
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.703373572899121e-05,
+      "loss": 0.1,
+      "step": 12680
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.7033267827063446e-05,
+      "loss": 0.1086,
+      "step": 12682
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.703279992513569e-05,
+      "loss": 0.0704,
+      "step": 12684
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.703233202320794e-05,
+      "loss": 0.0832,
+      "step": 12686
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.7031864121280184e-05,
+      "loss": 0.0761,
+      "step": 12688
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.703139621935242e-05,
+      "loss": 0.0811,
+      "step": 12690
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.703092831742467e-05,
+      "loss": 0.0842,
+      "step": 12692
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.7030460415496915e-05,
+      "loss": 0.0915,
+      "step": 12694
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.702999251356916e-05,
+      "loss": 0.0922,
+      "step": 12696
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.70295246116414e-05,
+      "loss": 0.0962,
+      "step": 12698
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.7029056709713646e-05,
+      "loss": 0.0808,
+      "step": 12700
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.702858880778589e-05,
+      "loss": 0.0769,
+      "step": 12702
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.702812090585814e-05,
+      "loss": 0.0768,
+      "step": 12704
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.702765300393038e-05,
+      "loss": 0.1067,
+      "step": 12706
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.702718510200262e-05,
+      "loss": 0.0711,
+      "step": 12708
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.702671720007487e-05,
+      "loss": 0.0673,
+      "step": 12710
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.7026249298147115e-05,
+      "loss": 0.087,
+      "step": 12712
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.7025781396219354e-05,
+      "loss": 0.0766,
+      "step": 12714
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.70253134942916e-05,
+      "loss": 0.079,
+      "step": 12716
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.702484559236384e-05,
+      "loss": 0.0754,
+      "step": 12718
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.702437769043609e-05,
+      "loss": 0.073,
+      "step": 12720
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.702390978850833e-05,
+      "loss": 0.066,
+      "step": 12722
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.702344188658058e-05,
+      "loss": 0.0853,
+      "step": 12724
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.7022973984652816e-05,
+      "loss": 0.0668,
+      "step": 12726
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.702250608272507e-05,
+      "loss": 0.0669,
+      "step": 12728
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.702203818079731e-05,
+      "loss": 0.0992,
+      "step": 12730
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.7021570278869554e-05,
+      "loss": 0.0826,
+      "step": 12732
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.702110237694179e-05,
+      "loss": 0.0781,
+      "step": 12734
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.702063447501404e-05,
+      "loss": 0.0865,
+      "step": 12736
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.7020166573086284e-05,
+      "loss": 0.0927,
+      "step": 12738
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.701969867115853e-05,
+      "loss": 0.0905,
+      "step": 12740
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.701923076923077e-05,
+      "loss": 0.0795,
+      "step": 12742
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.7018762867303015e-05,
+      "loss": 0.0623,
+      "step": 12744
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.701829496537526e-05,
+      "loss": 0.0874,
+      "step": 12746
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.701782706344751e-05,
+      "loss": 0.0702,
+      "step": 12748
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.7017359161519746e-05,
+      "loss": 0.0783,
+      "step": 12750
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.701689125959199e-05,
+      "loss": 0.0527,
+      "step": 12752
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.701642335766424e-05,
+      "loss": 0.0763,
+      "step": 12754
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.7015955455736484e-05,
+      "loss": 0.0902,
+      "step": 12756
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.701548755380872e-05,
+      "loss": 0.0715,
+      "step": 12758
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.701501965188097e-05,
+      "loss": 0.0658,
+      "step": 12760
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.7014551749953215e-05,
+      "loss": 0.0704,
+      "step": 12762
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.701408384802546e-05,
+      "loss": 0.0947,
+      "step": 12764
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.70136159460977e-05,
+      "loss": 0.0937,
+      "step": 12766
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.7013148044169946e-05,
+      "loss": 0.0731,
+      "step": 12768
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.7012680142242185e-05,
+      "loss": 0.0846,
+      "step": 12770
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.701221224031444e-05,
+      "loss": 0.0867,
+      "step": 12772
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.701174433838668e-05,
+      "loss": 0.0654,
+      "step": 12774
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.701127643645892e-05,
+      "loss": 0.086,
+      "step": 12776
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.701080853453116e-05,
+      "loss": 0.0658,
+      "step": 12778
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.701034063260341e-05,
+      "loss": 0.0805,
+      "step": 12780
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.7009872730675654e-05,
+      "loss": 0.0665,
+      "step": 12782
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.70094048287479e-05,
+      "loss": 0.0894,
+      "step": 12784
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.700893692682014e-05,
+      "loss": 0.0769,
+      "step": 12786
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.7008469024892385e-05,
+      "loss": 0.0898,
+      "step": 12788
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.700800112296463e-05,
+      "loss": 0.0977,
+      "step": 12790
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.700753322103687e-05,
+      "loss": 0.0745,
+      "step": 12792
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.7007065319109116e-05,
+      "loss": 0.0907,
+      "step": 12794
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.700659741718136e-05,
+      "loss": 0.0701,
+      "step": 12796
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.700612951525361e-05,
+      "loss": 0.0849,
+      "step": 12798
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.700566161332585e-05,
+      "loss": 0.1005,
+      "step": 12800
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.700519371139809e-05,
+      "loss": 0.0653,
+      "step": 12802
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.700472580947033e-05,
+      "loss": 0.0792,
+      "step": 12804
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.7004257907542585e-05,
+      "loss": 0.1003,
+      "step": 12806
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.7003790005614824e-05,
+      "loss": 0.1019,
+      "step": 12808
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.700332210368707e-05,
+      "loss": 0.1073,
+      "step": 12810
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.700285420175931e-05,
+      "loss": 0.0973,
+      "step": 12812
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.7002386299831555e-05,
+      "loss": 0.1095,
+      "step": 12814
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.70019183979038e-05,
+      "loss": 0.0908,
+      "step": 12816
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.7001450495976046e-05,
+      "loss": 0.1097,
+      "step": 12818
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.7000982594048286e-05,
+      "loss": 0.088,
+      "step": 12820
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.700051469212053e-05,
+      "loss": 0.0861,
+      "step": 12822
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.700004679019278e-05,
+      "loss": 0.0737,
+      "step": 12824
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.699957888826502e-05,
+      "loss": 0.082,
+      "step": 12826
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.699911098633726e-05,
+      "loss": 0.0848,
+      "step": 12828
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.699864308440951e-05,
+      "loss": 0.1033,
+      "step": 12830
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.6998175182481754e-05,
+      "loss": 0.091,
+      "step": 12832
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.6997707280554e-05,
+      "loss": 0.0813,
+      "step": 12834
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.699723937862624e-05,
+      "loss": 0.0833,
+      "step": 12836
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.6996771476698485e-05,
+      "loss": 0.0817,
+      "step": 12838
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.699630357477073e-05,
+      "loss": 0.07,
+      "step": 12840
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.699583567284298e-05,
+      "loss": 0.0913,
+      "step": 12842
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.6995367770915216e-05,
+      "loss": 0.0844,
+      "step": 12844
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.699489986898746e-05,
+      "loss": 0.0987,
+      "step": 12846
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.69944319670597e-05,
+      "loss": 0.0743,
+      "step": 12848
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.6993964065131954e-05,
+      "loss": 0.1001,
+      "step": 12850
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.699349616320419e-05,
+      "loss": 0.0953,
+      "step": 12852
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.699302826127644e-05,
+      "loss": 0.0662,
+      "step": 12854
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.699256035934868e-05,
+      "loss": 0.0551,
+      "step": 12856
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.699209245742093e-05,
+      "loss": 0.101,
+      "step": 12858
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.699162455549317e-05,
+      "loss": 0.083,
+      "step": 12860
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6991156653565416e-05,
+      "loss": 0.077,
+      "step": 12862
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6990688751637655e-05,
+      "loss": 0.0892,
+      "step": 12864
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.69902208497099e-05,
+      "loss": 0.0972,
+      "step": 12866
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.698975294778215e-05,
+      "loss": 0.0699,
+      "step": 12868
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.698928504585439e-05,
+      "loss": 0.0878,
+      "step": 12870
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.698881714392663e-05,
+      "loss": 0.0886,
+      "step": 12872
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.698834924199888e-05,
+      "loss": 0.0696,
+      "step": 12874
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6987881340071124e-05,
+      "loss": 0.0991,
+      "step": 12876
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.698741343814337e-05,
+      "loss": 0.0653,
+      "step": 12878
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.698694553621561e-05,
+      "loss": 0.074,
+      "step": 12880
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6986477634287855e-05,
+      "loss": 0.0856,
+      "step": 12882
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.69860097323601e-05,
+      "loss": 0.0951,
+      "step": 12884
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6985541830432346e-05,
+      "loss": 0.0908,
+      "step": 12886
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6985073928504586e-05,
+      "loss": 0.0951,
+      "step": 12888
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.698460602657683e-05,
+      "loss": 0.0747,
+      "step": 12890
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.698413812464908e-05,
+      "loss": 0.0823,
+      "step": 12892
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.698367022272132e-05,
+      "loss": 0.0784,
+      "step": 12894
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.698320232079356e-05,
+      "loss": 0.0788,
+      "step": 12896
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.698273441886581e-05,
+      "loss": 0.0754,
+      "step": 12898
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.698226651693805e-05,
+      "loss": 0.08,
+      "step": 12900
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.69817986150103e-05,
+      "loss": 0.1191,
+      "step": 12902
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.698133071308254e-05,
+      "loss": 0.0671,
+      "step": 12904
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6980862811154785e-05,
+      "loss": 0.0759,
+      "step": 12906
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6980394909227024e-05,
+      "loss": 0.0832,
+      "step": 12908
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.697992700729927e-05,
+      "loss": 0.0873,
+      "step": 12910
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6979459105371516e-05,
+      "loss": 0.0718,
+      "step": 12912
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.697899120344376e-05,
+      "loss": 0.0744,
+      "step": 12914
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6978523301516e-05,
+      "loss": 0.0865,
+      "step": 12916
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.697805539958825e-05,
+      "loss": 0.0906,
+      "step": 12918
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.697758749766049e-05,
+      "loss": 0.1134,
+      "step": 12920
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.697711959573274e-05,
+      "loss": 0.0586,
+      "step": 12922
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.697665169380498e-05,
+      "loss": 0.0672,
+      "step": 12924
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6976183791877224e-05,
+      "loss": 0.0727,
+      "step": 12926
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.697571588994947e-05,
+      "loss": 0.0823,
+      "step": 12928
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6975247988021716e-05,
+      "loss": 0.0665,
+      "step": 12930
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.6974780086093955e-05,
+      "loss": 0.0841,
+      "step": 12932
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.69743121841662e-05,
+      "loss": 0.073,
+      "step": 12934
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.697384428223845e-05,
+      "loss": 0.0805,
+      "step": 12936
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.697337638031069e-05,
+      "loss": 0.0971,
+      "step": 12938
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.697290847838293e-05,
+      "loss": 0.0782,
+      "step": 12940
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.697244057645518e-05,
+      "loss": 0.0666,
+      "step": 12942
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.697197267452742e-05,
+      "loss": 0.0874,
+      "step": 12944
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.697150477259967e-05,
+      "loss": 0.087,
+      "step": 12946
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.697103687067191e-05,
+      "loss": 0.067,
+      "step": 12948
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.6970568968744155e-05,
+      "loss": 0.0841,
+      "step": 12950
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.6970101066816394e-05,
+      "loss": 0.0915,
+      "step": 12952
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.6969633164888646e-05,
+      "loss": 0.0671,
+      "step": 12954
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.6969165262960886e-05,
+      "loss": 0.0947,
+      "step": 12956
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.696869736103313e-05,
+      "loss": 0.0843,
+      "step": 12958
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.696822945910537e-05,
+      "loss": 0.0736,
+      "step": 12960
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.6967761557177617e-05,
+      "loss": 0.0629,
+      "step": 12962
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.696729365524986e-05,
+      "loss": 0.119,
+      "step": 12964
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.696682575332211e-05,
+      "loss": 0.0911,
+      "step": 12966
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.696635785139435e-05,
+      "loss": 0.0837,
+      "step": 12968
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.6965889949466593e-05,
+      "loss": 0.0801,
+      "step": 12970
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.696542204753884e-05,
+      "loss": 0.07,
+      "step": 12972
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.6964954145611085e-05,
+      "loss": 0.0721,
+      "step": 12974
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.6964486243683324e-05,
+      "loss": 0.0727,
+      "step": 12976
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.696401834175557e-05,
+      "loss": 0.092,
+      "step": 12978
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.6963550439827816e-05,
+      "loss": 0.0906,
+      "step": 12980
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.696308253790006e-05,
+      "loss": 0.0842,
+      "step": 12982
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.69626146359723e-05,
+      "loss": 0.0721,
+      "step": 12984
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.696214673404455e-05,
+      "loss": 0.0861,
+      "step": 12986
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.696167883211679e-05,
+      "loss": 0.0657,
+      "step": 12988
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.696121093018904e-05,
+      "loss": 0.0798,
+      "step": 12990
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.696074302826128e-05,
+      "loss": 0.1028,
+      "step": 12992
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.6960275126333524e-05,
+      "loss": 0.0893,
+      "step": 12994
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.695980722440576e-05,
+      "loss": 0.0804,
+      "step": 12996
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.6959339322478016e-05,
+      "loss": 0.071,
+      "step": 12998
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.6958871420550255e-05,
+      "loss": 0.0995,
+      "step": 13000
+    },
+    {
+      "epoch": 1.82,
+      "eval_gen_len": 29.2069,
+      "eval_loss": 1.0517786741256714,
+      "eval_meteor": 0.0455,
+      "eval_runtime": 14.382,
+      "eval_samples_per_second": 4.033,
+      "eval_steps_per_second": 0.556,
+      "step": 13000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.69584035186225e-05,
+      "loss": 0.0864,
+      "step": 13002
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.695793561669474e-05,
+      "loss": 0.062,
+      "step": 13004
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.695746771476699e-05,
+      "loss": 0.0822,
+      "step": 13006
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.695699981283923e-05,
+      "loss": 0.071,
+      "step": 13008
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.695653191091148e-05,
+      "loss": 0.1013,
+      "step": 13010
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.695606400898372e-05,
+      "loss": 0.0641,
+      "step": 13012
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.695559610705596e-05,
+      "loss": 0.0776,
+      "step": 13014
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.695512820512821e-05,
+      "loss": 0.081,
+      "step": 13016
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.6954660303200455e-05,
+      "loss": 0.0809,
+      "step": 13018
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.6954192401272694e-05,
+      "loss": 0.0897,
+      "step": 13020
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.695372449934494e-05,
+      "loss": 0.0887,
+      "step": 13022
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.6953256597417186e-05,
+      "loss": 0.0946,
+      "step": 13024
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.695278869548943e-05,
+      "loss": 0.0753,
+      "step": 13026
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.695232079356167e-05,
+      "loss": 0.079,
+      "step": 13028
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.6951852891633917e-05,
+      "loss": 0.0981,
+      "step": 13030
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.695138498970616e-05,
+      "loss": 0.1006,
+      "step": 13032
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.695091708777841e-05,
+      "loss": 0.0621,
+      "step": 13034
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.695044918585065e-05,
+      "loss": 0.0791,
+      "step": 13036
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.6949981283922893e-05,
+      "loss": 0.1089,
+      "step": 13038
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.694951338199514e-05,
+      "loss": 0.0858,
+      "step": 13040
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.694904548006738e-05,
+      "loss": 0.0971,
+      "step": 13042
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.6948577578139624e-05,
+      "loss": 0.0803,
+      "step": 13044
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.6948109676211864e-05,
+      "loss": 0.0635,
+      "step": 13046
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.694764177428411e-05,
+      "loss": 0.0969,
+      "step": 13048
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.6947173872356355e-05,
+      "loss": 0.0941,
+      "step": 13050
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.69467059704286e-05,
+      "loss": 0.0975,
+      "step": 13052
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.694623806850084e-05,
+      "loss": 0.0821,
+      "step": 13054
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.6945770166573086e-05,
+      "loss": 0.0905,
+      "step": 13056
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.694530226464533e-05,
+      "loss": 0.0847,
+      "step": 13058
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.694483436271758e-05,
+      "loss": 0.0875,
+      "step": 13060
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.694436646078982e-05,
+      "loss": 0.0798,
+      "step": 13062
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.694389855886206e-05,
+      "loss": 0.0898,
+      "step": 13064
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.694343065693431e-05,
+      "loss": 0.0901,
+      "step": 13066
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.6942962755006555e-05,
+      "loss": 0.1074,
+      "step": 13068
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.6942494853078794e-05,
+      "loss": 0.0895,
+      "step": 13070
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.694202695115104e-05,
+      "loss": 0.0819,
+      "step": 13072
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.6941559049223286e-05,
+      "loss": 0.1017,
+      "step": 13074
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.694109114729553e-05,
+      "loss": 0.0857,
+      "step": 13076
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.694062324536777e-05,
+      "loss": 0.0768,
+      "step": 13078
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.694015534344002e-05,
+      "loss": 0.0742,
+      "step": 13080
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.6939687441512256e-05,
+      "loss": 0.1015,
+      "step": 13082
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.693921953958451e-05,
+      "loss": 0.0989,
+      "step": 13084
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.693875163765675e-05,
+      "loss": 0.0934,
+      "step": 13086
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.6938283735728994e-05,
+      "loss": 0.0767,
+      "step": 13088
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.693781583380123e-05,
+      "loss": 0.0917,
+      "step": 13090
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.693734793187348e-05,
+      "loss": 0.0758,
+      "step": 13092
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.6936880029945725e-05,
+      "loss": 0.1066,
+      "step": 13094
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.693641212801797e-05,
+      "loss": 0.0982,
+      "step": 13096
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.693594422609021e-05,
+      "loss": 0.0868,
+      "step": 13098
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.6935476324162456e-05,
+      "loss": 0.0634,
+      "step": 13100
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.69350084222347e-05,
+      "loss": 0.0647,
+      "step": 13102
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.693454052030695e-05,
+      "loss": 0.075,
+      "step": 13104
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.693407261837919e-05,
+      "loss": 0.08,
+      "step": 13106
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.693360471645143e-05,
+      "loss": 0.0741,
+      "step": 13108
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.693313681452368e-05,
+      "loss": 0.0889,
+      "step": 13110
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.6932668912595924e-05,
+      "loss": 0.0747,
+      "step": 13112
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.6932201010668164e-05,
+      "loss": 0.071,
+      "step": 13114
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.693173310874041e-05,
+      "loss": 0.0843,
+      "step": 13116
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.6931265206812655e-05,
+      "loss": 0.0856,
+      "step": 13118
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.69307973048849e-05,
+      "loss": 0.1067,
+      "step": 13120
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.693032940295714e-05,
+      "loss": 0.0665,
+      "step": 13122
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.6929861501029386e-05,
+      "loss": 0.0728,
+      "step": 13124
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.6929393599101625e-05,
+      "loss": 0.0911,
+      "step": 13126
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.692892569717388e-05,
+      "loss": 0.1013,
+      "step": 13128
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.692845779524612e-05,
+      "loss": 0.0608,
+      "step": 13130
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.692798989331836e-05,
+      "loss": 0.0821,
+      "step": 13132
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.69275219913906e-05,
+      "loss": 0.1001,
+      "step": 13134
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.6927054089462855e-05,
+      "loss": 0.0867,
+      "step": 13136
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.6926586187535094e-05,
+      "loss": 0.0828,
+      "step": 13138
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.692611828560734e-05,
+      "loss": 0.0871,
+      "step": 13140
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.692565038367958e-05,
+      "loss": 0.095,
+      "step": 13142
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6925182481751825e-05,
+      "loss": 0.0667,
+      "step": 13144
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.692471457982407e-05,
+      "loss": 0.0967,
+      "step": 13146
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.692424667789632e-05,
+      "loss": 0.0768,
+      "step": 13148
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6923778775968556e-05,
+      "loss": 0.0924,
+      "step": 13150
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.69233108740408e-05,
+      "loss": 0.0602,
+      "step": 13152
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.692284297211305e-05,
+      "loss": 0.098,
+      "step": 13154
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6922375070185294e-05,
+      "loss": 0.0871,
+      "step": 13156
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.692190716825753e-05,
+      "loss": 0.0651,
+      "step": 13158
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.692167321729366e-05,
+      "loss": 0.169,
+      "step": 13160
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.69212053153659e-05,
+      "loss": 0.0942,
+      "step": 13162
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.692073741343815e-05,
+      "loss": 0.0786,
+      "step": 13164
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.692026951151039e-05,
+      "loss": 0.0576,
+      "step": 13166
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6919801609582636e-05,
+      "loss": 0.0804,
+      "step": 13168
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6919333707654875e-05,
+      "loss": 0.0876,
+      "step": 13170
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.691886580572712e-05,
+      "loss": 0.0728,
+      "step": 13172
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.691839790379937e-05,
+      "loss": 0.0686,
+      "step": 13174
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6917930001871606e-05,
+      "loss": 0.0839,
+      "step": 13176
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.691746209994385e-05,
+      "loss": 0.0708,
+      "step": 13178
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.69169941980161e-05,
+      "loss": 0.0886,
+      "step": 13180
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6916526296088344e-05,
+      "loss": 0.0977,
+      "step": 13182
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.691605839416058e-05,
+      "loss": 0.0758,
+      "step": 13184
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.691559049223283e-05,
+      "loss": 0.0708,
+      "step": 13186
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6915122590305075e-05,
+      "loss": 0.0796,
+      "step": 13188
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.691465468837732e-05,
+      "loss": 0.0808,
+      "step": 13190
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.691418678644956e-05,
+      "loss": 0.0826,
+      "step": 13192
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6913718884521806e-05,
+      "loss": 0.0899,
+      "step": 13194
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6913250982594045e-05,
+      "loss": 0.0659,
+      "step": 13196
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.69127830806663e-05,
+      "loss": 0.0826,
+      "step": 13198
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.691231517873854e-05,
+      "loss": 0.0948,
+      "step": 13200
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.691184727681078e-05,
+      "loss": 0.074,
+      "step": 13202
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.691137937488302e-05,
+      "loss": 0.0961,
+      "step": 13204
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.691091147295527e-05,
+      "loss": 0.0726,
+      "step": 13206
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6910443571027514e-05,
+      "loss": 0.0844,
+      "step": 13208
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.690997566909976e-05,
+      "loss": 0.0673,
+      "step": 13210
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6909507767172e-05,
+      "loss": 0.0706,
+      "step": 13212
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6909039865244245e-05,
+      "loss": 0.1078,
+      "step": 13214
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.690857196331649e-05,
+      "loss": 0.0711,
+      "step": 13216
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.6908104061388737e-05,
+      "loss": 0.0799,
+      "step": 13218
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.6907636159460976e-05,
+      "loss": 0.071,
+      "step": 13220
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.690716825753322e-05,
+      "loss": 0.0848,
+      "step": 13222
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.690670035560547e-05,
+      "loss": 0.0854,
+      "step": 13224
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.6906232453677713e-05,
+      "loss": 0.0851,
+      "step": 13226
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.690576455174995e-05,
+      "loss": 0.0686,
+      "step": 13228
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.69052966498222e-05,
+      "loss": 0.1077,
+      "step": 13230
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.6904828747894444e-05,
+      "loss": 0.0749,
+      "step": 13232
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.690436084596669e-05,
+      "loss": 0.0885,
+      "step": 13234
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.690389294403893e-05,
+      "loss": 0.0927,
+      "step": 13236
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.6903425042111175e-05,
+      "loss": 0.0764,
+      "step": 13238
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.6902957140183414e-05,
+      "loss": 0.0975,
+      "step": 13240
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.690248923825567e-05,
+      "loss": 0.0649,
+      "step": 13242
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.6902021336327906e-05,
+      "loss": 0.0905,
+      "step": 13244
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.690155343440015e-05,
+      "loss": 0.0837,
+      "step": 13246
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.690108553247239e-05,
+      "loss": 0.0855,
+      "step": 13248
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.6900617630544644e-05,
+      "loss": 0.0862,
+      "step": 13250
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.690014972861688e-05,
+      "loss": 0.0878,
+      "step": 13252
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.689968182668913e-05,
+      "loss": 0.0909,
+      "step": 13254
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.689921392476137e-05,
+      "loss": 0.0998,
+      "step": 13256
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.6898746022833614e-05,
+      "loss": 0.0955,
+      "step": 13258
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.689827812090586e-05,
+      "loss": 0.1101,
+      "step": 13260
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.6897810218978106e-05,
+      "loss": 0.0848,
+      "step": 13262
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.6897342317050345e-05,
+      "loss": 0.095,
+      "step": 13264
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.689687441512259e-05,
+      "loss": 0.086,
+      "step": 13266
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.689640651319484e-05,
+      "loss": 0.096,
+      "step": 13268
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.689593861126708e-05,
+      "loss": 0.0887,
+      "step": 13270
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.689547070933932e-05,
+      "loss": 0.0871,
+      "step": 13272
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.689500280741157e-05,
+      "loss": 0.1057,
+      "step": 13274
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.6894534905483814e-05,
+      "loss": 0.0844,
+      "step": 13276
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.689406700355606e-05,
+      "loss": 0.0959,
+      "step": 13278
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.68935991016283e-05,
+      "loss": 0.0793,
+      "step": 13280
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.6893131199700545e-05,
+      "loss": 0.0894,
+      "step": 13282
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.689266329777279e-05,
+      "loss": 0.1122,
+      "step": 13284
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.6892195395845037e-05,
+      "loss": 0.0787,
+      "step": 13286
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.6891727493917276e-05,
+      "loss": 0.0738,
+      "step": 13288
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.689125959198952e-05,
+      "loss": 0.1069,
+      "step": 13290
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.689079169006176e-05,
+      "loss": 0.0706,
+      "step": 13292
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.6890323788134013e-05,
+      "loss": 0.0879,
+      "step": 13294
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.688985588620625e-05,
+      "loss": 0.0904,
+      "step": 13296
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.68893879842785e-05,
+      "loss": 0.1037,
+      "step": 13298
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.688892008235074e-05,
+      "loss": 0.0918,
+      "step": 13300
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.688845218042299e-05,
+      "loss": 0.1026,
+      "step": 13302
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.688798427849523e-05,
+      "loss": 0.0706,
+      "step": 13304
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.6887516376567475e-05,
+      "loss": 0.0817,
+      "step": 13306
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.6887048474639714e-05,
+      "loss": 0.0742,
+      "step": 13308
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.688658057271196e-05,
+      "loss": 0.0702,
+      "step": 13310
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.6886112670784206e-05,
+      "loss": 0.1123,
+      "step": 13312
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.688564476885645e-05,
+      "loss": 0.1076,
+      "step": 13314
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.688517686692869e-05,
+      "loss": 0.108,
+      "step": 13316
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.688470896500094e-05,
+      "loss": 0.1129,
+      "step": 13318
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.688424106307318e-05,
+      "loss": 0.0735,
+      "step": 13320
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.688377316114543e-05,
+      "loss": 0.0797,
+      "step": 13322
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.688330525921767e-05,
+      "loss": 0.0802,
+      "step": 13324
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.6882837357289914e-05,
+      "loss": 0.0914,
+      "step": 13326
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.688236945536216e-05,
+      "loss": 0.0757,
+      "step": 13328
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.6881901553434406e-05,
+      "loss": 0.0616,
+      "step": 13330
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.6881433651506645e-05,
+      "loss": 0.0916,
+      "step": 13332
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.688096574957889e-05,
+      "loss": 0.0731,
+      "step": 13334
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.688049784765114e-05,
+      "loss": 0.0945,
+      "step": 13336
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.688002994572338e-05,
+      "loss": 0.1139,
+      "step": 13338
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.687956204379562e-05,
+      "loss": 0.0869,
+      "step": 13340
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.687909414186787e-05,
+      "loss": 0.0807,
+      "step": 13342
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.687862623994011e-05,
+      "loss": 0.0698,
+      "step": 13344
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.687815833801236e-05,
+      "loss": 0.1004,
+      "step": 13346
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.68776904360846e-05,
+      "loss": 0.0944,
+      "step": 13348
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.6877222534156845e-05,
+      "loss": 0.0938,
+      "step": 13350
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.6876754632229084e-05,
+      "loss": 0.0711,
+      "step": 13352
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.687628673030133e-05,
+      "loss": 0.0859,
+      "step": 13354
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.6875818828373576e-05,
+      "loss": 0.0783,
+      "step": 13356
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.687535092644582e-05,
+      "loss": 0.076,
+      "step": 13358
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.687488302451806e-05,
+      "loss": 0.0679,
+      "step": 13360
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.687441512259031e-05,
+      "loss": 0.0665,
+      "step": 13362
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.687394722066255e-05,
+      "loss": 0.0869,
+      "step": 13364
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.68734793187348e-05,
+      "loss": 0.0705,
+      "step": 13366
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.687301141680704e-05,
+      "loss": 0.1087,
+      "step": 13368
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.6872543514879284e-05,
+      "loss": 0.0877,
+      "step": 13370
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.687207561295153e-05,
+      "loss": 0.0842,
+      "step": 13372
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.6871607711023775e-05,
+      "loss": 0.0827,
+      "step": 13374
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.6871139809096015e-05,
+      "loss": 0.0682,
+      "step": 13376
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.687067190716826e-05,
+      "loss": 0.0734,
+      "step": 13378
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.6870204005240506e-05,
+      "loss": 0.1011,
+      "step": 13380
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.686973610331275e-05,
+      "loss": 0.0867,
+      "step": 13382
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.686926820138499e-05,
+      "loss": 0.0709,
+      "step": 13384
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.686880029945724e-05,
+      "loss": 0.0679,
+      "step": 13386
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.6868332397529476e-05,
+      "loss": 0.0864,
+      "step": 13388
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.686786449560173e-05,
+      "loss": 0.0858,
+      "step": 13390
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.686739659367397e-05,
+      "loss": 0.0901,
+      "step": 13392
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.6866928691746214e-05,
+      "loss": 0.0751,
+      "step": 13394
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.686646078981845e-05,
+      "loss": 0.0931,
+      "step": 13396
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.6865992887890706e-05,
+      "loss": 0.1245,
+      "step": 13398
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.6865524985962945e-05,
+      "loss": 0.0741,
+      "step": 13400
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.686505708403519e-05,
+      "loss": 0.0923,
+      "step": 13402
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.686458918210743e-05,
+      "loss": 0.0938,
+      "step": 13404
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.6864121280179676e-05,
+      "loss": 0.0802,
+      "step": 13406
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.686365337825192e-05,
+      "loss": 0.096,
+      "step": 13408
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.686318547632417e-05,
+      "loss": 0.1033,
+      "step": 13410
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.686271757439641e-05,
+      "loss": 0.0859,
+      "step": 13412
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.686224967246865e-05,
+      "loss": 0.1024,
+      "step": 13414
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.68617817705409e-05,
+      "loss": 0.0806,
+      "step": 13416
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.6861313868613145e-05,
+      "loss": 0.1089,
+      "step": 13418
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.6860845966685384e-05,
+      "loss": 0.1144,
+      "step": 13420
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.686037806475763e-05,
+      "loss": 0.089,
+      "step": 13422
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.6859910162829876e-05,
+      "loss": 0.1023,
+      "step": 13424
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.6859442260902115e-05,
+      "loss": 0.0765,
+      "step": 13426
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.685897435897436e-05,
+      "loss": 0.0894,
+      "step": 13428
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.68585064570466e-05,
+      "loss": 0.0809,
+      "step": 13430
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.685803855511885e-05,
+      "loss": 0.0723,
+      "step": 13432
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.685757065319109e-05,
+      "loss": 0.0896,
+      "step": 13434
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.685710275126334e-05,
+      "loss": 0.0995,
+      "step": 13436
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.685663484933558e-05,
+      "loss": 0.0846,
+      "step": 13438
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.685616694740782e-05,
+      "loss": 0.074,
+      "step": 13440
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.685569904548007e-05,
+      "loss": 0.1063,
+      "step": 13442
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.6855231143552315e-05,
+      "loss": 0.0902,
+      "step": 13444
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.6854763241624554e-05,
+      "loss": 0.0906,
+      "step": 13446
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.68542953396968e-05,
+      "loss": 0.0996,
+      "step": 13448
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.6853827437769045e-05,
+      "loss": 0.1113,
+      "step": 13450
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.685335953584129e-05,
+      "loss": 0.0751,
+      "step": 13452
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.685289163391353e-05,
+      "loss": 0.0791,
+      "step": 13454
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.6852423731985776e-05,
+      "loss": 0.0825,
+      "step": 13456
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.685195583005802e-05,
+      "loss": 0.0831,
+      "step": 13458
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.685148792813027e-05,
+      "loss": 0.0823,
+      "step": 13460
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.685102002620251e-05,
+      "loss": 0.0809,
+      "step": 13462
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.685055212427475e-05,
+      "loss": 0.0869,
+      "step": 13464
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.6850084222347e-05,
+      "loss": 0.0849,
+      "step": 13466
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.6849616320419245e-05,
+      "loss": 0.079,
+      "step": 13468
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.6849148418491484e-05,
+      "loss": 0.0844,
+      "step": 13470
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.684868051656373e-05,
+      "loss": 0.0985,
+      "step": 13472
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.684821261463597e-05,
+      "loss": 0.0731,
+      "step": 13474
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.684774471270822e-05,
+      "loss": 0.0709,
+      "step": 13476
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.684727681078046e-05,
+      "loss": 0.0901,
+      "step": 13478
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.684680890885271e-05,
+      "loss": 0.0801,
+      "step": 13480
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.6846341006924946e-05,
+      "loss": 0.0704,
+      "step": 13482
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.68458731049972e-05,
+      "loss": 0.0871,
+      "step": 13484
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.684540520306944e-05,
+      "loss": 0.1005,
+      "step": 13486
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.6844937301141684e-05,
+      "loss": 0.0727,
+      "step": 13488
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.684446939921392e-05,
+      "loss": 0.0639,
+      "step": 13490
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.684400149728617e-05,
+      "loss": 0.0865,
+      "step": 13492
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.6843533595358415e-05,
+      "loss": 0.069,
+      "step": 13494
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.684306569343066e-05,
+      "loss": 0.0564,
+      "step": 13496
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.68425977915029e-05,
+      "loss": 0.0919,
+      "step": 13498
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.6842129889575146e-05,
+      "loss": 0.0692,
+      "step": 13500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.684166198764739e-05,
+      "loss": 0.0836,
+      "step": 13502
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.684119408571964e-05,
+      "loss": 0.0906,
+      "step": 13504
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.684072618379188e-05,
+      "loss": 0.0736,
+      "step": 13506
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.684025828186412e-05,
+      "loss": 0.0942,
+      "step": 13508
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.683979037993637e-05,
+      "loss": 0.0831,
+      "step": 13510
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.6839322478008615e-05,
+      "loss": 0.0843,
+      "step": 13512
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.6838854576080854e-05,
+      "loss": 0.0816,
+      "step": 13514
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.68383866741531e-05,
+      "loss": 0.0848,
+      "step": 13516
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.683791877222534e-05,
+      "loss": 0.0976,
+      "step": 13518
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.683745087029759e-05,
+      "loss": 0.0681,
+      "step": 13520
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.683698296836983e-05,
+      "loss": 0.0999,
+      "step": 13522
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.6836515066442076e-05,
+      "loss": 0.0753,
+      "step": 13524
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.6836047164514316e-05,
+      "loss": 0.0841,
+      "step": 13526
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.683557926258657e-05,
+      "loss": 0.0818,
+      "step": 13528
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.683511136065881e-05,
+      "loss": 0.0804,
+      "step": 13530
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.683464345873105e-05,
+      "loss": 0.0679,
+      "step": 13532
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.683417555680329e-05,
+      "loss": 0.091,
+      "step": 13534
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.683370765487554e-05,
+      "loss": 0.0876,
+      "step": 13536
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.6833239752947784e-05,
+      "loss": 0.0914,
+      "step": 13538
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.683277185102003e-05,
+      "loss": 0.1029,
+      "step": 13540
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.683230394909227e-05,
+      "loss": 0.0933,
+      "step": 13542
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.6831836047164515e-05,
+      "loss": 0.0913,
+      "step": 13544
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.683136814523676e-05,
+      "loss": 0.1064,
+      "step": 13546
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.683090024330901e-05,
+      "loss": 0.0847,
+      "step": 13548
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.6830432341381246e-05,
+      "loss": 0.101,
+      "step": 13550
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.682996443945349e-05,
+      "loss": 0.0917,
+      "step": 13552
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.682949653752574e-05,
+      "loss": 0.0954,
+      "step": 13554
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.6829028635597984e-05,
+      "loss": 0.0735,
+      "step": 13556
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.682856073367022e-05,
+      "loss": 0.0715,
+      "step": 13558
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.682809283174247e-05,
+      "loss": 0.0936,
+      "step": 13560
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.6827624929814715e-05,
+      "loss": 0.0892,
+      "step": 13562
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.682715702788696e-05,
+      "loss": 0.085,
+      "step": 13564
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.68266891259592e-05,
+      "loss": 0.0854,
+      "step": 13566
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.6826221224031446e-05,
+      "loss": 0.0997,
+      "step": 13568
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.6825753322103685e-05,
+      "loss": 0.0778,
+      "step": 13570
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.682528542017594e-05,
+      "loss": 0.0742,
+      "step": 13572
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.682481751824818e-05,
+      "loss": 0.0959,
+      "step": 13574
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.682434961632042e-05,
+      "loss": 0.0921,
+      "step": 13576
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.682388171439266e-05,
+      "loss": 0.0991,
+      "step": 13578
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.6823413812464915e-05,
+      "loss": 0.0825,
+      "step": 13580
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.6822945910537154e-05,
+      "loss": 0.0795,
+      "step": 13582
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.68224780086094e-05,
+      "loss": 0.0881,
+      "step": 13584
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.682201010668164e-05,
+      "loss": 0.0804,
+      "step": 13586
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.6821542204753885e-05,
+      "loss": 0.0766,
+      "step": 13588
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.682107430282613e-05,
+      "loss": 0.0833,
+      "step": 13590
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.6820606400898376e-05,
+      "loss": 0.0696,
+      "step": 13592
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.6820138498970616e-05,
+      "loss": 0.0769,
+      "step": 13594
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.681967059704286e-05,
+      "loss": 0.0775,
+      "step": 13596
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.681920269511511e-05,
+      "loss": 0.0818,
+      "step": 13598
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.681873479318735e-05,
+      "loss": 0.0834,
+      "step": 13600
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.681826689125959e-05,
+      "loss": 0.1017,
+      "step": 13602
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.681779898933184e-05,
+      "loss": 0.077,
+      "step": 13604
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.6817331087404084e-05,
+      "loss": 0.0925,
+      "step": 13606
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.681686318547633e-05,
+      "loss": 0.0859,
+      "step": 13608
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.681639528354857e-05,
+      "loss": 0.0888,
+      "step": 13610
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.6815927381620815e-05,
+      "loss": 0.1129,
+      "step": 13612
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.681545947969306e-05,
+      "loss": 0.0808,
+      "step": 13614
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.681499157776531e-05,
+      "loss": 0.0833,
+      "step": 13616
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.6814523675837546e-05,
+      "loss": 0.0898,
+      "step": 13618
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.681405577390979e-05,
+      "loss": 0.0736,
+      "step": 13620
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.681358787198203e-05,
+      "loss": 0.0834,
+      "step": 13622
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.6813119970054284e-05,
+      "loss": 0.0705,
+      "step": 13624
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.681265206812652e-05,
+      "loss": 0.0742,
+      "step": 13626
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.681241811716264e-05,
+      "loss": 0.0926,
+      "step": 13628
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.681195021523489e-05,
+      "loss": 0.0784,
+      "step": 13630
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.6811482313307134e-05,
+      "loss": 0.0995,
+      "step": 13632
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.681101441137938e-05,
+      "loss": 0.0833,
+      "step": 13634
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.681054650945162e-05,
+      "loss": 0.0907,
+      "step": 13636
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.6810078607523865e-05,
+      "loss": 0.0694,
+      "step": 13638
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.6809610705596105e-05,
+      "loss": 0.07,
+      "step": 13640
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.680914280366836e-05,
+      "loss": 0.0885,
+      "step": 13642
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6808674901740596e-05,
+      "loss": 0.0748,
+      "step": 13644
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.680820699981284e-05,
+      "loss": 0.07,
+      "step": 13646
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.680773909788508e-05,
+      "loss": 0.0625,
+      "step": 13648
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.680727119595733e-05,
+      "loss": 0.0632,
+      "step": 13650
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.680680329402957e-05,
+      "loss": 0.0726,
+      "step": 13652
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.680633539210182e-05,
+      "loss": 0.0738,
+      "step": 13654
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.680586749017406e-05,
+      "loss": 0.0956,
+      "step": 13656
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6805399588246304e-05,
+      "loss": 0.0864,
+      "step": 13658
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.680493168631855e-05,
+      "loss": 0.0802,
+      "step": 13660
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6804463784390796e-05,
+      "loss": 0.0783,
+      "step": 13662
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6803995882463035e-05,
+      "loss": 0.0808,
+      "step": 13664
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.680352798053528e-05,
+      "loss": 0.083,
+      "step": 13666
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.680306007860753e-05,
+      "loss": 0.0921,
+      "step": 13668
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.680259217667977e-05,
+      "loss": 0.0798,
+      "step": 13670
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.680212427475201e-05,
+      "loss": 0.0709,
+      "step": 13672
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.680165637282426e-05,
+      "loss": 0.0752,
+      "step": 13674
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6801188470896504e-05,
+      "loss": 0.0701,
+      "step": 13676
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.680072056896875e-05,
+      "loss": 0.1258,
+      "step": 13678
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.680025266704099e-05,
+      "loss": 0.0913,
+      "step": 13680
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6799784765113235e-05,
+      "loss": 0.0848,
+      "step": 13682
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6799316863185474e-05,
+      "loss": 0.0862,
+      "step": 13684
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.679884896125773e-05,
+      "loss": 0.1008,
+      "step": 13686
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6798381059329966e-05,
+      "loss": 0.0804,
+      "step": 13688
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.679791315740221e-05,
+      "loss": 0.0801,
+      "step": 13690
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.679744525547445e-05,
+      "loss": 0.0761,
+      "step": 13692
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6796977353546704e-05,
+      "loss": 0.1032,
+      "step": 13694
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.679650945161894e-05,
+      "loss": 0.078,
+      "step": 13696
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.679604154969119e-05,
+      "loss": 0.0774,
+      "step": 13698
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.679557364776343e-05,
+      "loss": 0.0831,
+      "step": 13700
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6795105745835674e-05,
+      "loss": 0.0748,
+      "step": 13702
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.679463784390792e-05,
+      "loss": 0.0989,
+      "step": 13704
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6794169941980165e-05,
+      "loss": 0.0964,
+      "step": 13706
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6793702040052405e-05,
+      "loss": 0.0724,
+      "step": 13708
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.679323413812465e-05,
+      "loss": 0.0925,
+      "step": 13710
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.6792766236196896e-05,
+      "loss": 0.0866,
+      "step": 13712
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.679229833426914e-05,
+      "loss": 0.0824,
+      "step": 13714
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.679183043234138e-05,
+      "loss": 0.0734,
+      "step": 13716
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.679136253041363e-05,
+      "loss": 0.0826,
+      "step": 13718
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.679089462848587e-05,
+      "loss": 0.0684,
+      "step": 13720
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.679042672655812e-05,
+      "loss": 0.0783,
+      "step": 13722
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.678995882463036e-05,
+      "loss": 0.0761,
+      "step": 13724
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6789490922702604e-05,
+      "loss": 0.0723,
+      "step": 13726
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.678902302077485e-05,
+      "loss": 0.093,
+      "step": 13728
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6788555118847096e-05,
+      "loss": 0.0994,
+      "step": 13730
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6788087216919335e-05,
+      "loss": 0.0745,
+      "step": 13732
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.678761931499158e-05,
+      "loss": 0.0751,
+      "step": 13734
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.678715141306382e-05,
+      "loss": 0.0986,
+      "step": 13736
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.678668351113607e-05,
+      "loss": 0.0797,
+      "step": 13738
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.678621560920831e-05,
+      "loss": 0.0839,
+      "step": 13740
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.678574770728056e-05,
+      "loss": 0.0674,
+      "step": 13742
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.67852798053528e-05,
+      "loss": 0.0608,
+      "step": 13744
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.678481190342505e-05,
+      "loss": 0.0822,
+      "step": 13746
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.678434400149729e-05,
+      "loss": 0.0805,
+      "step": 13748
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6783876099569535e-05,
+      "loss": 0.129,
+      "step": 13750
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6783408197641774e-05,
+      "loss": 0.089,
+      "step": 13752
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.678294029571402e-05,
+      "loss": 0.0838,
+      "step": 13754
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6782472393786266e-05,
+      "loss": 0.0747,
+      "step": 13756
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.678200449185851e-05,
+      "loss": 0.082,
+      "step": 13758
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.678153658993075e-05,
+      "loss": 0.0776,
+      "step": 13760
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6781068688003e-05,
+      "loss": 0.0814,
+      "step": 13762
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.678060078607524e-05,
+      "loss": 0.0779,
+      "step": 13764
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.678013288414749e-05,
+      "loss": 0.0998,
+      "step": 13766
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.677966498221973e-05,
+      "loss": 0.0682,
+      "step": 13768
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6779197080291974e-05,
+      "loss": 0.0751,
+      "step": 13770
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.677872917836422e-05,
+      "loss": 0.0841,
+      "step": 13772
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6778261276436465e-05,
+      "loss": 0.0675,
+      "step": 13774
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6777793374508705e-05,
+      "loss": 0.0807,
+      "step": 13776
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.677732547258095e-05,
+      "loss": 0.0729,
+      "step": 13778
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.6776857570653196e-05,
+      "loss": 0.0751,
+      "step": 13780
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.677638966872544e-05,
+      "loss": 0.0849,
+      "step": 13782
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.677592176679768e-05,
+      "loss": 0.099,
+      "step": 13784
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.677545386486993e-05,
+      "loss": 0.0918,
+      "step": 13786
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6774985962942167e-05,
+      "loss": 0.1082,
+      "step": 13788
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.677451806101442e-05,
+      "loss": 0.071,
+      "step": 13790
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.677405015908666e-05,
+      "loss": 0.1029,
+      "step": 13792
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6773582257158904e-05,
+      "loss": 0.1171,
+      "step": 13794
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6773114355231143e-05,
+      "loss": 0.0791,
+      "step": 13796
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.677264645330339e-05,
+      "loss": 0.0978,
+      "step": 13798
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6772178551375635e-05,
+      "loss": 0.0806,
+      "step": 13800
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.677171064944788e-05,
+      "loss": 0.0735,
+      "step": 13802
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.677124274752012e-05,
+      "loss": 0.0793,
+      "step": 13804
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6770774845592366e-05,
+      "loss": 0.0637,
+      "step": 13806
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.677030694366461e-05,
+      "loss": 0.0803,
+      "step": 13808
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.676983904173685e-05,
+      "loss": 0.0876,
+      "step": 13810
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.67693711398091e-05,
+      "loss": 0.081,
+      "step": 13812
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6768903237881336e-05,
+      "loss": 0.0778,
+      "step": 13814
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.676843533595359e-05,
+      "loss": 0.0915,
+      "step": 13816
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.676796743402583e-05,
+      "loss": 0.1146,
+      "step": 13818
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6767499532098074e-05,
+      "loss": 0.0943,
+      "step": 13820
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.676703163017031e-05,
+      "loss": 0.0681,
+      "step": 13822
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6766563728242566e-05,
+      "loss": 0.0717,
+      "step": 13824
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6766095826314805e-05,
+      "loss": 0.0849,
+      "step": 13826
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.676562792438705e-05,
+      "loss": 0.0759,
+      "step": 13828
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.676516002245929e-05,
+      "loss": 0.0588,
+      "step": 13830
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6764692120531536e-05,
+      "loss": 0.0837,
+      "step": 13832
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.676422421860378e-05,
+      "loss": 0.0741,
+      "step": 13834
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.676375631667603e-05,
+      "loss": 0.0802,
+      "step": 13836
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.676328841474827e-05,
+      "loss": 0.0932,
+      "step": 13838
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.676282051282051e-05,
+      "loss": 0.0816,
+      "step": 13840
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.676235261089276e-05,
+      "loss": 0.0683,
+      "step": 13842
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6761884708965005e-05,
+      "loss": 0.0657,
+      "step": 13844
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6761416807037244e-05,
+      "loss": 0.0708,
+      "step": 13846
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.676094890510949e-05,
+      "loss": 0.0979,
+      "step": 13848
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6760481003181736e-05,
+      "loss": 0.075,
+      "step": 13850
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.676001310125398e-05,
+      "loss": 0.0686,
+      "step": 13852
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.675954519932622e-05,
+      "loss": 0.0878,
+      "step": 13854
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6759077297398467e-05,
+      "loss": 0.0896,
+      "step": 13856
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.675860939547071e-05,
+      "loss": 0.0891,
+      "step": 13858
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.675814149354296e-05,
+      "loss": 0.0731,
+      "step": 13860
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.67576735916152e-05,
+      "loss": 0.0842,
+      "step": 13862
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6757205689687443e-05,
+      "loss": 0.0897,
+      "step": 13864
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.675673778775968e-05,
+      "loss": 0.116,
+      "step": 13866
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6756269885831935e-05,
+      "loss": 0.0838,
+      "step": 13868
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6755801983904174e-05,
+      "loss": 0.0742,
+      "step": 13870
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.675533408197642e-05,
+      "loss": 0.0804,
+      "step": 13872
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.675486618004866e-05,
+      "loss": 0.0811,
+      "step": 13874
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.675439827812091e-05,
+      "loss": 0.0847,
+      "step": 13876
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.675393037619315e-05,
+      "loss": 0.1092,
+      "step": 13878
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.67534624742654e-05,
+      "loss": 0.0723,
+      "step": 13880
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6752994572337636e-05,
+      "loss": 0.0644,
+      "step": 13882
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.675252667040988e-05,
+      "loss": 0.0887,
+      "step": 13884
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.675205876848213e-05,
+      "loss": 0.1023,
+      "step": 13886
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6751590866554374e-05,
+      "loss": 0.075,
+      "step": 13888
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.675112296462661e-05,
+      "loss": 0.0983,
+      "step": 13890
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.675065506269886e-05,
+      "loss": 0.1074,
+      "step": 13892
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6750187160771105e-05,
+      "loss": 0.0783,
+      "step": 13894
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.674971925884335e-05,
+      "loss": 0.0913,
+      "step": 13896
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.674925135691559e-05,
+      "loss": 0.0891,
+      "step": 13898
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6748783454987836e-05,
+      "loss": 0.087,
+      "step": 13900
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.674831555306008e-05,
+      "loss": 0.0637,
+      "step": 13902
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.674784765113233e-05,
+      "loss": 0.0705,
+      "step": 13904
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.674737974920457e-05,
+      "loss": 0.0833,
+      "step": 13906
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.674691184727681e-05,
+      "loss": 0.0829,
+      "step": 13908
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.674644394534906e-05,
+      "loss": 0.0869,
+      "step": 13910
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6745976043421305e-05,
+      "loss": 0.0823,
+      "step": 13912
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6745508141493544e-05,
+      "loss": 0.0874,
+      "step": 13914
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.674504023956579e-05,
+      "loss": 0.0743,
+      "step": 13916
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.674457233763803e-05,
+      "loss": 0.1075,
+      "step": 13918
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.674410443571028e-05,
+      "loss": 0.0799,
+      "step": 13920
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.674363653378252e-05,
+      "loss": 0.1071,
+      "step": 13922
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6743168631854767e-05,
+      "loss": 0.0778,
+      "step": 13924
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6742700729927006e-05,
+      "loss": 0.0834,
+      "step": 13926
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.674223282799925e-05,
+      "loss": 0.0907,
+      "step": 13928
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.67417649260715e-05,
+      "loss": 0.0761,
+      "step": 13930
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.6741297024143743e-05,
+      "loss": 0.1059,
+      "step": 13932
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.674082912221598e-05,
+      "loss": 0.0793,
+      "step": 13934
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.674036122028823e-05,
+      "loss": 0.0738,
+      "step": 13936
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.6739893318360474e-05,
+      "loss": 0.1102,
+      "step": 13938
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.673942541643272e-05,
+      "loss": 0.0918,
+      "step": 13940
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.673895751450496e-05,
+      "loss": 0.0736,
+      "step": 13942
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.6738489612577205e-05,
+      "loss": 0.0583,
+      "step": 13944
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.673802171064945e-05,
+      "loss": 0.114,
+      "step": 13946
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.67375538087217e-05,
+      "loss": 0.1063,
+      "step": 13948
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.6737085906793936e-05,
+      "loss": 0.0842,
+      "step": 13950
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.673661800486618e-05,
+      "loss": 0.074,
+      "step": 13952
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.673615010293843e-05,
+      "loss": 0.0677,
+      "step": 13954
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.6735682201010674e-05,
+      "loss": 0.0717,
+      "step": 13956
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.673521429908291e-05,
+      "loss": 0.0879,
+      "step": 13958
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.673474639715516e-05,
+      "loss": 0.069,
+      "step": 13960
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.67342784952274e-05,
+      "loss": 0.0787,
+      "step": 13962
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.673381059329965e-05,
+      "loss": 0.0953,
+      "step": 13964
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.673334269137189e-05,
+      "loss": 0.0925,
+      "step": 13966
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.6732874789444136e-05,
+      "loss": 0.0702,
+      "step": 13968
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.6732406887516375e-05,
+      "loss": 0.1242,
+      "step": 13970
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.673193898558863e-05,
+      "loss": 0.0616,
+      "step": 13972
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.673147108366087e-05,
+      "loss": 0.1254,
+      "step": 13974
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.673100318173311e-05,
+      "loss": 0.1014,
+      "step": 13976
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.673053527980535e-05,
+      "loss": 0.0924,
+      "step": 13978
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.67300673778776e-05,
+      "loss": 0.0804,
+      "step": 13980
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.6729599475949844e-05,
+      "loss": 0.0965,
+      "step": 13982
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.672913157402209e-05,
+      "loss": 0.0666,
+      "step": 13984
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.672866367209433e-05,
+      "loss": 0.0786,
+      "step": 13986
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.6728195770166575e-05,
+      "loss": 0.0736,
+      "step": 13988
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.672772786823882e-05,
+      "loss": 0.0724,
+      "step": 13990
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.6727259966311067e-05,
+      "loss": 0.069,
+      "step": 13992
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.6726792064383306e-05,
+      "loss": 0.0954,
+      "step": 13994
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.672632416245555e-05,
+      "loss": 0.0802,
+      "step": 13996
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.67258562605278e-05,
+      "loss": 0.0789,
+      "step": 13998
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.6725388358600043e-05,
+      "loss": 0.1104,
+      "step": 14000
+    },
+    {
+      "epoch": 1.97,
+      "eval_gen_len": 28.2931,
+      "eval_loss": 1.0474748611450195,
+      "eval_meteor": 0.048,
+      "eval_runtime": 13.517,
+      "eval_samples_per_second": 4.291,
+      "eval_steps_per_second": 0.592,
+      "step": 14000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.672492045667228e-05,
+      "loss": 0.119,
+      "step": 14002
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.672445255474453e-05,
+      "loss": 0.0843,
+      "step": 14004
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.6723984652816774e-05,
+      "loss": 0.0611,
+      "step": 14006
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.672351675088902e-05,
+      "loss": 0.0763,
+      "step": 14008
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.672304884896126e-05,
+      "loss": 0.0992,
+      "step": 14010
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.6722580947033505e-05,
+      "loss": 0.1185,
+      "step": 14012
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.6722113045105745e-05,
+      "loss": 0.0689,
+      "step": 14014
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.6721645143178e-05,
+      "loss": 0.0783,
+      "step": 14016
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.6721177241250236e-05,
+      "loss": 0.0736,
+      "step": 14018
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.672070933932248e-05,
+      "loss": 0.0734,
+      "step": 14020
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.672024143739472e-05,
+      "loss": 0.08,
+      "step": 14022
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.6719773535466974e-05,
+      "loss": 0.0764,
+      "step": 14024
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.671930563353921e-05,
+      "loss": 0.0828,
+      "step": 14026
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.671883773161146e-05,
+      "loss": 0.0766,
+      "step": 14028
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.67183698296837e-05,
+      "loss": 0.0948,
+      "step": 14030
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.6717901927755944e-05,
+      "loss": 0.1003,
+      "step": 14032
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.671743402582819e-05,
+      "loss": 0.0863,
+      "step": 14034
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.6716966123900436e-05,
+      "loss": 0.0965,
+      "step": 14036
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.6716498221972675e-05,
+      "loss": 0.0851,
+      "step": 14038
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.671603032004492e-05,
+      "loss": 0.1,
+      "step": 14040
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.671556241811717e-05,
+      "loss": 0.096,
+      "step": 14042
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.671509451618941e-05,
+      "loss": 0.0889,
+      "step": 14044
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.671462661426165e-05,
+      "loss": 0.0792,
+      "step": 14046
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.67141587123339e-05,
+      "loss": 0.082,
+      "step": 14048
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.6713690810406144e-05,
+      "loss": 0.0832,
+      "step": 14050
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.671322290847839e-05,
+      "loss": 0.0758,
+      "step": 14052
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.671275500655063e-05,
+      "loss": 0.0826,
+      "step": 14054
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.6712287104622875e-05,
+      "loss": 0.0842,
+      "step": 14056
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.671181920269512e-05,
+      "loss": 0.099,
+      "step": 14058
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.671135130076736e-05,
+      "loss": 0.0892,
+      "step": 14060
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.6710883398839606e-05,
+      "loss": 0.1079,
+      "step": 14062
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.6710415496911845e-05,
+      "loss": 0.0814,
+      "step": 14064
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.670994759498409e-05,
+      "loss": 0.1106,
+      "step": 14066
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.670947969305634e-05,
+      "loss": 0.0782,
+      "step": 14068
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.670901179112858e-05,
+      "loss": 0.1078,
+      "step": 14070
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.670854388920082e-05,
+      "loss": 0.0754,
+      "step": 14072
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.670807598727307e-05,
+      "loss": 0.0854,
+      "step": 14074
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.6707608085345314e-05,
+      "loss": 0.0843,
+      "step": 14076
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.670714018341756e-05,
+      "loss": 0.0886,
+      "step": 14078
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.67066722814898e-05,
+      "loss": 0.0833,
+      "step": 14080
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.6706204379562045e-05,
+      "loss": 0.0828,
+      "step": 14082
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.670573647763429e-05,
+      "loss": 0.079,
+      "step": 14084
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.6705268575706536e-05,
+      "loss": 0.1032,
+      "step": 14086
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.6704800673778775e-05,
+      "loss": 0.0608,
+      "step": 14088
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.670433277185102e-05,
+      "loss": 0.0956,
+      "step": 14090
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.670386486992327e-05,
+      "loss": 0.1089,
+      "step": 14092
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.670339696799551e-05,
+      "loss": 0.0756,
+      "step": 14094
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.670292906606775e-05,
+      "loss": 0.0905,
+      "step": 14096
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.670246116414e-05,
+      "loss": 0.1066,
+      "step": 14098
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.670199326221224e-05,
+      "loss": 0.096,
+      "step": 14100
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.670152536028449e-05,
+      "loss": 0.0998,
+      "step": 14102
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.670105745835673e-05,
+      "loss": 0.1179,
+      "step": 14104
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.6700589556428975e-05,
+      "loss": 0.1025,
+      "step": 14106
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.6700121654501214e-05,
+      "loss": 0.0868,
+      "step": 14108
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.669965375257346e-05,
+      "loss": 0.071,
+      "step": 14110
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.6699185850645706e-05,
+      "loss": 0.1024,
+      "step": 14112
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.669871794871795e-05,
+      "loss": 0.0889,
+      "step": 14114
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.669825004679019e-05,
+      "loss": 0.084,
+      "step": 14116
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.669778214486244e-05,
+      "loss": 0.1139,
+      "step": 14118
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.669731424293468e-05,
+      "loss": 0.0789,
+      "step": 14120
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.669684634100693e-05,
+      "loss": 0.1202,
+      "step": 14122
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.669637843907917e-05,
+      "loss": 0.0674,
+      "step": 14124
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.6695910537151414e-05,
+      "loss": 0.0751,
+      "step": 14126
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.669544263522366e-05,
+      "loss": 0.0863,
+      "step": 14128
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.6694974733295906e-05,
+      "loss": 0.0698,
+      "step": 14130
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.6694506831368145e-05,
+      "loss": 0.0945,
+      "step": 14132
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.669403892944039e-05,
+      "loss": 0.1053,
+      "step": 14134
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.669357102751264e-05,
+      "loss": 0.1102,
+      "step": 14136
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.669310312558488e-05,
+      "loss": 0.0887,
+      "step": 14138
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.669263522365712e-05,
+      "loss": 0.096,
+      "step": 14140
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.669216732172937e-05,
+      "loss": 0.0923,
+      "step": 14142
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.669169941980161e-05,
+      "loss": 0.0653,
+      "step": 14144
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.669123151787386e-05,
+      "loss": 0.0936,
+      "step": 14146
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.66907636159461e-05,
+      "loss": 0.1005,
+      "step": 14148
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.6690295714018345e-05,
+      "loss": 0.0783,
+      "step": 14150
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.6689827812090584e-05,
+      "loss": 0.0802,
+      "step": 14152
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.6689359910162836e-05,
+      "loss": 0.0849,
+      "step": 14154
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.6688892008235076e-05,
+      "loss": 0.0584,
+      "step": 14156
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.668842410630732e-05,
+      "loss": 0.0892,
+      "step": 14158
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.668795620437956e-05,
+      "loss": 0.0815,
+      "step": 14160
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.6687488302451806e-05,
+      "loss": 0.0966,
+      "step": 14162
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.668702040052405e-05,
+      "loss": 0.0969,
+      "step": 14164
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.66865524985963e-05,
+      "loss": 0.0947,
+      "step": 14166
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.668608459666854e-05,
+      "loss": 0.0803,
+      "step": 14168
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.668561669474078e-05,
+      "loss": 0.0776,
+      "step": 14170
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.668514879281303e-05,
+      "loss": 0.0828,
+      "step": 14172
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.6684680890885275e-05,
+      "loss": 0.0849,
+      "step": 14174
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.6684212988957514e-05,
+      "loss": 0.0927,
+      "step": 14176
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.668374508702976e-05,
+      "loss": 0.1052,
+      "step": 14178
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.6683277185102006e-05,
+      "loss": 0.1121,
+      "step": 14180
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.668280928317425e-05,
+      "loss": 0.0738,
+      "step": 14182
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.668234138124649e-05,
+      "loss": 0.0724,
+      "step": 14184
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.668187347931874e-05,
+      "loss": 0.0853,
+      "step": 14186
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.668140557739098e-05,
+      "loss": 0.086,
+      "step": 14188
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.668093767546323e-05,
+      "loss": 0.0838,
+      "step": 14190
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.668046977353547e-05,
+      "loss": 0.064,
+      "step": 14192
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.6680001871607714e-05,
+      "loss": 0.0782,
+      "step": 14194
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.667953396967995e-05,
+      "loss": 0.0698,
+      "step": 14196
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.6679066067752206e-05,
+      "loss": 0.099,
+      "step": 14198
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.6678598165824445e-05,
+      "loss": 0.089,
+      "step": 14200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.667813026389669e-05,
+      "loss": 0.0976,
+      "step": 14202
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.667766236196893e-05,
+      "loss": 0.0739,
+      "step": 14204
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.6677194460041176e-05,
+      "loss": 0.0998,
+      "step": 14206
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.667672655811342e-05,
+      "loss": 0.0815,
+      "step": 14208
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.667625865618567e-05,
+      "loss": 0.0717,
+      "step": 14210
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.667579075425791e-05,
+      "loss": 0.0927,
+      "step": 14212
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.667532285233015e-05,
+      "loss": 0.0788,
+      "step": 14214
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.66748549504024e-05,
+      "loss": 0.0717,
+      "step": 14216
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6674387048474645e-05,
+      "loss": 0.0824,
+      "step": 14218
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6673919146546884e-05,
+      "loss": 0.0815,
+      "step": 14220
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.667345124461913e-05,
+      "loss": 0.0733,
+      "step": 14222
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6672983342691376e-05,
+      "loss": 0.068,
+      "step": 14224
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.667251544076362e-05,
+      "loss": 0.0959,
+      "step": 14226
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.667204753883586e-05,
+      "loss": 0.0813,
+      "step": 14228
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6671579636908106e-05,
+      "loss": 0.1064,
+      "step": 14230
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.667111173498035e-05,
+      "loss": 0.0949,
+      "step": 14232
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.66706438330526e-05,
+      "loss": 0.104,
+      "step": 14234
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.667017593112484e-05,
+      "loss": 0.0782,
+      "step": 14236
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.666970802919708e-05,
+      "loss": 0.0857,
+      "step": 14238
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.666924012726932e-05,
+      "loss": 0.0854,
+      "step": 14240
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6668772225341575e-05,
+      "loss": 0.1081,
+      "step": 14242
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6668304323413814e-05,
+      "loss": 0.0633,
+      "step": 14244
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.666783642148606e-05,
+      "loss": 0.0887,
+      "step": 14246
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.666760247052218e-05,
+      "loss": 0.1334,
+      "step": 14248
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6667134568594426e-05,
+      "loss": 0.0486,
+      "step": 14250
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.666666666666667e-05,
+      "loss": 0.0616,
+      "step": 14252
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.666619876473891e-05,
+      "loss": 0.0566,
+      "step": 14254
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.666573086281116e-05,
+      "loss": 0.0476,
+      "step": 14256
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6665262960883396e-05,
+      "loss": 0.0517,
+      "step": 14258
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.666479505895565e-05,
+      "loss": 0.0507,
+      "step": 14260
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.666432715702789e-05,
+      "loss": 0.0484,
+      "step": 14262
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6663859255100134e-05,
+      "loss": 0.0614,
+      "step": 14264
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.666339135317237e-05,
+      "loss": 0.0486,
+      "step": 14266
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6662923451244625e-05,
+      "loss": 0.0598,
+      "step": 14268
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6662455549316864e-05,
+      "loss": 0.0537,
+      "step": 14270
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.666198764738911e-05,
+      "loss": 0.0677,
+      "step": 14272
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.666151974546135e-05,
+      "loss": 0.0623,
+      "step": 14274
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6661051843533595e-05,
+      "loss": 0.0255,
+      "step": 14276
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.666058394160584e-05,
+      "loss": 0.0463,
+      "step": 14278
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.666011603967809e-05,
+      "loss": 0.0572,
+      "step": 14280
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6659648137750326e-05,
+      "loss": 0.0471,
+      "step": 14282
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.665918023582257e-05,
+      "loss": 0.0744,
+      "step": 14284
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.665871233389482e-05,
+      "loss": 0.0569,
+      "step": 14286
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6658244431967064e-05,
+      "loss": 0.0458,
+      "step": 14288
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.66577765300393e-05,
+      "loss": 0.0442,
+      "step": 14290
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.665730862811155e-05,
+      "loss": 0.0391,
+      "step": 14292
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6656840726183795e-05,
+      "loss": 0.0603,
+      "step": 14294
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.665637282425604e-05,
+      "loss": 0.0438,
+      "step": 14296
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.665590492232828e-05,
+      "loss": 0.0394,
+      "step": 14298
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6655437020400526e-05,
+      "loss": 0.0503,
+      "step": 14300
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.665496911847277e-05,
+      "loss": 0.0505,
+      "step": 14302
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.665450121654502e-05,
+      "loss": 0.0406,
+      "step": 14304
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.665403331461726e-05,
+      "loss": 0.0445,
+      "step": 14306
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.66535654126895e-05,
+      "loss": 0.0477,
+      "step": 14308
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.665309751076174e-05,
+      "loss": 0.0571,
+      "step": 14310
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6652629608833995e-05,
+      "loss": 0.0366,
+      "step": 14312
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6652161706906234e-05,
+      "loss": 0.0402,
+      "step": 14314
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.665169380497848e-05,
+      "loss": 0.0381,
+      "step": 14316
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.665122590305072e-05,
+      "loss": 0.0498,
+      "step": 14318
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.665075800112297e-05,
+      "loss": 0.0558,
+      "step": 14320
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.665029009919521e-05,
+      "loss": 0.0462,
+      "step": 14322
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.664982219726746e-05,
+      "loss": 0.0783,
+      "step": 14324
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6649354295339696e-05,
+      "loss": 0.061,
+      "step": 14326
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.664888639341194e-05,
+      "loss": 0.0652,
+      "step": 14328
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.664841849148419e-05,
+      "loss": 0.0561,
+      "step": 14330
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6647950589556434e-05,
+      "loss": 0.0597,
+      "step": 14332
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.664748268762867e-05,
+      "loss": 0.0651,
+      "step": 14334
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.664701478570092e-05,
+      "loss": 0.0486,
+      "step": 14336
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6646546883773165e-05,
+      "loss": 0.0443,
+      "step": 14338
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.664607898184541e-05,
+      "loss": 0.0463,
+      "step": 14340
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.664561107991765e-05,
+      "loss": 0.0457,
+      "step": 14342
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6645143177989895e-05,
+      "loss": 0.0446,
+      "step": 14344
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.664467527606214e-05,
+      "loss": 0.0688,
+      "step": 14346
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.664420737413439e-05,
+      "loss": 0.04,
+      "step": 14348
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6643739472206626e-05,
+      "loss": 0.048,
+      "step": 14350
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.664327157027887e-05,
+      "loss": 0.0438,
+      "step": 14352
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.664280366835112e-05,
+      "loss": 0.0598,
+      "step": 14354
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6642335766423364e-05,
+      "loss": 0.051,
+      "step": 14356
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.66418678644956e-05,
+      "loss": 0.0522,
+      "step": 14358
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.664139996256785e-05,
+      "loss": 0.0475,
+      "step": 14360
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.664093206064009e-05,
+      "loss": 0.0378,
+      "step": 14362
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.664046415871234e-05,
+      "loss": 0.0474,
+      "step": 14364
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.663999625678458e-05,
+      "loss": 0.0579,
+      "step": 14366
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6639528354856826e-05,
+      "loss": 0.0463,
+      "step": 14368
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6639060452929065e-05,
+      "loss": 0.0532,
+      "step": 14370
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.663859255100131e-05,
+      "loss": 0.0422,
+      "step": 14372
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.663812464907356e-05,
+      "loss": 0.0569,
+      "step": 14374
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.66376567471458e-05,
+      "loss": 0.0525,
+      "step": 14376
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.663718884521804e-05,
+      "loss": 0.0692,
+      "step": 14378
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.663672094329029e-05,
+      "loss": 0.0638,
+      "step": 14380
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6636253041362534e-05,
+      "loss": 0.0563,
+      "step": 14382
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.663578513943478e-05,
+      "loss": 0.0384,
+      "step": 14384
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.663531723750702e-05,
+      "loss": 0.0473,
+      "step": 14386
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6634849335579265e-05,
+      "loss": 0.0254,
+      "step": 14388
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.663438143365151e-05,
+      "loss": 0.0498,
+      "step": 14390
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.663391353172376e-05,
+      "loss": 0.0498,
+      "step": 14392
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6633445629795996e-05,
+      "loss": 0.0393,
+      "step": 14394
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.663297772786824e-05,
+      "loss": 0.0416,
+      "step": 14396
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.663250982594049e-05,
+      "loss": 0.0513,
+      "step": 14398
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6632041924012734e-05,
+      "loss": 0.0553,
+      "step": 14400
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.663157402208497e-05,
+      "loss": 0.0621,
+      "step": 14402
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.663110612015722e-05,
+      "loss": 0.0484,
+      "step": 14404
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.663063821822946e-05,
+      "loss": 0.0625,
+      "step": 14406
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.663017031630171e-05,
+      "loss": 0.0455,
+      "step": 14408
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.662970241437395e-05,
+      "loss": 0.0517,
+      "step": 14410
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6629234512446195e-05,
+      "loss": 0.053,
+      "step": 14412
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6628766610518435e-05,
+      "loss": 0.0519,
+      "step": 14414
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.662829870859069e-05,
+      "loss": 0.0597,
+      "step": 14416
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6627830806662926e-05,
+      "loss": 0.0446,
+      "step": 14418
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.662736290473517e-05,
+      "loss": 0.036,
+      "step": 14420
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.662689500280741e-05,
+      "loss": 0.0692,
+      "step": 14422
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.662642710087966e-05,
+      "loss": 0.0571,
+      "step": 14424
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.66259591989519e-05,
+      "loss": 0.064,
+      "step": 14426
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.662549129702415e-05,
+      "loss": 0.0612,
+      "step": 14428
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.662502339509639e-05,
+      "loss": 0.0411,
+      "step": 14430
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6624555493168634e-05,
+      "loss": 0.0502,
+      "step": 14432
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.662408759124088e-05,
+      "loss": 0.0466,
+      "step": 14434
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6623619689313126e-05,
+      "loss": 0.0592,
+      "step": 14436
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6623151787385365e-05,
+      "loss": 0.0625,
+      "step": 14438
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6622683885457604e-05,
+      "loss": 0.0395,
+      "step": 14440
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.662221598352986e-05,
+      "loss": 0.0376,
+      "step": 14442
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6621748081602096e-05,
+      "loss": 0.0531,
+      "step": 14444
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.662128017967434e-05,
+      "loss": 0.0522,
+      "step": 14446
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.662081227774658e-05,
+      "loss": 0.0555,
+      "step": 14448
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6620344375818834e-05,
+      "loss": 0.0436,
+      "step": 14450
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.661987647389107e-05,
+      "loss": 0.036,
+      "step": 14452
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.661940857196332e-05,
+      "loss": 0.0492,
+      "step": 14454
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.661894067003556e-05,
+      "loss": 0.0557,
+      "step": 14456
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6618472768107804e-05,
+      "loss": 0.0332,
+      "step": 14458
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.661800486618005e-05,
+      "loss": 0.0464,
+      "step": 14460
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6617536964252296e-05,
+      "loss": 0.0492,
+      "step": 14462
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6617069062324535e-05,
+      "loss": 0.0456,
+      "step": 14464
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.661660116039678e-05,
+      "loss": 0.0545,
+      "step": 14466
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.661613325846903e-05,
+      "loss": 0.0428,
+      "step": 14468
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.661566535654127e-05,
+      "loss": 0.0435,
+      "step": 14470
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.661519745461351e-05,
+      "loss": 0.0409,
+      "step": 14472
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.661472955268576e-05,
+      "loss": 0.0642,
+      "step": 14474
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6614261650758004e-05,
+      "loss": 0.066,
+      "step": 14476
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.661379374883025e-05,
+      "loss": 0.064,
+      "step": 14478
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.661332584690249e-05,
+      "loss": 0.0546,
+      "step": 14480
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6612857944974735e-05,
+      "loss": 0.0944,
+      "step": 14482
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.661239004304698e-05,
+      "loss": 0.0726,
+      "step": 14484
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6611922141119226e-05,
+      "loss": 0.0468,
+      "step": 14486
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6611454239191466e-05,
+      "loss": 0.0404,
+      "step": 14488
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.661098633726371e-05,
+      "loss": 0.056,
+      "step": 14490
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.661051843533595e-05,
+      "loss": 0.0486,
+      "step": 14492
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.66100505334082e-05,
+      "loss": 0.0692,
+      "step": 14494
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.660958263148044e-05,
+      "loss": 0.0455,
+      "step": 14496
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.660911472955269e-05,
+      "loss": 0.042,
+      "step": 14498
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.660864682762493e-05,
+      "loss": 0.0303,
+      "step": 14500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6608178925697173e-05,
+      "loss": 0.0518,
+      "step": 14502
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.660771102376942e-05,
+      "loss": 0.0586,
+      "step": 14504
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6607243121841665e-05,
+      "loss": 0.0516,
+      "step": 14506
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6606775219913904e-05,
+      "loss": 0.0457,
+      "step": 14508
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.660630731798615e-05,
+      "loss": 0.07,
+      "step": 14510
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6605839416058396e-05,
+      "loss": 0.0537,
+      "step": 14512
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.660537151413064e-05,
+      "loss": 0.041,
+      "step": 14514
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.660490361220288e-05,
+      "loss": 0.0478,
+      "step": 14516
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.660443571027513e-05,
+      "loss": 0.0361,
+      "step": 14518
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.660396780834737e-05,
+      "loss": 0.0689,
+      "step": 14520
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.660349990641962e-05,
+      "loss": 0.0567,
+      "step": 14522
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.660303200449186e-05,
+      "loss": 0.0406,
+      "step": 14524
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6602564102564104e-05,
+      "loss": 0.0564,
+      "step": 14526
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.660209620063635e-05,
+      "loss": 0.0621,
+      "step": 14528
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6601628298708596e-05,
+      "loss": 0.063,
+      "step": 14530
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6601160396780835e-05,
+      "loss": 0.058,
+      "step": 14532
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.660069249485308e-05,
+      "loss": 0.0404,
+      "step": 14534
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.660022459292532e-05,
+      "loss": 0.0621,
+      "step": 14536
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.659975669099757e-05,
+      "loss": 0.0534,
+      "step": 14538
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.659928878906981e-05,
+      "loss": 0.07,
+      "step": 14540
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.659882088714206e-05,
+      "loss": 0.043,
+      "step": 14542
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.65983529852143e-05,
+      "loss": 0.0656,
+      "step": 14544
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.659788508328655e-05,
+      "loss": 0.05,
+      "step": 14546
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.659741718135879e-05,
+      "loss": 0.063,
+      "step": 14548
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6596949279431035e-05,
+      "loss": 0.0544,
+      "step": 14550
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6596481377503274e-05,
+      "loss": 0.0418,
+      "step": 14552
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.659601347557552e-05,
+      "loss": 0.0418,
+      "step": 14554
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6595545573647766e-05,
+      "loss": 0.0735,
+      "step": 14556
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.659507767172001e-05,
+      "loss": 0.0715,
+      "step": 14558
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.659460976979225e-05,
+      "loss": 0.0477,
+      "step": 14560
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6594141867864497e-05,
+      "loss": 0.0454,
+      "step": 14562
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.659367396593674e-05,
+      "loss": 0.0515,
+      "step": 14564
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.659320606400899e-05,
+      "loss": 0.0504,
+      "step": 14566
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.659273816208123e-05,
+      "loss": 0.0641,
+      "step": 14568
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6592270260153473e-05,
+      "loss": 0.0577,
+      "step": 14570
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.659180235822572e-05,
+      "loss": 0.0744,
+      "step": 14572
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6591334456297965e-05,
+      "loss": 0.0584,
+      "step": 14574
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6590866554370204e-05,
+      "loss": 0.0404,
+      "step": 14576
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.659039865244245e-05,
+      "loss": 0.0616,
+      "step": 14578
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6589930750514696e-05,
+      "loss": 0.0401,
+      "step": 14580
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.658946284858694e-05,
+      "loss": 0.0437,
+      "step": 14582
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.658899494665918e-05,
+      "loss": 0.0641,
+      "step": 14584
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.658852704473143e-05,
+      "loss": 0.0548,
+      "step": 14586
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6588059142803666e-05,
+      "loss": 0.0503,
+      "step": 14588
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.658759124087592e-05,
+      "loss": 0.0467,
+      "step": 14590
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.658712333894816e-05,
+      "loss": 0.0517,
+      "step": 14592
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6586655437020404e-05,
+      "loss": 0.0374,
+      "step": 14594
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.658618753509264e-05,
+      "loss": 0.0573,
+      "step": 14596
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6585719633164896e-05,
+      "loss": 0.0446,
+      "step": 14598
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6585251731237135e-05,
+      "loss": 0.0477,
+      "step": 14600
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.658478382930938e-05,
+      "loss": 0.056,
+      "step": 14602
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.658431592738162e-05,
+      "loss": 0.0613,
+      "step": 14604
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6583848025453866e-05,
+      "loss": 0.0651,
+      "step": 14606
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.658338012352611e-05,
+      "loss": 0.0466,
+      "step": 14608
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.658291222159836e-05,
+      "loss": 0.0451,
+      "step": 14610
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.65824443196706e-05,
+      "loss": 0.0697,
+      "step": 14612
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.658197641774284e-05,
+      "loss": 0.0467,
+      "step": 14614
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.658150851581509e-05,
+      "loss": 0.0579,
+      "step": 14616
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6581040613887335e-05,
+      "loss": 0.0528,
+      "step": 14618
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6580572711959574e-05,
+      "loss": 0.0398,
+      "step": 14620
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.658010481003182e-05,
+      "loss": 0.0467,
+      "step": 14622
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6579636908104066e-05,
+      "loss": 0.0417,
+      "step": 14624
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.657916900617631e-05,
+      "loss": 0.054,
+      "step": 14626
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.657870110424855e-05,
+      "loss": 0.0376,
+      "step": 14628
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6578233202320797e-05,
+      "loss": 0.0616,
+      "step": 14630
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.657776530039304e-05,
+      "loss": 0.0564,
+      "step": 14632
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.657729739846529e-05,
+      "loss": 0.06,
+      "step": 14634
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.657682949653753e-05,
+      "loss": 0.0604,
+      "step": 14636
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6576361594609773e-05,
+      "loss": 0.0494,
+      "step": 14638
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.657589369268201e-05,
+      "loss": 0.0466,
+      "step": 14640
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6575425790754265e-05,
+      "loss": 0.0515,
+      "step": 14642
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6574957888826504e-05,
+      "loss": 0.0423,
+      "step": 14644
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.657448998689875e-05,
+      "loss": 0.0596,
+      "step": 14646
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.657402208497099e-05,
+      "loss": 0.0368,
+      "step": 14648
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6573554183043235e-05,
+      "loss": 0.0494,
+      "step": 14650
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.657308628111548e-05,
+      "loss": 0.0445,
+      "step": 14652
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.657261837918773e-05,
+      "loss": 0.0517,
+      "step": 14654
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6572150477259966e-05,
+      "loss": 0.0482,
+      "step": 14656
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.657168257533221e-05,
+      "loss": 0.0476,
+      "step": 14658
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.657121467340446e-05,
+      "loss": 0.0343,
+      "step": 14660
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6570746771476704e-05,
+      "loss": 0.0455,
+      "step": 14662
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.657027886954894e-05,
+      "loss": 0.0417,
+      "step": 14664
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.656981096762119e-05,
+      "loss": 0.0496,
+      "step": 14666
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6569343065693435e-05,
+      "loss": 0.0366,
+      "step": 14668
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.656887516376568e-05,
+      "loss": 0.0536,
+      "step": 14670
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.656840726183792e-05,
+      "loss": 0.0497,
+      "step": 14672
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6567939359910166e-05,
+      "loss": 0.0493,
+      "step": 14674
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.656747145798241e-05,
+      "loss": 0.0546,
+      "step": 14676
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.656700355605466e-05,
+      "loss": 0.0519,
+      "step": 14678
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.65665356541269e-05,
+      "loss": 0.0568,
+      "step": 14680
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.656606775219914e-05,
+      "loss": 0.0414,
+      "step": 14682
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.656559985027138e-05,
+      "loss": 0.0476,
+      "step": 14684
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6565131948343635e-05,
+      "loss": 0.0621,
+      "step": 14686
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6564664046415874e-05,
+      "loss": 0.0392,
+      "step": 14688
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.656419614448812e-05,
+      "loss": 0.0544,
+      "step": 14690
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.656372824256036e-05,
+      "loss": 0.0524,
+      "step": 14692
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6563260340632605e-05,
+      "loss": 0.0526,
+      "step": 14694
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.656279243870485e-05,
+      "loss": 0.0513,
+      "step": 14696
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.656232453677709e-05,
+      "loss": 0.0559,
+      "step": 14698
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6561856634849336e-05,
+      "loss": 0.0438,
+      "step": 14700
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.656138873292158e-05,
+      "loss": 0.055,
+      "step": 14702
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.656092083099383e-05,
+      "loss": 0.0487,
+      "step": 14704
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.656045292906607e-05,
+      "loss": 0.0491,
+      "step": 14706
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.655998502713831e-05,
+      "loss": 0.0372,
+      "step": 14708
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.655951712521056e-05,
+      "loss": 0.056,
+      "step": 14710
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6559049223282804e-05,
+      "loss": 0.054,
+      "step": 14712
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6558581321355044e-05,
+      "loss": 0.0396,
+      "step": 14714
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.655811341942729e-05,
+      "loss": 0.0446,
+      "step": 14716
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.655764551749953e-05,
+      "loss": 0.05,
+      "step": 14718
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.655717761557178e-05,
+      "loss": 0.0512,
+      "step": 14720
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.655670971364402e-05,
+      "loss": 0.0376,
+      "step": 14722
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6556241811716266e-05,
+      "loss": 0.0605,
+      "step": 14724
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6555773909788505e-05,
+      "loss": 0.0378,
+      "step": 14726
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.655530600786076e-05,
+      "loss": 0.0732,
+      "step": 14728
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6554838105933e-05,
+      "loss": 0.0635,
+      "step": 14730
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.655437020400524e-05,
+      "loss": 0.0476,
+      "step": 14732
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.655390230207748e-05,
+      "loss": 0.0382,
+      "step": 14734
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.655343440014973e-05,
+      "loss": 0.0371,
+      "step": 14736
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6552966498221974e-05,
+      "loss": 0.0559,
+      "step": 14738
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.655249859629422e-05,
+      "loss": 0.0363,
+      "step": 14740
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.655203069436646e-05,
+      "loss": 0.0508,
+      "step": 14742
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6551562792438705e-05,
+      "loss": 0.0466,
+      "step": 14744
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.655109489051095e-05,
+      "loss": 0.0489,
+      "step": 14746
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.65506269885832e-05,
+      "loss": 0.066,
+      "step": 14748
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6550159086655436e-05,
+      "loss": 0.0509,
+      "step": 14750
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.654969118472768e-05,
+      "loss": 0.0368,
+      "step": 14752
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.654922328279993e-05,
+      "loss": 0.0315,
+      "step": 14754
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6548755380872174e-05,
+      "loss": 0.0598,
+      "step": 14756
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.654828747894441e-05,
+      "loss": 0.0678,
+      "step": 14758
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.654781957701666e-05,
+      "loss": 0.0628,
+      "step": 14760
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6547351675088905e-05,
+      "loss": 0.0318,
+      "step": 14762
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.654688377316115e-05,
+      "loss": 0.0581,
+      "step": 14764
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.654641587123339e-05,
+      "loss": 0.0387,
+      "step": 14766
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6545947969305636e-05,
+      "loss": 0.0604,
+      "step": 14768
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6545480067377875e-05,
+      "loss": 0.0608,
+      "step": 14770
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.654501216545013e-05,
+      "loss": 0.0671,
+      "step": 14772
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.654454426352237e-05,
+      "loss": 0.0544,
+      "step": 14774
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.654407636159461e-05,
+      "loss": 0.0495,
+      "step": 14776
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.654360845966685e-05,
+      "loss": 0.0484,
+      "step": 14778
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6543140557739104e-05,
+      "loss": 0.0563,
+      "step": 14780
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6542672655811344e-05,
+      "loss": 0.0578,
+      "step": 14782
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.654220475388359e-05,
+      "loss": 0.0636,
+      "step": 14784
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.654173685195583e-05,
+      "loss": 0.0592,
+      "step": 14786
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6541268950028075e-05,
+      "loss": 0.0511,
+      "step": 14788
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.654080104810032e-05,
+      "loss": 0.0665,
+      "step": 14790
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6540333146172566e-05,
+      "loss": 0.0539,
+      "step": 14792
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6539865244244806e-05,
+      "loss": 0.059,
+      "step": 14794
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.653939734231705e-05,
+      "loss": 0.0649,
+      "step": 14796
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.65389294403893e-05,
+      "loss": 0.0448,
+      "step": 14798
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.653846153846154e-05,
+      "loss": 0.0811,
+      "step": 14800
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.653799363653378e-05,
+      "loss": 0.0506,
+      "step": 14802
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.653752573460603e-05,
+      "loss": 0.0577,
+      "step": 14804
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6537057832678274e-05,
+      "loss": 0.0445,
+      "step": 14806
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.653658993075052e-05,
+      "loss": 0.04,
+      "step": 14808
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.653612202882276e-05,
+      "loss": 0.0642,
+      "step": 14810
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6535654126895005e-05,
+      "loss": 0.0449,
+      "step": 14812
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6535186224967244e-05,
+      "loss": 0.0496,
+      "step": 14814
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.65347183230395e-05,
+      "loss": 0.0708,
+      "step": 14816
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6534250421111736e-05,
+      "loss": 0.0607,
+      "step": 14818
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.653378251918398e-05,
+      "loss": 0.0564,
+      "step": 14820
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.653331461725622e-05,
+      "loss": 0.0459,
+      "step": 14822
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6532846715328474e-05,
+      "loss": 0.0708,
+      "step": 14824
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.653237881340071e-05,
+      "loss": 0.0561,
+      "step": 14826
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.653191091147296e-05,
+      "loss": 0.0717,
+      "step": 14828
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.65314430095452e-05,
+      "loss": 0.0797,
+      "step": 14830
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6530975107617444e-05,
+      "loss": 0.0392,
+      "step": 14832
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.653050720568969e-05,
+      "loss": 0.0628,
+      "step": 14834
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6530039303761936e-05,
+      "loss": 0.0623,
+      "step": 14836
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6529571401834175e-05,
+      "loss": 0.0539,
+      "step": 14838
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.652910349990642e-05,
+      "loss": 0.0444,
+      "step": 14840
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.652863559797867e-05,
+      "loss": 0.0597,
+      "step": 14842
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.652816769605091e-05,
+      "loss": 0.046,
+      "step": 14844
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.652769979412315e-05,
+      "loss": 0.0736,
+      "step": 14846
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.65272318921954e-05,
+      "loss": 0.0652,
+      "step": 14848
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6526763990267644e-05,
+      "loss": 0.0419,
+      "step": 14850
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.652629608833989e-05,
+      "loss": 0.0417,
+      "step": 14852
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.652582818641213e-05,
+      "loss": 0.0475,
+      "step": 14854
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.6525360284484375e-05,
+      "loss": 0.0626,
+      "step": 14856
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.652489238255662e-05,
+      "loss": 0.0506,
+      "step": 14858
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.6524424480628866e-05,
+      "loss": 0.0565,
+      "step": 14860
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.6523956578701106e-05,
+      "loss": 0.0566,
+      "step": 14862
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.652348867677335e-05,
+      "loss": 0.0614,
+      "step": 14864
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.652302077484559e-05,
+      "loss": 0.049,
+      "step": 14866
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.652255287291784e-05,
+      "loss": 0.0626,
+      "step": 14868
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.652208497099008e-05,
+      "loss": 0.0448,
+      "step": 14870
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.652161706906233e-05,
+      "loss": 0.0537,
+      "step": 14872
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.652114916713457e-05,
+      "loss": 0.0431,
+      "step": 14874
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.652068126520682e-05,
+      "loss": 0.047,
+      "step": 14876
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.652021336327906e-05,
+      "loss": 0.0668,
+      "step": 14878
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.6519745461351305e-05,
+      "loss": 0.0505,
+      "step": 14880
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.6519277559423544e-05,
+      "loss": 0.0652,
+      "step": 14882
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.651880965749579e-05,
+      "loss": 0.0646,
+      "step": 14884
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.6518341755568036e-05,
+      "loss": 0.0597,
+      "step": 14886
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.651787385364028e-05,
+      "loss": 0.0604,
+      "step": 14888
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.651740595171252e-05,
+      "loss": 0.0598,
+      "step": 14890
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.651693804978477e-05,
+      "loss": 0.0547,
+      "step": 14892
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.651647014785701e-05,
+      "loss": 0.0596,
+      "step": 14894
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.651600224592926e-05,
+      "loss": 0.0751,
+      "step": 14896
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.65155343440015e-05,
+      "loss": 0.0511,
+      "step": 14898
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.6515066442073744e-05,
+      "loss": 0.0748,
+      "step": 14900
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.651459854014599e-05,
+      "loss": 0.0514,
+      "step": 14902
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.6514130638218236e-05,
+      "loss": 0.0444,
+      "step": 14904
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.6513662736290475e-05,
+      "loss": 0.0424,
+      "step": 14906
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.651319483436272e-05,
+      "loss": 0.0598,
+      "step": 14908
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.651272693243497e-05,
+      "loss": 0.0619,
+      "step": 14910
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.651225903050721e-05,
+      "loss": 0.0885,
+      "step": 14912
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.651179112857945e-05,
+      "loss": 0.0641,
+      "step": 14914
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.65113232266517e-05,
+      "loss": 0.0525,
+      "step": 14916
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.651085532472394e-05,
+      "loss": 0.0531,
+      "step": 14918
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.651038742279619e-05,
+      "loss": 0.0573,
+      "step": 14920
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.650991952086843e-05,
+      "loss": 0.0528,
+      "step": 14922
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.6509451618940675e-05,
+      "loss": 0.0504,
+      "step": 14924
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.6508983717012914e-05,
+      "loss": 0.0463,
+      "step": 14926
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.650851581508516e-05,
+      "loss": 0.0617,
+      "step": 14928
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.6508047913157406e-05,
+      "loss": 0.0479,
+      "step": 14930
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.650758001122965e-05,
+      "loss": 0.0539,
+      "step": 14932
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.650711210930189e-05,
+      "loss": 0.0714,
+      "step": 14934
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.6506644207374137e-05,
+      "loss": 0.0437,
+      "step": 14936
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.650617630544638e-05,
+      "loss": 0.0589,
+      "step": 14938
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.650570840351863e-05,
+      "loss": 0.0713,
+      "step": 14940
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.650524050159087e-05,
+      "loss": 0.055,
+      "step": 14942
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.650477259966311e-05,
+      "loss": 0.053,
+      "step": 14944
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.650430469773536e-05,
+      "loss": 0.0679,
+      "step": 14946
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.65038367958076e-05,
+      "loss": 0.066,
+      "step": 14948
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.6503368893879844e-05,
+      "loss": 0.0706,
+      "step": 14950
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.6502900991952083e-05,
+      "loss": 0.0536,
+      "step": 14952
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.6502433090024336e-05,
+      "loss": 0.0565,
+      "step": 14954
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.6501965188096575e-05,
+      "loss": 0.0618,
+      "step": 14956
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.650149728616882e-05,
+      "loss": 0.0537,
+      "step": 14958
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.650102938424106e-05,
+      "loss": 0.0347,
+      "step": 14960
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.6500561482313306e-05,
+      "loss": 0.0448,
+      "step": 14962
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.650009358038555e-05,
+      "loss": 0.0494,
+      "step": 14964
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.64996256784578e-05,
+      "loss": 0.0313,
+      "step": 14966
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.649915777653004e-05,
+      "loss": 0.0549,
+      "step": 14968
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.649868987460228e-05,
+      "loss": 0.0697,
+      "step": 14970
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.649822197267453e-05,
+      "loss": 0.0573,
+      "step": 14972
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.6497754070746775e-05,
+      "loss": 0.057,
+      "step": 14974
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.6497286168819014e-05,
+      "loss": 0.0701,
+      "step": 14976
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.649681826689126e-05,
+      "loss": 0.0424,
+      "step": 14978
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.6496350364963506e-05,
+      "loss": 0.0565,
+      "step": 14980
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.649588246303575e-05,
+      "loss": 0.0474,
+      "step": 14982
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.649541456110799e-05,
+      "loss": 0.0531,
+      "step": 14984
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.649494665918024e-05,
+      "loss": 0.0607,
+      "step": 14986
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.649447875725248e-05,
+      "loss": 0.0822,
+      "step": 14988
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.649401085532473e-05,
+      "loss": 0.0507,
+      "step": 14990
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.649354295339697e-05,
+      "loss": 0.0385,
+      "step": 14992
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.6493075051469214e-05,
+      "loss": 0.0401,
+      "step": 14994
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.649260714954145e-05,
+      "loss": 0.0452,
+      "step": 14996
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.6492139247613706e-05,
+      "loss": 0.0527,
+      "step": 14998
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.6491671345685945e-05,
+      "loss": 0.0805,
+      "step": 15000
+    },
+    {
+      "epoch": 2.11,
+      "eval_gen_len": 30.6379,
+      "eval_loss": 1.070984959602356,
+      "eval_meteor": 0.0449,
+      "eval_runtime": 14.582,
+      "eval_samples_per_second": 3.978,
+      "eval_steps_per_second": 0.549,
+      "step": 15000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.649120344375819e-05,
+      "loss": 0.0477,
+      "step": 15002
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.649073554183043e-05,
+      "loss": 0.0534,
+      "step": 15004
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.649026763990268e-05,
+      "loss": 0.0488,
+      "step": 15006
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.648979973797492e-05,
+      "loss": 0.0484,
+      "step": 15008
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.648933183604717e-05,
+      "loss": 0.0495,
+      "step": 15010
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.648886393411941e-05,
+      "loss": 0.0635,
+      "step": 15012
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.648839603219165e-05,
+      "loss": 0.0501,
+      "step": 15014
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.64879281302639e-05,
+      "loss": 0.0803,
+      "step": 15016
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.6487460228336144e-05,
+      "loss": 0.0396,
+      "step": 15018
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.6486992326408383e-05,
+      "loss": 0.0516,
+      "step": 15020
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.648652442448063e-05,
+      "loss": 0.0522,
+      "step": 15022
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.6486056522552875e-05,
+      "loss": 0.0452,
+      "step": 15024
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.648558862062512e-05,
+      "loss": 0.0603,
+      "step": 15026
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.648512071869736e-05,
+      "loss": 0.0586,
+      "step": 15028
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.6484652816769606e-05,
+      "loss": 0.0389,
+      "step": 15030
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.648418491484185e-05,
+      "loss": 0.0535,
+      "step": 15032
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.64837170129141e-05,
+      "loss": 0.0582,
+      "step": 15034
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.648324911098634e-05,
+      "loss": 0.0479,
+      "step": 15036
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.648278120905858e-05,
+      "loss": 0.057,
+      "step": 15038
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.648231330713083e-05,
+      "loss": 0.0658,
+      "step": 15040
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.6481845405203075e-05,
+      "loss": 0.0594,
+      "step": 15042
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.6481377503275314e-05,
+      "loss": 0.0462,
+      "step": 15044
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.648090960134756e-05,
+      "loss": 0.0506,
+      "step": 15046
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.64804416994198e-05,
+      "loss": 0.0614,
+      "step": 15048
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.647997379749205e-05,
+      "loss": 0.0835,
+      "step": 15050
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.647950589556429e-05,
+      "loss": 0.0583,
+      "step": 15052
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.647903799363654e-05,
+      "loss": 0.0607,
+      "step": 15054
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.6478570091708776e-05,
+      "loss": 0.0574,
+      "step": 15056
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.647810218978103e-05,
+      "loss": 0.0558,
+      "step": 15058
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.647763428785327e-05,
+      "loss": 0.0432,
+      "step": 15060
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.6477166385925514e-05,
+      "loss": 0.0598,
+      "step": 15062
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.647669848399775e-05,
+      "loss": 0.0699,
+      "step": 15064
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.647623058207e-05,
+      "loss": 0.051,
+      "step": 15066
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.6475762680142245e-05,
+      "loss": 0.0436,
+      "step": 15068
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.647529477821449e-05,
+      "loss": 0.0566,
+      "step": 15070
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.647482687628673e-05,
+      "loss": 0.0457,
+      "step": 15072
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.6474358974358976e-05,
+      "loss": 0.0543,
+      "step": 15074
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.647389107243122e-05,
+      "loss": 0.0551,
+      "step": 15076
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.647342317050347e-05,
+      "loss": 0.0436,
+      "step": 15078
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.647295526857571e-05,
+      "loss": 0.0405,
+      "step": 15080
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.647248736664795e-05,
+      "loss": 0.0448,
+      "step": 15082
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.64720194647202e-05,
+      "loss": 0.0623,
+      "step": 15084
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.6471551562792444e-05,
+      "loss": 0.0591,
+      "step": 15086
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.6471083660864684e-05,
+      "loss": 0.0362,
+      "step": 15088
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.647061575893693e-05,
+      "loss": 0.0531,
+      "step": 15090
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.6470147857009175e-05,
+      "loss": 0.0579,
+      "step": 15092
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.646967995508142e-05,
+      "loss": 0.045,
+      "step": 15094
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.646921205315366e-05,
+      "loss": 0.0718,
+      "step": 15096
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.6468744151225906e-05,
+      "loss": 0.0662,
+      "step": 15098
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.6468276249298145e-05,
+      "loss": 0.0498,
+      "step": 15100
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.64678083473704e-05,
+      "loss": 0.0631,
+      "step": 15102
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.646734044544264e-05,
+      "loss": 0.0572,
+      "step": 15104
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.646687254351488e-05,
+      "loss": 0.0628,
+      "step": 15106
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.646640464158712e-05,
+      "loss": 0.045,
+      "step": 15108
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.646593673965937e-05,
+      "loss": 0.0361,
+      "step": 15110
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.6465468837731614e-05,
+      "loss": 0.0522,
+      "step": 15112
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.646500093580386e-05,
+      "loss": 0.0517,
+      "step": 15114
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.64645330338761e-05,
+      "loss": 0.0408,
+      "step": 15116
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.6464065131948345e-05,
+      "loss": 0.0477,
+      "step": 15118
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.646359723002059e-05,
+      "loss": 0.0487,
+      "step": 15120
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.646312932809284e-05,
+      "loss": 0.0528,
+      "step": 15122
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.6462661426165076e-05,
+      "loss": 0.0603,
+      "step": 15124
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.646219352423732e-05,
+      "loss": 0.0583,
+      "step": 15126
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.646172562230957e-05,
+      "loss": 0.0488,
+      "step": 15128
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.6461257720381814e-05,
+      "loss": 0.06,
+      "step": 15130
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.646078981845405e-05,
+      "loss": 0.0546,
+      "step": 15132
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.64603219165263e-05,
+      "loss": 0.0397,
+      "step": 15134
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.6459854014598545e-05,
+      "loss": 0.063,
+      "step": 15136
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.645938611267079e-05,
+      "loss": 0.0425,
+      "step": 15138
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.645891821074303e-05,
+      "loss": 0.0743,
+      "step": 15140
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.6458450308815276e-05,
+      "loss": 0.0632,
+      "step": 15142
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.6457982406887515e-05,
+      "loss": 0.054,
+      "step": 15144
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.645751450495977e-05,
+      "loss": 0.0592,
+      "step": 15146
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.645704660303201e-05,
+      "loss": 0.0652,
+      "step": 15148
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.645657870110425e-05,
+      "loss": 0.0531,
+      "step": 15150
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.645611079917649e-05,
+      "loss": 0.0572,
+      "step": 15152
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.6455642897248744e-05,
+      "loss": 0.0478,
+      "step": 15154
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.6455174995320984e-05,
+      "loss": 0.0781,
+      "step": 15156
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.645470709339323e-05,
+      "loss": 0.0587,
+      "step": 15158
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.645423919146547e-05,
+      "loss": 0.0478,
+      "step": 15160
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.6453771289537714e-05,
+      "loss": 0.0492,
+      "step": 15162
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.645330338760996e-05,
+      "loss": 0.0515,
+      "step": 15164
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.6452835485682206e-05,
+      "loss": 0.0323,
+      "step": 15166
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.6452367583754445e-05,
+      "loss": 0.0474,
+      "step": 15168
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.645189968182669e-05,
+      "loss": 0.0621,
+      "step": 15170
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.645143177989894e-05,
+      "loss": 0.0607,
+      "step": 15172
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.645096387797118e-05,
+      "loss": 0.062,
+      "step": 15174
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.645049597604342e-05,
+      "loss": 0.0483,
+      "step": 15176
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.645002807411567e-05,
+      "loss": 0.0559,
+      "step": 15178
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.6449560172187914e-05,
+      "loss": 0.0626,
+      "step": 15180
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.644909227026016e-05,
+      "loss": 0.0545,
+      "step": 15182
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.64486243683324e-05,
+      "loss": 0.0888,
+      "step": 15184
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.6448156466404645e-05,
+      "loss": 0.1247,
+      "step": 15186
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.644768856447689e-05,
+      "loss": 0.046,
+      "step": 15188
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.644722066254914e-05,
+      "loss": 0.0655,
+      "step": 15190
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.6446752760621376e-05,
+      "loss": 0.062,
+      "step": 15192
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.644628485869362e-05,
+      "loss": 0.047,
+      "step": 15194
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.644581695676586e-05,
+      "loss": 0.0581,
+      "step": 15196
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.644534905483811e-05,
+      "loss": 0.058,
+      "step": 15198
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.644488115291035e-05,
+      "loss": 0.0498,
+      "step": 15200
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.644441325098259e-05,
+      "loss": 0.0443,
+      "step": 15202
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.644394534905484e-05,
+      "loss": 0.0454,
+      "step": 15204
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.6443477447127084e-05,
+      "loss": 0.0568,
+      "step": 15206
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.644300954519933e-05,
+      "loss": 0.053,
+      "step": 15208
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.644254164327157e-05,
+      "loss": 0.059,
+      "step": 15210
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.6442073741343815e-05,
+      "loss": 0.062,
+      "step": 15212
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.644160583941606e-05,
+      "loss": 0.0483,
+      "step": 15214
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.644113793748831e-05,
+      "loss": 0.0564,
+      "step": 15216
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.6440670035560546e-05,
+      "loss": 0.0625,
+      "step": 15218
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.644020213363279e-05,
+      "loss": 0.0437,
+      "step": 15220
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.643973423170504e-05,
+      "loss": 0.0526,
+      "step": 15222
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.6439266329777284e-05,
+      "loss": 0.0499,
+      "step": 15224
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.643879842784952e-05,
+      "loss": 0.0658,
+      "step": 15226
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.643833052592177e-05,
+      "loss": 0.0556,
+      "step": 15228
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.643786262399401e-05,
+      "loss": 0.0673,
+      "step": 15230
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.643739472206626e-05,
+      "loss": 0.0553,
+      "step": 15232
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.64369268201385e-05,
+      "loss": 0.0614,
+      "step": 15234
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.6436458918210745e-05,
+      "loss": 0.0884,
+      "step": 15236
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.6435991016282985e-05,
+      "loss": 0.0455,
+      "step": 15238
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.643552311435523e-05,
+      "loss": 0.0463,
+      "step": 15240
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.6435055212427476e-05,
+      "loss": 0.0636,
+      "step": 15242
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.643458731049972e-05,
+      "loss": 0.0556,
+      "step": 15244
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.643411940857196e-05,
+      "loss": 0.0569,
+      "step": 15246
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.643365150664421e-05,
+      "loss": 0.0563,
+      "step": 15248
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.643318360471645e-05,
+      "loss": 0.059,
+      "step": 15250
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.64327157027887e-05,
+      "loss": 0.0442,
+      "step": 15252
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.643224780086094e-05,
+      "loss": 0.0612,
+      "step": 15254
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.6431779898933184e-05,
+      "loss": 0.0571,
+      "step": 15256
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.643131199700543e-05,
+      "loss": 0.0552,
+      "step": 15258
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.6430844095077676e-05,
+      "loss": 0.0851,
+      "step": 15260
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.6430376193149915e-05,
+      "loss": 0.0524,
+      "step": 15262
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.642990829122216e-05,
+      "loss": 0.0518,
+      "step": 15264
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.642944038929441e-05,
+      "loss": 0.0576,
+      "step": 15266
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.642897248736665e-05,
+      "loss": 0.039,
+      "step": 15268
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.642850458543889e-05,
+      "loss": 0.046,
+      "step": 15270
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.642803668351114e-05,
+      "loss": 0.0452,
+      "step": 15272
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.642756878158338e-05,
+      "loss": 0.0541,
+      "step": 15274
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.642710087965563e-05,
+      "loss": 0.094,
+      "step": 15276
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.642663297772787e-05,
+      "loss": 0.054,
+      "step": 15278
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.6426165075800115e-05,
+      "loss": 0.0606,
+      "step": 15280
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.6425697173872354e-05,
+      "loss": 0.0469,
+      "step": 15282
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.642522927194461e-05,
+      "loss": 0.0518,
+      "step": 15284
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.6424761370016846e-05,
+      "loss": 0.0563,
+      "step": 15286
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.642429346808909e-05,
+      "loss": 0.0486,
+      "step": 15288
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.642382556616133e-05,
+      "loss": 0.0513,
+      "step": 15290
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.642335766423358e-05,
+      "loss": 0.0571,
+      "step": 15292
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.642288976230582e-05,
+      "loss": 0.0572,
+      "step": 15294
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.642242186037807e-05,
+      "loss": 0.0496,
+      "step": 15296
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.642195395845031e-05,
+      "loss": 0.0754,
+      "step": 15298
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.6421486056522554e-05,
+      "loss": 0.0485,
+      "step": 15300
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.64210181545948e-05,
+      "loss": 0.0543,
+      "step": 15302
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.6420550252667045e-05,
+      "loss": 0.0426,
+      "step": 15304
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.6420082350739285e-05,
+      "loss": 0.066,
+      "step": 15306
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.641961444881153e-05,
+      "loss": 0.0548,
+      "step": 15308
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.6419146546883776e-05,
+      "loss": 0.0523,
+      "step": 15310
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.641867864495602e-05,
+      "loss": 0.0636,
+      "step": 15312
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.641821074302826e-05,
+      "loss": 0.058,
+      "step": 15314
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.641774284110051e-05,
+      "loss": 0.0528,
+      "step": 15316
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.641727493917275e-05,
+      "loss": 0.0574,
+      "step": 15318
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.6416807037245e-05,
+      "loss": 0.0475,
+      "step": 15320
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.641633913531724e-05,
+      "loss": 0.0413,
+      "step": 15322
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.6415871233389484e-05,
+      "loss": 0.0584,
+      "step": 15324
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.6415403331461723e-05,
+      "loss": 0.0588,
+      "step": 15326
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.6414935429533976e-05,
+      "loss": 0.05,
+      "step": 15328
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.6414467527606215e-05,
+      "loss": 0.0552,
+      "step": 15330
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.641399962567846e-05,
+      "loss": 0.0602,
+      "step": 15332
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.64135317237507e-05,
+      "loss": 0.0541,
+      "step": 15334
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.641306382182295e-05,
+      "loss": 0.0666,
+      "step": 15336
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.641259591989519e-05,
+      "loss": 0.0538,
+      "step": 15338
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.641212801796744e-05,
+      "loss": 0.0529,
+      "step": 15340
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.641166011603968e-05,
+      "loss": 0.0535,
+      "step": 15342
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.641119221411192e-05,
+      "loss": 0.0572,
+      "step": 15344
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.641072431218417e-05,
+      "loss": 0.0448,
+      "step": 15346
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.6410256410256415e-05,
+      "loss": 0.0659,
+      "step": 15348
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.6409788508328654e-05,
+      "loss": 0.0585,
+      "step": 15350
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.64093206064009e-05,
+      "loss": 0.0567,
+      "step": 15352
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6408852704473146e-05,
+      "loss": 0.0545,
+      "step": 15354
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.640838480254539e-05,
+      "loss": 0.0571,
+      "step": 15356
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.640791690061763e-05,
+      "loss": 0.0446,
+      "step": 15358
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.640744899868988e-05,
+      "loss": 0.0552,
+      "step": 15360
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.640698109676212e-05,
+      "loss": 0.0586,
+      "step": 15362
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.640651319483437e-05,
+      "loss": 0.0646,
+      "step": 15364
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.640604529290661e-05,
+      "loss": 0.0577,
+      "step": 15366
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6405577390978854e-05,
+      "loss": 0.0524,
+      "step": 15368
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.64051094890511e-05,
+      "loss": 0.0343,
+      "step": 15370
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6404641587123345e-05,
+      "loss": 0.0588,
+      "step": 15372
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6404173685195585e-05,
+      "loss": 0.0514,
+      "step": 15374
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.640370578326783e-05,
+      "loss": 0.0534,
+      "step": 15376
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.640323788134007e-05,
+      "loss": 0.0449,
+      "step": 15378
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.640276997941232e-05,
+      "loss": 0.0473,
+      "step": 15380
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.640230207748456e-05,
+      "loss": 0.0454,
+      "step": 15382
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.640183417555681e-05,
+      "loss": 0.0465,
+      "step": 15384
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6401366273629047e-05,
+      "loss": 0.0691,
+      "step": 15386
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.640089837170129e-05,
+      "loss": 0.0444,
+      "step": 15388
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.640043046977354e-05,
+      "loss": 0.0527,
+      "step": 15390
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6399962567845784e-05,
+      "loss": 0.047,
+      "step": 15392
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6399494665918023e-05,
+      "loss": 0.0557,
+      "step": 15394
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.639902676399027e-05,
+      "loss": 0.0504,
+      "step": 15396
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6398558862062515e-05,
+      "loss": 0.0716,
+      "step": 15398
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.639809096013476e-05,
+      "loss": 0.0585,
+      "step": 15400
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6397623058207e-05,
+      "loss": 0.0546,
+      "step": 15402
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6397155156279246e-05,
+      "loss": 0.0537,
+      "step": 15404
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.639668725435149e-05,
+      "loss": 0.0489,
+      "step": 15406
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.639621935242374e-05,
+      "loss": 0.0468,
+      "step": 15408
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.639575145049598e-05,
+      "loss": 0.0536,
+      "step": 15410
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.639528354856822e-05,
+      "loss": 0.0616,
+      "step": 15412
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.639481564664047e-05,
+      "loss": 0.0497,
+      "step": 15414
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6394347744712715e-05,
+      "loss": 0.0816,
+      "step": 15416
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.6393879842784954e-05,
+      "loss": 0.0655,
+      "step": 15418
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.63934119408572e-05,
+      "loss": 0.0619,
+      "step": 15420
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.639294403892944e-05,
+      "loss": 0.0545,
+      "step": 15422
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.639247613700169e-05,
+      "loss": 0.0582,
+      "step": 15424
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.639200823507393e-05,
+      "loss": 0.0548,
+      "step": 15426
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.639154033314618e-05,
+      "loss": 0.057,
+      "step": 15428
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6391072431218416e-05,
+      "loss": 0.0554,
+      "step": 15430
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.639060452929067e-05,
+      "loss": 0.0669,
+      "step": 15432
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.639013662736291e-05,
+      "loss": 0.0555,
+      "step": 15434
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6389668725435154e-05,
+      "loss": 0.0639,
+      "step": 15436
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.638920082350739e-05,
+      "loss": 0.0716,
+      "step": 15438
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.638873292157964e-05,
+      "loss": 0.0643,
+      "step": 15440
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6388265019651885e-05,
+      "loss": 0.0448,
+      "step": 15442
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.638779711772413e-05,
+      "loss": 0.0432,
+      "step": 15444
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.638732921579637e-05,
+      "loss": 0.0594,
+      "step": 15446
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6386861313868616e-05,
+      "loss": 0.0545,
+      "step": 15448
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.638639341194086e-05,
+      "loss": 0.0535,
+      "step": 15450
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.63859255100131e-05,
+      "loss": 0.0547,
+      "step": 15452
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6385457608085347e-05,
+      "loss": 0.0533,
+      "step": 15454
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6384989706157586e-05,
+      "loss": 0.0608,
+      "step": 15456
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.638452180422984e-05,
+      "loss": 0.0604,
+      "step": 15458
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.638405390230208e-05,
+      "loss": 0.0497,
+      "step": 15460
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6383586000374323e-05,
+      "loss": 0.0541,
+      "step": 15462
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.638311809844656e-05,
+      "loss": 0.0518,
+      "step": 15464
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6382650196518815e-05,
+      "loss": 0.05,
+      "step": 15466
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6382182294591054e-05,
+      "loss": 0.0651,
+      "step": 15468
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.63817143926633e-05,
+      "loss": 0.0633,
+      "step": 15470
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.638124649073554e-05,
+      "loss": 0.0395,
+      "step": 15472
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6380778588807785e-05,
+      "loss": 0.0576,
+      "step": 15474
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.638031068688003e-05,
+      "loss": 0.0611,
+      "step": 15476
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.637984278495228e-05,
+      "loss": 0.0513,
+      "step": 15478
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6379374883024516e-05,
+      "loss": 0.0625,
+      "step": 15480
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.637890698109676e-05,
+      "loss": 0.0531,
+      "step": 15482
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.637843907916901e-05,
+      "loss": 0.063,
+      "step": 15484
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6377971177241254e-05,
+      "loss": 0.0545,
+      "step": 15486
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.637750327531349e-05,
+      "loss": 0.0325,
+      "step": 15488
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.637703537338574e-05,
+      "loss": 0.056,
+      "step": 15490
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.6376567471457985e-05,
+      "loss": 0.04,
+      "step": 15492
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.637609956953023e-05,
+      "loss": 0.0766,
+      "step": 15494
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.637563166760247e-05,
+      "loss": 0.0583,
+      "step": 15496
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.6375163765674716e-05,
+      "loss": 0.0625,
+      "step": 15498
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.637469586374696e-05,
+      "loss": 0.0655,
+      "step": 15500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.637422796181921e-05,
+      "loss": 0.039,
+      "step": 15502
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.637376005989145e-05,
+      "loss": 0.0538,
+      "step": 15504
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.637329215796369e-05,
+      "loss": 0.0585,
+      "step": 15506
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.637282425603593e-05,
+      "loss": 0.0564,
+      "step": 15508
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.6372356354108185e-05,
+      "loss": 0.0359,
+      "step": 15510
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.6371888452180424e-05,
+      "loss": 0.0466,
+      "step": 15512
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.637142055025267e-05,
+      "loss": 0.0471,
+      "step": 15514
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.637095264832491e-05,
+      "loss": 0.0497,
+      "step": 15516
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.6370484746397155e-05,
+      "loss": 0.0497,
+      "step": 15518
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.63700168444694e-05,
+      "loss": 0.0579,
+      "step": 15520
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.6369548942541647e-05,
+      "loss": 0.0632,
+      "step": 15522
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.6369081040613886e-05,
+      "loss": 0.0546,
+      "step": 15524
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.636861313868613e-05,
+      "loss": 0.0536,
+      "step": 15526
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.636814523675838e-05,
+      "loss": 0.0689,
+      "step": 15528
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.6367677334830623e-05,
+      "loss": 0.0513,
+      "step": 15530
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.636720943290286e-05,
+      "loss": 0.0614,
+      "step": 15532
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.636674153097511e-05,
+      "loss": 0.0648,
+      "step": 15534
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.6366273629047354e-05,
+      "loss": 0.0535,
+      "step": 15536
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.63658057271196e-05,
+      "loss": 0.0775,
+      "step": 15538
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.636533782519184e-05,
+      "loss": 0.0444,
+      "step": 15540
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.6364869923264085e-05,
+      "loss": 0.0726,
+      "step": 15542
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.636440202133633e-05,
+      "loss": 0.0636,
+      "step": 15544
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.636393411940858e-05,
+      "loss": 0.0536,
+      "step": 15546
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.6363466217480816e-05,
+      "loss": 0.0899,
+      "step": 15548
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.636299831555306e-05,
+      "loss": 0.0589,
+      "step": 15550
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.63625304136253e-05,
+      "loss": 0.0591,
+      "step": 15552
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.6362062511697554e-05,
+      "loss": 0.0833,
+      "step": 15554
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.636159460976979e-05,
+      "loss": 0.056,
+      "step": 15556
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.636112670784204e-05,
+      "loss": 0.0591,
+      "step": 15558
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.636065880591428e-05,
+      "loss": 0.0595,
+      "step": 15560
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.636019090398653e-05,
+      "loss": 0.0524,
+      "step": 15562
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.635972300205877e-05,
+      "loss": 0.0499,
+      "step": 15564
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.6359255100131016e-05,
+      "loss": 0.0536,
+      "step": 15566
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.6358787198203255e-05,
+      "loss": 0.0428,
+      "step": 15568
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.63583192962755e-05,
+      "loss": 0.0692,
+      "step": 15570
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.635785139434775e-05,
+      "loss": 0.0556,
+      "step": 15572
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.635738349241999e-05,
+      "loss": 0.0433,
+      "step": 15574
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.635691559049223e-05,
+      "loss": 0.0503,
+      "step": 15576
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.635644768856448e-05,
+      "loss": 0.049,
+      "step": 15578
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.6355979786636724e-05,
+      "loss": 0.0619,
+      "step": 15580
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.635551188470897e-05,
+      "loss": 0.0455,
+      "step": 15582
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.635504398278121e-05,
+      "loss": 0.0665,
+      "step": 15584
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.6354576080853455e-05,
+      "loss": 0.0644,
+      "step": 15586
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.63541081789257e-05,
+      "loss": 0.0636,
+      "step": 15588
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.6353640276997947e-05,
+      "loss": 0.0647,
+      "step": 15590
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.6353172375070186e-05,
+      "loss": 0.063,
+      "step": 15592
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.635270447314243e-05,
+      "loss": 0.1058,
+      "step": 15594
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.635223657121468e-05,
+      "loss": 0.0463,
+      "step": 15596
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.6351768669286923e-05,
+      "loss": 0.0686,
+      "step": 15598
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.635130076735916e-05,
+      "loss": 0.0621,
+      "step": 15600
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.635083286543141e-05,
+      "loss": 0.0481,
+      "step": 15602
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.635036496350365e-05,
+      "loss": 0.0616,
+      "step": 15604
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.63498970615759e-05,
+      "loss": 0.0559,
+      "step": 15606
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.634942915964814e-05,
+      "loss": 0.0528,
+      "step": 15608
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.6348961257720385e-05,
+      "loss": 0.0511,
+      "step": 15610
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.6348493355792625e-05,
+      "loss": 0.0385,
+      "step": 15612
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.634802545386488e-05,
+      "loss": 0.0599,
+      "step": 15614
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.6347557551937116e-05,
+      "loss": 0.0457,
+      "step": 15616
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.634708965000936e-05,
+      "loss": 0.0578,
+      "step": 15618
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.63466217480816e-05,
+      "loss": 0.05,
+      "step": 15620
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.634615384615385e-05,
+      "loss": 0.0419,
+      "step": 15622
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.634568594422609e-05,
+      "loss": 0.082,
+      "step": 15624
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.634521804229834e-05,
+      "loss": 0.0633,
+      "step": 15626
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.634475014037058e-05,
+      "loss": 0.0507,
+      "step": 15628
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.6344282238442824e-05,
+      "loss": 0.0467,
+      "step": 15630
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.634381433651507e-05,
+      "loss": 0.0654,
+      "step": 15632
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.6343346434587316e-05,
+      "loss": 0.0531,
+      "step": 15634
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.6342878532659555e-05,
+      "loss": 0.0542,
+      "step": 15636
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.63424106307318e-05,
+      "loss": 0.0747,
+      "step": 15638
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.634194272880405e-05,
+      "loss": 0.0491,
+      "step": 15640
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.634147482687629e-05,
+      "loss": 0.0477,
+      "step": 15642
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.634100692494853e-05,
+      "loss": 0.0781,
+      "step": 15644
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.634053902302078e-05,
+      "loss": 0.0875,
+      "step": 15646
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.6340071121093024e-05,
+      "loss": 0.0636,
+      "step": 15648
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.633960321916527e-05,
+      "loss": 0.0407,
+      "step": 15650
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.633913531723751e-05,
+      "loss": 0.0512,
+      "step": 15652
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.6338667415309755e-05,
+      "loss": 0.0644,
+      "step": 15654
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.6338199513381994e-05,
+      "loss": 0.0408,
+      "step": 15656
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.633773161145425e-05,
+      "loss": 0.0575,
+      "step": 15658
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.6337263709526486e-05,
+      "loss": 0.0611,
+      "step": 15660
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.633679580759873e-05,
+      "loss": 0.0643,
+      "step": 15662
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.633632790567097e-05,
+      "loss": 0.1125,
+      "step": 15664
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.633586000374322e-05,
+      "loss": 0.0515,
+      "step": 15666
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.633539210181546e-05,
+      "loss": 0.0465,
+      "step": 15668
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.633492419988771e-05,
+      "loss": 0.073,
+      "step": 15670
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.633445629795995e-05,
+      "loss": 0.0571,
+      "step": 15672
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.6333988396032194e-05,
+      "loss": 0.0672,
+      "step": 15674
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.633352049410444e-05,
+      "loss": 0.0739,
+      "step": 15676
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.6333052592176685e-05,
+      "loss": 0.0513,
+      "step": 15678
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.6332584690248925e-05,
+      "loss": 0.0375,
+      "step": 15680
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.633211678832117e-05,
+      "loss": 0.0816,
+      "step": 15682
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.6331648886393416e-05,
+      "loss": 0.0483,
+      "step": 15684
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.633118098446566e-05,
+      "loss": 0.0672,
+      "step": 15686
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.63307130825379e-05,
+      "loss": 0.0518,
+      "step": 15688
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.633024518061015e-05,
+      "loss": 0.0551,
+      "step": 15690
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.632977727868239e-05,
+      "loss": 0.053,
+      "step": 15692
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.632930937675464e-05,
+      "loss": 0.0707,
+      "step": 15694
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.632884147482688e-05,
+      "loss": 0.0538,
+      "step": 15696
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.6328373572899124e-05,
+      "loss": 0.0524,
+      "step": 15698
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.632790567097136e-05,
+      "loss": 0.0526,
+      "step": 15700
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.6327437769043616e-05,
+      "loss": 0.0603,
+      "step": 15702
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.6326969867115855e-05,
+      "loss": 0.0508,
+      "step": 15704
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.6326501965188094e-05,
+      "loss": 0.0504,
+      "step": 15706
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.632603406326034e-05,
+      "loss": 0.0759,
+      "step": 15708
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.6325566161332586e-05,
+      "loss": 0.0701,
+      "step": 15710
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.632509825940483e-05,
+      "loss": 0.0525,
+      "step": 15712
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.632463035747707e-05,
+      "loss": 0.0752,
+      "step": 15714
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.632416245554932e-05,
+      "loss": 0.0466,
+      "step": 15716
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.632369455362156e-05,
+      "loss": 0.0748,
+      "step": 15718
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.632322665169381e-05,
+      "loss": 0.0522,
+      "step": 15720
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.632275874976605e-05,
+      "loss": 0.0591,
+      "step": 15722
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.6322290847838294e-05,
+      "loss": 0.0554,
+      "step": 15724
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.632182294591054e-05,
+      "loss": 0.069,
+      "step": 15726
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.6321355043982786e-05,
+      "loss": 0.043,
+      "step": 15728
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.6320887142055025e-05,
+      "loss": 0.0566,
+      "step": 15730
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.632041924012727e-05,
+      "loss": 0.0616,
+      "step": 15732
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.631995133819951e-05,
+      "loss": 0.0504,
+      "step": 15734
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.631948343627176e-05,
+      "loss": 0.038,
+      "step": 15736
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.6319015534344e-05,
+      "loss": 0.0668,
+      "step": 15738
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.631854763241625e-05,
+      "loss": 0.0513,
+      "step": 15740
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.631807973048849e-05,
+      "loss": 0.0499,
+      "step": 15742
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.631761182856074e-05,
+      "loss": 0.0645,
+      "step": 15744
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.631714392663298e-05,
+      "loss": 0.0597,
+      "step": 15746
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.6316676024705225e-05,
+      "loss": 0.0603,
+      "step": 15748
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.6316208122777464e-05,
+      "loss": 0.0607,
+      "step": 15750
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.631574022084971e-05,
+      "loss": 0.072,
+      "step": 15752
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.6315272318921956e-05,
+      "loss": 0.0732,
+      "step": 15754
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.63148044169942e-05,
+      "loss": 0.0526,
+      "step": 15756
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.631433651506644e-05,
+      "loss": 0.06,
+      "step": 15758
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.6313868613138686e-05,
+      "loss": 0.0551,
+      "step": 15760
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.631340071121093e-05,
+      "loss": 0.0413,
+      "step": 15762
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.631293280928318e-05,
+      "loss": 0.0723,
+      "step": 15764
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.631246490735542e-05,
+      "loss": 0.0541,
+      "step": 15766
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.631199700542766e-05,
+      "loss": 0.0587,
+      "step": 15768
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.631152910349991e-05,
+      "loss": 0.0835,
+      "step": 15770
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.6311061201572155e-05,
+      "loss": 0.0397,
+      "step": 15772
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.6310593299644394e-05,
+      "loss": 0.073,
+      "step": 15774
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.631012539771664e-05,
+      "loss": 0.066,
+      "step": 15776
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.6309657495788886e-05,
+      "loss": 0.0558,
+      "step": 15778
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.630918959386113e-05,
+      "loss": 0.0481,
+      "step": 15780
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.630872169193337e-05,
+      "loss": 0.0479,
+      "step": 15782
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.630825379000562e-05,
+      "loss": 0.0566,
+      "step": 15784
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.6307785888077856e-05,
+      "loss": 0.0748,
+      "step": 15786
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.630731798615011e-05,
+      "loss": 0.0521,
+      "step": 15788
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.630685008422235e-05,
+      "loss": 0.07,
+      "step": 15790
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.6306382182294594e-05,
+      "loss": 0.0689,
+      "step": 15792
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.630591428036683e-05,
+      "loss": 0.0412,
+      "step": 15794
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.6305446378439086e-05,
+      "loss": 0.0472,
+      "step": 15796
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.6304978476511325e-05,
+      "loss": 0.0605,
+      "step": 15798
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.630451057458357e-05,
+      "loss": 0.0562,
+      "step": 15800
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.630404267265581e-05,
+      "loss": 0.0456,
+      "step": 15802
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.6303574770728056e-05,
+      "loss": 0.0515,
+      "step": 15804
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.63031068688003e-05,
+      "loss": 0.0508,
+      "step": 15806
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.630263896687255e-05,
+      "loss": 0.0717,
+      "step": 15808
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.630217106494479e-05,
+      "loss": 0.0532,
+      "step": 15810
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.630170316301703e-05,
+      "loss": 0.0498,
+      "step": 15812
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.630123526108928e-05,
+      "loss": 0.0589,
+      "step": 15814
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.6300767359161525e-05,
+      "loss": 0.0525,
+      "step": 15816
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.6300299457233764e-05,
+      "loss": 0.0576,
+      "step": 15818
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.629983155530601e-05,
+      "loss": 0.0634,
+      "step": 15820
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.6299363653378256e-05,
+      "loss": 0.0351,
+      "step": 15822
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.62988957514505e-05,
+      "loss": 0.0523,
+      "step": 15824
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.629842784952274e-05,
+      "loss": 0.0544,
+      "step": 15826
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.6297959947594986e-05,
+      "loss": 0.0492,
+      "step": 15828
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.6297492045667226e-05,
+      "loss": 0.0475,
+      "step": 15830
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.629702414373948e-05,
+      "loss": 0.043,
+      "step": 15832
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.629655624181172e-05,
+      "loss": 0.0636,
+      "step": 15834
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.629608833988396e-05,
+      "loss": 0.0614,
+      "step": 15836
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.62956204379562e-05,
+      "loss": 0.0643,
+      "step": 15838
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.6295152536028455e-05,
+      "loss": 0.0549,
+      "step": 15840
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.6294684634100694e-05,
+      "loss": 0.0648,
+      "step": 15842
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.629421673217294e-05,
+      "loss": 0.0555,
+      "step": 15844
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.629374883024518e-05,
+      "loss": 0.047,
+      "step": 15846
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.6293280928317425e-05,
+      "loss": 0.0622,
+      "step": 15848
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.629281302638967e-05,
+      "loss": 0.0447,
+      "step": 15850
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.629234512446192e-05,
+      "loss": 0.0678,
+      "step": 15852
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.6291877222534156e-05,
+      "loss": 0.0686,
+      "step": 15854
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.62914093206064e-05,
+      "loss": 0.0524,
+      "step": 15856
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.629094141867865e-05,
+      "loss": 0.0439,
+      "step": 15858
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.6290473516750894e-05,
+      "loss": 0.0538,
+      "step": 15860
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.629000561482313e-05,
+      "loss": 0.0602,
+      "step": 15862
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.628953771289538e-05,
+      "loss": 0.05,
+      "step": 15864
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.6289069810967625e-05,
+      "loss": 0.0514,
+      "step": 15866
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.628860190903987e-05,
+      "loss": 0.0504,
+      "step": 15868
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.628813400711211e-05,
+      "loss": 0.0541,
+      "step": 15870
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.6287666105184356e-05,
+      "loss": 0.0529,
+      "step": 15872
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.62871982032566e-05,
+      "loss": 0.0441,
+      "step": 15874
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.628673030132885e-05,
+      "loss": 0.0595,
+      "step": 15876
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.628626239940109e-05,
+      "loss": 0.0677,
+      "step": 15878
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.628579449747333e-05,
+      "loss": 0.0418,
+      "step": 15880
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.628532659554557e-05,
+      "loss": 0.0485,
+      "step": 15882
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.6284858693617825e-05,
+      "loss": 0.0765,
+      "step": 15884
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.6284390791690064e-05,
+      "loss": 0.0502,
+      "step": 15886
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.628392288976231e-05,
+      "loss": 0.0586,
+      "step": 15888
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.628345498783455e-05,
+      "loss": 0.0559,
+      "step": 15890
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.62829870859068e-05,
+      "loss": 0.0506,
+      "step": 15892
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.628251918397904e-05,
+      "loss": 0.0451,
+      "step": 15894
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.6282051282051287e-05,
+      "loss": 0.0502,
+      "step": 15896
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.6281583380123526e-05,
+      "loss": 0.0514,
+      "step": 15898
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.628111547819577e-05,
+      "loss": 0.0448,
+      "step": 15900
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.628064757626802e-05,
+      "loss": 0.0621,
+      "step": 15902
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.628017967434026e-05,
+      "loss": 0.0698,
+      "step": 15904
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.62797117724125e-05,
+      "loss": 0.0967,
+      "step": 15906
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.627924387048475e-05,
+      "loss": 0.0529,
+      "step": 15908
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.6278775968556994e-05,
+      "loss": 0.0491,
+      "step": 15910
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.627830806662924e-05,
+      "loss": 0.0508,
+      "step": 15912
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.627784016470148e-05,
+      "loss": 0.0655,
+      "step": 15914
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.6277372262773725e-05,
+      "loss": 0.0729,
+      "step": 15916
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.627690436084597e-05,
+      "loss": 0.0731,
+      "step": 15918
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.627643645891822e-05,
+      "loss": 0.0545,
+      "step": 15920
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.6275968556990456e-05,
+      "loss": 0.0717,
+      "step": 15922
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.62755006550627e-05,
+      "loss": 0.0521,
+      "step": 15924
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.627503275313495e-05,
+      "loss": 0.0446,
+      "step": 15926
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.6274564851207194e-05,
+      "loss": 0.0545,
+      "step": 15928
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.627409694927943e-05,
+      "loss": 0.0442,
+      "step": 15930
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.627362904735168e-05,
+      "loss": 0.0688,
+      "step": 15932
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.627316114542392e-05,
+      "loss": 0.0469,
+      "step": 15934
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.627269324349617e-05,
+      "loss": 0.0658,
+      "step": 15936
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.627222534156841e-05,
+      "loss": 0.0646,
+      "step": 15938
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.6271757439640656e-05,
+      "loss": 0.0568,
+      "step": 15940
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.6271289537712895e-05,
+      "loss": 0.0526,
+      "step": 15942
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.627082163578514e-05,
+      "loss": 0.0592,
+      "step": 15944
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.627035373385739e-05,
+      "loss": 0.0532,
+      "step": 15946
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.626988583192963e-05,
+      "loss": 0.0567,
+      "step": 15948
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.626941793000187e-05,
+      "loss": 0.0497,
+      "step": 15950
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.626895002807412e-05,
+      "loss": 0.0551,
+      "step": 15952
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.6268482126146364e-05,
+      "loss": 0.061,
+      "step": 15954
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.626801422421861e-05,
+      "loss": 0.0532,
+      "step": 15956
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.626754632229085e-05,
+      "loss": 0.0593,
+      "step": 15958
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.6267078420363095e-05,
+      "loss": 0.0664,
+      "step": 15960
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.626661051843534e-05,
+      "loss": 0.0697,
+      "step": 15962
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.626614261650758e-05,
+      "loss": 0.0644,
+      "step": 15964
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.6265674714579826e-05,
+      "loss": 0.0635,
+      "step": 15966
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.6265206812652065e-05,
+      "loss": 0.0528,
+      "step": 15968
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.626473891072432e-05,
+      "loss": 0.0927,
+      "step": 15970
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.626427100879656e-05,
+      "loss": 0.0519,
+      "step": 15972
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.62638031068688e-05,
+      "loss": 0.0556,
+      "step": 15974
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.626333520494104e-05,
+      "loss": 0.0586,
+      "step": 15976
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.626286730301329e-05,
+      "loss": 0.063,
+      "step": 15978
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.6262399401085533e-05,
+      "loss": 0.0455,
+      "step": 15980
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.626193149915778e-05,
+      "loss": 0.0621,
+      "step": 15982
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.626146359723002e-05,
+      "loss": 0.0717,
+      "step": 15984
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.6260995695302264e-05,
+      "loss": 0.0523,
+      "step": 15986
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.626052779337451e-05,
+      "loss": 0.0531,
+      "step": 15988
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.6260059891446756e-05,
+      "loss": 0.0528,
+      "step": 15990
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.6259591989518995e-05,
+      "loss": 0.0591,
+      "step": 15992
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.625912408759124e-05,
+      "loss": 0.049,
+      "step": 15994
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.625865618566349e-05,
+      "loss": 0.0585,
+      "step": 15996
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.625818828373573e-05,
+      "loss": 0.0465,
+      "step": 15998
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.625772038180797e-05,
+      "loss": 0.0579,
+      "step": 16000
+    },
+    {
+      "epoch": 2.25,
+      "eval_gen_len": 31.2759,
+      "eval_loss": 1.0764782428741455,
+      "eval_meteor": 0.0444,
+      "eval_runtime": 16.106,
+      "eval_samples_per_second": 3.601,
+      "eval_steps_per_second": 0.497,
+      "step": 16000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.625725247988022e-05,
+      "loss": 0.0719,
+      "step": 16002
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.6256784577952464e-05,
+      "loss": 0.0528,
+      "step": 16004
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.625631667602471e-05,
+      "loss": 0.0561,
+      "step": 16006
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.625584877409695e-05,
+      "loss": 0.076,
+      "step": 16008
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.6255380872169195e-05,
+      "loss": 0.078,
+      "step": 16010
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.6254912970241434e-05,
+      "loss": 0.059,
+      "step": 16012
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.625444506831369e-05,
+      "loss": 0.0592,
+      "step": 16014
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.6253977166385926e-05,
+      "loss": 0.0463,
+      "step": 16016
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.625350926445817e-05,
+      "loss": 0.0571,
+      "step": 16018
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.625304136253041e-05,
+      "loss": 0.0479,
+      "step": 16020
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.6252573460602664e-05,
+      "loss": 0.0529,
+      "step": 16022
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.62521055586749e-05,
+      "loss": 0.0784,
+      "step": 16024
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.625163765674715e-05,
+      "loss": 0.0507,
+      "step": 16026
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.625116975481939e-05,
+      "loss": 0.0442,
+      "step": 16028
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.6250701852891634e-05,
+      "loss": 0.0547,
+      "step": 16030
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.625023395096388e-05,
+      "loss": 0.0469,
+      "step": 16032
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.6249766049036126e-05,
+      "loss": 0.0338,
+      "step": 16034
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.6249298147108365e-05,
+      "loss": 0.0775,
+      "step": 16036
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.624883024518061e-05,
+      "loss": 0.0377,
+      "step": 16038
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.624836234325286e-05,
+      "loss": 0.062,
+      "step": 16040
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.62478944413251e-05,
+      "loss": 0.0591,
+      "step": 16042
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.624742653939734e-05,
+      "loss": 0.0792,
+      "step": 16044
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.624695863746959e-05,
+      "loss": 0.0659,
+      "step": 16046
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.6246490735541834e-05,
+      "loss": 0.0681,
+      "step": 16048
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.624602283361408e-05,
+      "loss": 0.0465,
+      "step": 16050
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.624555493168632e-05,
+      "loss": 0.0468,
+      "step": 16052
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.6245087029758564e-05,
+      "loss": 0.0424,
+      "step": 16054
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.624461912783081e-05,
+      "loss": 0.0625,
+      "step": 16056
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.6244151225903056e-05,
+      "loss": 0.0727,
+      "step": 16058
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.6243683323975295e-05,
+      "loss": 0.0527,
+      "step": 16060
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.624321542204754e-05,
+      "loss": 0.0547,
+      "step": 16062
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.624274752011978e-05,
+      "loss": 0.0608,
+      "step": 16064
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.624227961819203e-05,
+      "loss": 0.0504,
+      "step": 16066
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.624181171626427e-05,
+      "loss": 0.0585,
+      "step": 16068
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.624134381433652e-05,
+      "loss": 0.0524,
+      "step": 16070
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.624087591240876e-05,
+      "loss": 0.0604,
+      "step": 16072
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.624040801048101e-05,
+      "loss": 0.0621,
+      "step": 16074
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.623994010855325e-05,
+      "loss": 0.0873,
+      "step": 16076
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.6239472206625495e-05,
+      "loss": 0.0553,
+      "step": 16078
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.6239004304697734e-05,
+      "loss": 0.0554,
+      "step": 16080
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.623853640276998e-05,
+      "loss": 0.0522,
+      "step": 16082
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.6238068500842226e-05,
+      "loss": 0.0606,
+      "step": 16084
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.623760059891447e-05,
+      "loss": 0.0513,
+      "step": 16086
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.623713269698671e-05,
+      "loss": 0.0511,
+      "step": 16088
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.623666479505896e-05,
+      "loss": 0.0557,
+      "step": 16090
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.62361968931312e-05,
+      "loss": 0.0704,
+      "step": 16092
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.623572899120345e-05,
+      "loss": 0.049,
+      "step": 16094
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.623526108927569e-05,
+      "loss": 0.0575,
+      "step": 16096
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.6234793187347934e-05,
+      "loss": 0.0567,
+      "step": 16098
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.623432528542018e-05,
+      "loss": 0.0604,
+      "step": 16100
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.6233857383492426e-05,
+      "loss": 0.0472,
+      "step": 16102
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.6233389481564665e-05,
+      "loss": 0.0522,
+      "step": 16104
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.623292157963691e-05,
+      "loss": 0.063,
+      "step": 16106
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.623245367770915e-05,
+      "loss": 0.0545,
+      "step": 16108
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.62319857757814e-05,
+      "loss": 0.0519,
+      "step": 16110
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.623151787385364e-05,
+      "loss": 0.0685,
+      "step": 16112
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.623104997192589e-05,
+      "loss": 0.0856,
+      "step": 16114
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.623058206999813e-05,
+      "loss": 0.069,
+      "step": 16116
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.623011416807038e-05,
+      "loss": 0.0548,
+      "step": 16118
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.622964626614262e-05,
+      "loss": 0.0451,
+      "step": 16120
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.6229178364214864e-05,
+      "loss": 0.0634,
+      "step": 16122
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.6228710462287104e-05,
+      "loss": 0.0577,
+      "step": 16124
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.622824256035935e-05,
+      "loss": 0.0692,
+      "step": 16126
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.6227774658431595e-05,
+      "loss": 0.0533,
+      "step": 16128
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.622730675650384e-05,
+      "loss": 0.0734,
+      "step": 16130
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.622683885457608e-05,
+      "loss": 0.0567,
+      "step": 16132
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.6226370952648326e-05,
+      "loss": 0.0465,
+      "step": 16134
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.622590305072057e-05,
+      "loss": 0.0433,
+      "step": 16136
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.622543514879282e-05,
+      "loss": 0.0711,
+      "step": 16138
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.622496724686506e-05,
+      "loss": 0.0676,
+      "step": 16140
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.62244993449373e-05,
+      "loss": 0.0466,
+      "step": 16142
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.622403144300955e-05,
+      "loss": 0.0457,
+      "step": 16144
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.6223563541081795e-05,
+      "loss": 0.076,
+      "step": 16146
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.6223095639154034e-05,
+      "loss": 0.049,
+      "step": 16148
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.622262773722628e-05,
+      "loss": 0.066,
+      "step": 16150
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.6222159835298526e-05,
+      "loss": 0.0499,
+      "step": 16152
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.622169193337077e-05,
+      "loss": 0.0615,
+      "step": 16154
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.622122403144301e-05,
+      "loss": 0.0845,
+      "step": 16156
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.622075612951526e-05,
+      "loss": 0.0564,
+      "step": 16158
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.6220288227587496e-05,
+      "loss": 0.0744,
+      "step": 16160
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.621982032565975e-05,
+      "loss": 0.0741,
+      "step": 16162
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.621935242373199e-05,
+      "loss": 0.0589,
+      "step": 16164
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.6218884521804234e-05,
+      "loss": 0.048,
+      "step": 16166
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.621841661987647e-05,
+      "loss": 0.0606,
+      "step": 16168
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.6217948717948726e-05,
+      "loss": 0.0448,
+      "step": 16170
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.6217480816020965e-05,
+      "loss": 0.0572,
+      "step": 16172
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.621701291409321e-05,
+      "loss": 0.0815,
+      "step": 16174
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.621654501216545e-05,
+      "loss": 0.055,
+      "step": 16176
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.6216077110237696e-05,
+      "loss": 0.0532,
+      "step": 16178
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.621560920830994e-05,
+      "loss": 0.0919,
+      "step": 16180
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.621514130638219e-05,
+      "loss": 0.0571,
+      "step": 16182
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.621467340445443e-05,
+      "loss": 0.0619,
+      "step": 16184
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.621420550252667e-05,
+      "loss": 0.0578,
+      "step": 16186
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.621373760059892e-05,
+      "loss": 0.0743,
+      "step": 16188
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.6213269698671165e-05,
+      "loss": 0.0552,
+      "step": 16190
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.6212801796743404e-05,
+      "loss": 0.0691,
+      "step": 16192
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.621233389481565e-05,
+      "loss": 0.0478,
+      "step": 16194
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.6211865992887895e-05,
+      "loss": 0.0586,
+      "step": 16196
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.621139809096014e-05,
+      "loss": 0.076,
+      "step": 16198
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.621093018903238e-05,
+      "loss": 0.0608,
+      "step": 16200
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.6210462287104626e-05,
+      "loss": 0.0567,
+      "step": 16202
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.620999438517687e-05,
+      "loss": 0.0479,
+      "step": 16204
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.620952648324912e-05,
+      "loss": 0.0485,
+      "step": 16206
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.620905858132136e-05,
+      "loss": 0.0612,
+      "step": 16208
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.62085906793936e-05,
+      "loss": 0.0601,
+      "step": 16210
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.620812277746584e-05,
+      "loss": 0.0503,
+      "step": 16212
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.620765487553809e-05,
+      "loss": 0.0633,
+      "step": 16214
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.6207186973610334e-05,
+      "loss": 0.0617,
+      "step": 16216
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.6206719071682573e-05,
+      "loss": 0.0533,
+      "step": 16218
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.620625116975482e-05,
+      "loss": 0.0569,
+      "step": 16220
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.6205783267827065e-05,
+      "loss": 0.0582,
+      "step": 16222
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.620531536589931e-05,
+      "loss": 0.0469,
+      "step": 16224
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.620484746397155e-05,
+      "loss": 0.1163,
+      "step": 16226
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.6204379562043796e-05,
+      "loss": 0.0874,
+      "step": 16228
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.620391166011604e-05,
+      "loss": 0.0459,
+      "step": 16230
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.620344375818829e-05,
+      "loss": 0.0586,
+      "step": 16232
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.620297585626053e-05,
+      "loss": 0.0801,
+      "step": 16234
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.620250795433277e-05,
+      "loss": 0.0812,
+      "step": 16236
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.620204005240502e-05,
+      "loss": 0.0583,
+      "step": 16238
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.6201572150477265e-05,
+      "loss": 0.0785,
+      "step": 16240
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.6201104248549504e-05,
+      "loss": 0.0425,
+      "step": 16242
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.620063634662175e-05,
+      "loss": 0.0945,
+      "step": 16244
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.620016844469399e-05,
+      "loss": 0.0518,
+      "step": 16246
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.619970054276624e-05,
+      "loss": 0.0438,
+      "step": 16248
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.619923264083848e-05,
+      "loss": 0.0681,
+      "step": 16250
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.619876473891073e-05,
+      "loss": 0.0466,
+      "step": 16252
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.6198296836982966e-05,
+      "loss": 0.0776,
+      "step": 16254
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.619782893505521e-05,
+      "loss": 0.0648,
+      "step": 16256
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.619736103312746e-05,
+      "loss": 0.0503,
+      "step": 16258
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.6196893131199704e-05,
+      "loss": 0.0909,
+      "step": 16260
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.619642522927194e-05,
+      "loss": 0.0644,
+      "step": 16262
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.619595732734419e-05,
+      "loss": 0.0481,
+      "step": 16264
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.6195489425416435e-05,
+      "loss": 0.0545,
+      "step": 16266
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.619502152348868e-05,
+      "loss": 0.0611,
+      "step": 16268
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.619455362156092e-05,
+      "loss": 0.0729,
+      "step": 16270
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.6194085719633166e-05,
+      "loss": 0.0636,
+      "step": 16272
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.619361781770541e-05,
+      "loss": 0.0756,
+      "step": 16274
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.619314991577766e-05,
+      "loss": 0.0516,
+      "step": 16276
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.6192682013849897e-05,
+      "loss": 0.0693,
+      "step": 16278
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.619221411192214e-05,
+      "loss": 0.0627,
+      "step": 16280
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.619174620999439e-05,
+      "loss": 0.062,
+      "step": 16282
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.6191278308066634e-05,
+      "loss": 0.0543,
+      "step": 16284
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.6190810406138873e-05,
+      "loss": 0.0464,
+      "step": 16286
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.619034250421112e-05,
+      "loss": 0.0653,
+      "step": 16288
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.618987460228336e-05,
+      "loss": 0.0498,
+      "step": 16290
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.618940670035561e-05,
+      "loss": 0.062,
+      "step": 16292
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.618893879842785e-05,
+      "loss": 0.0462,
+      "step": 16294
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.6188470896500096e-05,
+      "loss": 0.0664,
+      "step": 16296
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.6188002994572335e-05,
+      "loss": 0.0627,
+      "step": 16298
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.618753509264459e-05,
+      "loss": 0.0553,
+      "step": 16300
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.618706719071683e-05,
+      "loss": 0.0562,
+      "step": 16302
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.618659928878907e-05,
+      "loss": 0.0546,
+      "step": 16304
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.618613138686131e-05,
+      "loss": 0.0595,
+      "step": 16306
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.618566348493356e-05,
+      "loss": 0.0518,
+      "step": 16308
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.6185195583005804e-05,
+      "loss": 0.0709,
+      "step": 16310
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.618472768107805e-05,
+      "loss": 0.055,
+      "step": 16312
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.618425977915029e-05,
+      "loss": 0.0624,
+      "step": 16314
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.6183791877222535e-05,
+      "loss": 0.0775,
+      "step": 16316
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.618332397529478e-05,
+      "loss": 0.0514,
+      "step": 16318
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.618285607336703e-05,
+      "loss": 0.0788,
+      "step": 16320
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.6182388171439266e-05,
+      "loss": 0.0521,
+      "step": 16322
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.618192026951151e-05,
+      "loss": 0.0568,
+      "step": 16324
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.618145236758376e-05,
+      "loss": 0.0472,
+      "step": 16326
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.6180984465656004e-05,
+      "loss": 0.0519,
+      "step": 16328
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.618051656372824e-05,
+      "loss": 0.0589,
+      "step": 16330
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.618004866180049e-05,
+      "loss": 0.0736,
+      "step": 16332
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.6179580759872735e-05,
+      "loss": 0.0629,
+      "step": 16334
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.617911285794498e-05,
+      "loss": 0.0486,
+      "step": 16336
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.617864495601722e-05,
+      "loss": 0.0601,
+      "step": 16338
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.6178177054089466e-05,
+      "loss": 0.0581,
+      "step": 16340
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.6177709152161705e-05,
+      "loss": 0.0396,
+      "step": 16342
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.617724125023396e-05,
+      "loss": 0.0938,
+      "step": 16344
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.6176773348306197e-05,
+      "loss": 0.0623,
+      "step": 16346
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.617630544637844e-05,
+      "loss": 0.0791,
+      "step": 16348
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.617583754445068e-05,
+      "loss": 0.057,
+      "step": 16350
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6175369642522934e-05,
+      "loss": 0.0463,
+      "step": 16352
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6174901740595173e-05,
+      "loss": 0.046,
+      "step": 16354
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.617443383866742e-05,
+      "loss": 0.0841,
+      "step": 16356
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.617396593673966e-05,
+      "loss": 0.0467,
+      "step": 16358
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6173498034811904e-05,
+      "loss": 0.0535,
+      "step": 16360
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.617303013288415e-05,
+      "loss": 0.047,
+      "step": 16362
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6172562230956396e-05,
+      "loss": 0.1096,
+      "step": 16364
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6172094329028635e-05,
+      "loss": 0.068,
+      "step": 16366
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.617162642710088e-05,
+      "loss": 0.0582,
+      "step": 16368
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.617115852517313e-05,
+      "loss": 0.0677,
+      "step": 16370
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.617069062324537e-05,
+      "loss": 0.0565,
+      "step": 16372
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.617022272131761e-05,
+      "loss": 0.07,
+      "step": 16374
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.616975481938986e-05,
+      "loss": 0.0644,
+      "step": 16376
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6169286917462104e-05,
+      "loss": 0.0604,
+      "step": 16378
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.616881901553435e-05,
+      "loss": 0.0556,
+      "step": 16380
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.616835111360659e-05,
+      "loss": 0.0456,
+      "step": 16382
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6167883211678835e-05,
+      "loss": 0.0548,
+      "step": 16384
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.616741530975108e-05,
+      "loss": 0.0578,
+      "step": 16386
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.616694740782333e-05,
+      "loss": 0.0583,
+      "step": 16388
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6166479505895566e-05,
+      "loss": 0.0528,
+      "step": 16390
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.616601160396781e-05,
+      "loss": 0.064,
+      "step": 16392
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.616554370204005e-05,
+      "loss": 0.058,
+      "step": 16394
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6165075800112304e-05,
+      "loss": 0.0615,
+      "step": 16396
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.616460789818454e-05,
+      "loss": 0.0498,
+      "step": 16398
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.616413999625679e-05,
+      "loss": 0.0625,
+      "step": 16400
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.616367209432903e-05,
+      "loss": 0.0654,
+      "step": 16402
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6163204192401274e-05,
+      "loss": 0.0408,
+      "step": 16404
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.616273629047352e-05,
+      "loss": 0.057,
+      "step": 16406
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6162268388545766e-05,
+      "loss": 0.0513,
+      "step": 16408
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6161800486618005e-05,
+      "loss": 0.0735,
+      "step": 16410
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.616133258469025e-05,
+      "loss": 0.0664,
+      "step": 16412
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6160864682762497e-05,
+      "loss": 0.0707,
+      "step": 16414
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.616039678083474e-05,
+      "loss": 0.0516,
+      "step": 16416
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.615992887890698e-05,
+      "loss": 0.0514,
+      "step": 16418
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.615946097697923e-05,
+      "loss": 0.0582,
+      "step": 16420
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6158993075051473e-05,
+      "loss": 0.0472,
+      "step": 16422
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.615852517312372e-05,
+      "loss": 0.0581,
+      "step": 16424
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.615805727119596e-05,
+      "loss": 0.0539,
+      "step": 16426
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6157589369268204e-05,
+      "loss": 0.0751,
+      "step": 16428
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.615712146734045e-05,
+      "loss": 0.0493,
+      "step": 16430
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6156653565412696e-05,
+      "loss": 0.0724,
+      "step": 16432
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6156185663484935e-05,
+      "loss": 0.0493,
+      "step": 16434
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.615571776155718e-05,
+      "loss": 0.0519,
+      "step": 16436
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.615524985962942e-05,
+      "loss": 0.0593,
+      "step": 16438
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.615478195770167e-05,
+      "loss": 0.048,
+      "step": 16440
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.615431405577391e-05,
+      "loss": 0.0508,
+      "step": 16442
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.615384615384616e-05,
+      "loss": 0.046,
+      "step": 16444
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.61533782519184e-05,
+      "loss": 0.0685,
+      "step": 16446
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.615291034999065e-05,
+      "loss": 0.0537,
+      "step": 16448
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.615244244806289e-05,
+      "loss": 0.0448,
+      "step": 16450
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6151974546135135e-05,
+      "loss": 0.0671,
+      "step": 16452
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6151506644207374e-05,
+      "loss": 0.0448,
+      "step": 16454
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.615103874227962e-05,
+      "loss": 0.0539,
+      "step": 16456
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6150570840351866e-05,
+      "loss": 0.0512,
+      "step": 16458
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.615010293842411e-05,
+      "loss": 0.0685,
+      "step": 16460
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.614963503649635e-05,
+      "loss": 0.0577,
+      "step": 16462
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.61491671345686e-05,
+      "loss": 0.0531,
+      "step": 16464
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.614869923264084e-05,
+      "loss": 0.0841,
+      "step": 16466
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.614823133071308e-05,
+      "loss": 0.0471,
+      "step": 16468
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.614776342878533e-05,
+      "loss": 0.0877,
+      "step": 16470
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.614729552685757e-05,
+      "loss": 0.07,
+      "step": 16472
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.614682762492982e-05,
+      "loss": 0.0508,
+      "step": 16474
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.614635972300206e-05,
+      "loss": 0.0608,
+      "step": 16476
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6145891821074305e-05,
+      "loss": 0.0594,
+      "step": 16478
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6145423919146544e-05,
+      "loss": 0.0865,
+      "step": 16480
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6144956017218797e-05,
+      "loss": 0.0721,
+      "step": 16482
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6144488115291036e-05,
+      "loss": 0.0509,
+      "step": 16484
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.614402021336328e-05,
+      "loss": 0.0556,
+      "step": 16486
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.614355231143552e-05,
+      "loss": 0.0462,
+      "step": 16488
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.614308440950777e-05,
+      "loss": 0.0593,
+      "step": 16490
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.614261650758001e-05,
+      "loss": 0.0748,
+      "step": 16492
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.614214860565226e-05,
+      "loss": 0.0551,
+      "step": 16494
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.61416807037245e-05,
+      "loss": 0.0499,
+      "step": 16496
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6141212801796744e-05,
+      "loss": 0.0581,
+      "step": 16498
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.614074489986899e-05,
+      "loss": 0.0746,
+      "step": 16500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6140276997941235e-05,
+      "loss": 0.0484,
+      "step": 16502
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6139809096013475e-05,
+      "loss": 0.0661,
+      "step": 16504
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.613934119408572e-05,
+      "loss": 0.0562,
+      "step": 16506
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6138873292157966e-05,
+      "loss": 0.0686,
+      "step": 16508
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.613840539023021e-05,
+      "loss": 0.0752,
+      "step": 16510
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.613793748830245e-05,
+      "loss": 0.0491,
+      "step": 16512
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.61374695863747e-05,
+      "loss": 0.0613,
+      "step": 16514
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.613700168444694e-05,
+      "loss": 0.0764,
+      "step": 16516
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.613653378251919e-05,
+      "loss": 0.0659,
+      "step": 16518
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.613606588059143e-05,
+      "loss": 0.0626,
+      "step": 16520
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6135597978663674e-05,
+      "loss": 0.0582,
+      "step": 16522
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.613513007673591e-05,
+      "loss": 0.0557,
+      "step": 16524
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6134662174808166e-05,
+      "loss": 0.0657,
+      "step": 16526
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6134194272880405e-05,
+      "loss": 0.0618,
+      "step": 16528
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.613372637095265e-05,
+      "loss": 0.072,
+      "step": 16530
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.613325846902489e-05,
+      "loss": 0.062,
+      "step": 16532
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6132790567097136e-05,
+      "loss": 0.0533,
+      "step": 16534
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.613232266516938e-05,
+      "loss": 0.0631,
+      "step": 16536
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.613185476324163e-05,
+      "loss": 0.0842,
+      "step": 16538
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.613138686131387e-05,
+      "loss": 0.0708,
+      "step": 16540
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.613091895938611e-05,
+      "loss": 0.0484,
+      "step": 16542
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.613045105745836e-05,
+      "loss": 0.0605,
+      "step": 16544
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6129983155530605e-05,
+      "loss": 0.0594,
+      "step": 16546
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6129515253602844e-05,
+      "loss": 0.0636,
+      "step": 16548
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.612904735167509e-05,
+      "loss": 0.0584,
+      "step": 16550
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6128579449747336e-05,
+      "loss": 0.0532,
+      "step": 16552
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.612811154781958e-05,
+      "loss": 0.073,
+      "step": 16554
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.612764364589182e-05,
+      "loss": 0.0531,
+      "step": 16556
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.612717574396407e-05,
+      "loss": 0.0558,
+      "step": 16558
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.612670784203631e-05,
+      "loss": 0.0779,
+      "step": 16560
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.612623994010856e-05,
+      "loss": 0.0613,
+      "step": 16562
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.61257720381808e-05,
+      "loss": 0.0509,
+      "step": 16564
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6125304136253044e-05,
+      "loss": 0.0541,
+      "step": 16566
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.612483623432528e-05,
+      "loss": 0.0821,
+      "step": 16568
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6124368332397535e-05,
+      "loss": 0.0678,
+      "step": 16570
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6123900430469775e-05,
+      "loss": 0.0801,
+      "step": 16572
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.612343252854202e-05,
+      "loss": 0.0473,
+      "step": 16574
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.612296462661426e-05,
+      "loss": 0.0618,
+      "step": 16576
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.612249672468651e-05,
+      "loss": 0.0674,
+      "step": 16578
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.612202882275875e-05,
+      "loss": 0.0641,
+      "step": 16580
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6121560920831e-05,
+      "loss": 0.0386,
+      "step": 16582
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6121093018903236e-05,
+      "loss": 0.0587,
+      "step": 16584
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.612062511697548e-05,
+      "loss": 0.0815,
+      "step": 16586
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.612015721504773e-05,
+      "loss": 0.0677,
+      "step": 16588
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6119689313119974e-05,
+      "loss": 0.0587,
+      "step": 16590
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.611922141119221e-05,
+      "loss": 0.0756,
+      "step": 16592
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.611875350926446e-05,
+      "loss": 0.052,
+      "step": 16594
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6118285607336705e-05,
+      "loss": 0.0793,
+      "step": 16596
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.611781770540895e-05,
+      "loss": 0.0521,
+      "step": 16598
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.611734980348119e-05,
+      "loss": 0.0568,
+      "step": 16600
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6116881901553436e-05,
+      "loss": 0.0561,
+      "step": 16602
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.611641399962568e-05,
+      "loss": 0.0714,
+      "step": 16604
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.611594609769793e-05,
+      "loss": 0.0629,
+      "step": 16606
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.611547819577017e-05,
+      "loss": 0.0693,
+      "step": 16608
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.611501029384241e-05,
+      "loss": 0.0908,
+      "step": 16610
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.611454239191466e-05,
+      "loss": 0.0742,
+      "step": 16612
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6114074489986905e-05,
+      "loss": 0.0683,
+      "step": 16614
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6113606588059144e-05,
+      "loss": 0.0545,
+      "step": 16616
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.611313868613139e-05,
+      "loss": 0.0542,
+      "step": 16618
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.611267078420363e-05,
+      "loss": 0.074,
+      "step": 16620
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.611220288227588e-05,
+      "loss": 0.051,
+      "step": 16622
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.611173498034812e-05,
+      "loss": 0.0808,
+      "step": 16624
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.611126707842037e-05,
+      "loss": 0.0537,
+      "step": 16626
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6110799176492606e-05,
+      "loss": 0.0457,
+      "step": 16628
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.611033127456486e-05,
+      "loss": 0.087,
+      "step": 16630
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.61098633726371e-05,
+      "loss": 0.0687,
+      "step": 16632
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6109395470709344e-05,
+      "loss": 0.056,
+      "step": 16634
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.610892756878158e-05,
+      "loss": 0.0518,
+      "step": 16636
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.610845966685383e-05,
+      "loss": 0.0531,
+      "step": 16638
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.6107991764926075e-05,
+      "loss": 0.0789,
+      "step": 16640
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.610752386299832e-05,
+      "loss": 0.0628,
+      "step": 16642
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.610705596107056e-05,
+      "loss": 0.0588,
+      "step": 16644
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.6106588059142806e-05,
+      "loss": 0.0638,
+      "step": 16646
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.610612015721505e-05,
+      "loss": 0.061,
+      "step": 16648
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.61056522552873e-05,
+      "loss": 0.0556,
+      "step": 16650
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.6105184353359536e-05,
+      "loss": 0.0462,
+      "step": 16652
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.610471645143178e-05,
+      "loss": 0.0714,
+      "step": 16654
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.610424854950403e-05,
+      "loss": 0.0501,
+      "step": 16656
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.6103780647576274e-05,
+      "loss": 0.0516,
+      "step": 16658
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.610331274564851e-05,
+      "loss": 0.0599,
+      "step": 16660
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.610284484372076e-05,
+      "loss": 0.0821,
+      "step": 16662
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.6102376941793005e-05,
+      "loss": 0.0811,
+      "step": 16664
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.610190903986525e-05,
+      "loss": 0.0708,
+      "step": 16666
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.610144113793749e-05,
+      "loss": 0.0741,
+      "step": 16668
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.6100973236009736e-05,
+      "loss": 0.0713,
+      "step": 16670
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.6100505334081975e-05,
+      "loss": 0.0618,
+      "step": 16672
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.610003743215423e-05,
+      "loss": 0.0671,
+      "step": 16674
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.609956953022647e-05,
+      "loss": 0.0601,
+      "step": 16676
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.609910162829871e-05,
+      "loss": 0.0528,
+      "step": 16678
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.609863372637095e-05,
+      "loss": 0.0526,
+      "step": 16680
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.60981658244432e-05,
+      "loss": 0.0576,
+      "step": 16682
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.6097697922515444e-05,
+      "loss": 0.0426,
+      "step": 16684
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.609723002058769e-05,
+      "loss": 0.0726,
+      "step": 16686
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.609676211865993e-05,
+      "loss": 0.0705,
+      "step": 16688
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.6096294216732175e-05,
+      "loss": 0.0621,
+      "step": 16690
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.609582631480442e-05,
+      "loss": 0.0509,
+      "step": 16692
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.609535841287667e-05,
+      "loss": 0.0782,
+      "step": 16694
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.6094890510948906e-05,
+      "loss": 0.0638,
+      "step": 16696
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.609442260902115e-05,
+      "loss": 0.0521,
+      "step": 16698
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.60939547070934e-05,
+      "loss": 0.0451,
+      "step": 16700
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.6093486805165644e-05,
+      "loss": 0.0591,
+      "step": 16702
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.609301890323788e-05,
+      "loss": 0.0667,
+      "step": 16704
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.609255100131013e-05,
+      "loss": 0.0561,
+      "step": 16706
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6092083099382375e-05,
+      "loss": 0.0688,
+      "step": 16708
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.609161519745462e-05,
+      "loss": 0.0598,
+      "step": 16710
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.609114729552686e-05,
+      "loss": 0.0753,
+      "step": 16712
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6090679393599106e-05,
+      "loss": 0.0469,
+      "step": 16714
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6090211491671345e-05,
+      "loss": 0.0704,
+      "step": 16716
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.608974358974359e-05,
+      "loss": 0.077,
+      "step": 16718
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6089275687815836e-05,
+      "loss": 0.0525,
+      "step": 16720
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6088807785888076e-05,
+      "loss": 0.0562,
+      "step": 16722
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.608833988396032e-05,
+      "loss": 0.052,
+      "step": 16724
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.608787198203257e-05,
+      "loss": 0.0595,
+      "step": 16726
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.608740408010481e-05,
+      "loss": 0.0414,
+      "step": 16728
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.608693617817705e-05,
+      "loss": 0.0572,
+      "step": 16730
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.60864682762493e-05,
+      "loss": 0.0627,
+      "step": 16732
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6086000374321544e-05,
+      "loss": 0.0548,
+      "step": 16734
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.608553247239379e-05,
+      "loss": 0.0753,
+      "step": 16736
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.608506457046603e-05,
+      "loss": 0.0451,
+      "step": 16738
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6084596668538275e-05,
+      "loss": 0.0617,
+      "step": 16740
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.608412876661052e-05,
+      "loss": 0.053,
+      "step": 16742
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.608366086468277e-05,
+      "loss": 0.0546,
+      "step": 16744
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6083192962755006e-05,
+      "loss": 0.0679,
+      "step": 16746
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.608272506082725e-05,
+      "loss": 0.0787,
+      "step": 16748
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.608225715889949e-05,
+      "loss": 0.0687,
+      "step": 16750
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6081789256971744e-05,
+      "loss": 0.0536,
+      "step": 16752
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.608132135504398e-05,
+      "loss": 0.0475,
+      "step": 16754
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.608085345311623e-05,
+      "loss": 0.0521,
+      "step": 16756
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.608038555118847e-05,
+      "loss": 0.0566,
+      "step": 16758
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.607991764926072e-05,
+      "loss": 0.0481,
+      "step": 16760
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.607944974733296e-05,
+      "loss": 0.0376,
+      "step": 16762
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6078981845405206e-05,
+      "loss": 0.0459,
+      "step": 16764
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.6078513943477445e-05,
+      "loss": 0.0415,
+      "step": 16766
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.607804604154969e-05,
+      "loss": 0.0703,
+      "step": 16768
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.607757813962194e-05,
+      "loss": 0.0706,
+      "step": 16770
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.607711023769418e-05,
+      "loss": 0.064,
+      "step": 16772
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.607664233576642e-05,
+      "loss": 0.0482,
+      "step": 16774
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.607617443383867e-05,
+      "loss": 0.0616,
+      "step": 16776
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.6075706531910914e-05,
+      "loss": 0.0534,
+      "step": 16778
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.607523862998316e-05,
+      "loss": 0.0702,
+      "step": 16780
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.60747707280554e-05,
+      "loss": 0.0701,
+      "step": 16782
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.6074302826127645e-05,
+      "loss": 0.0727,
+      "step": 16784
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.607383492419989e-05,
+      "loss": 0.0622,
+      "step": 16786
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.6073367022272137e-05,
+      "loss": 0.0635,
+      "step": 16788
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.6072899120344376e-05,
+      "loss": 0.0579,
+      "step": 16790
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.607243121841662e-05,
+      "loss": 0.0479,
+      "step": 16792
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.607196331648887e-05,
+      "loss": 0.0494,
+      "step": 16794
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.607149541456111e-05,
+      "loss": 0.0831,
+      "step": 16796
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.607102751263335e-05,
+      "loss": 0.0533,
+      "step": 16798
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.60705596107056e-05,
+      "loss": 0.0537,
+      "step": 16800
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.607009170877784e-05,
+      "loss": 0.0819,
+      "step": 16802
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.606962380685009e-05,
+      "loss": 0.0561,
+      "step": 16804
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.606915590492233e-05,
+      "loss": 0.06,
+      "step": 16806
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.6068688002994575e-05,
+      "loss": 0.0597,
+      "step": 16808
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.6068220101066814e-05,
+      "loss": 0.0701,
+      "step": 16810
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.606775219913906e-05,
+      "loss": 0.0679,
+      "step": 16812
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.6067284297211306e-05,
+      "loss": 0.0591,
+      "step": 16814
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.606681639528355e-05,
+      "loss": 0.0587,
+      "step": 16816
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.606634849335579e-05,
+      "loss": 0.0487,
+      "step": 16818
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.606588059142804e-05,
+      "loss": 0.0621,
+      "step": 16820
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.606541268950028e-05,
+      "loss": 0.0735,
+      "step": 16822
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.606494478757253e-05,
+      "loss": 0.0654,
+      "step": 16824
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.606447688564477e-05,
+      "loss": 0.0682,
+      "step": 16826
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.6064008983717014e-05,
+      "loss": 0.0346,
+      "step": 16828
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.606354108178926e-05,
+      "loss": 0.0667,
+      "step": 16830
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.6063073179861506e-05,
+      "loss": 0.0453,
+      "step": 16832
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.6062605277933745e-05,
+      "loss": 0.0623,
+      "step": 16834
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.606213737600599e-05,
+      "loss": 0.0471,
+      "step": 16836
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.606166947407824e-05,
+      "loss": 0.0709,
+      "step": 16838
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.606120157215048e-05,
+      "loss": 0.0435,
+      "step": 16840
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.606073367022272e-05,
+      "loss": 0.0719,
+      "step": 16842
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.606026576829497e-05,
+      "loss": 0.0587,
+      "step": 16844
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.605979786636721e-05,
+      "loss": 0.0595,
+      "step": 16846
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.605932996443946e-05,
+      "loss": 0.0608,
+      "step": 16848
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.60588620625117e-05,
+      "loss": 0.0538,
+      "step": 16850
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.6058394160583945e-05,
+      "loss": 0.0636,
+      "step": 16852
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.6057926258656184e-05,
+      "loss": 0.0599,
+      "step": 16854
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.6057458356728437e-05,
+      "loss": 0.0517,
+      "step": 16856
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.6056990454800676e-05,
+      "loss": 0.0577,
+      "step": 16858
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.605652255287292e-05,
+      "loss": 0.0617,
+      "step": 16860
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.605605465094516e-05,
+      "loss": 0.0459,
+      "step": 16862
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.605558674901741e-05,
+      "loss": 0.0593,
+      "step": 16864
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.605511884708965e-05,
+      "loss": 0.0528,
+      "step": 16866
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.60546509451619e-05,
+      "loss": 0.069,
+      "step": 16868
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.605418304323414e-05,
+      "loss": 0.0485,
+      "step": 16870
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.6053715141306383e-05,
+      "loss": 0.0633,
+      "step": 16872
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.605324723937863e-05,
+      "loss": 0.0693,
+      "step": 16874
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.6052779337450875e-05,
+      "loss": 0.0531,
+      "step": 16876
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.6052311435523114e-05,
+      "loss": 0.0652,
+      "step": 16878
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.605184353359536e-05,
+      "loss": 0.0575,
+      "step": 16880
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.6051375631667606e-05,
+      "loss": 0.0489,
+      "step": 16882
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.605090772973985e-05,
+      "loss": 0.0633,
+      "step": 16884
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.605043982781209e-05,
+      "loss": 0.0673,
+      "step": 16886
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.604997192588434e-05,
+      "loss": 0.0693,
+      "step": 16888
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.604950402395658e-05,
+      "loss": 0.0557,
+      "step": 16890
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.604903612202883e-05,
+      "loss": 0.07,
+      "step": 16892
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.604856822010107e-05,
+      "loss": 0.0546,
+      "step": 16894
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.6048100318173314e-05,
+      "loss": 0.0549,
+      "step": 16896
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.604763241624555e-05,
+      "loss": 0.0663,
+      "step": 16898
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.6047164514317806e-05,
+      "loss": 0.0624,
+      "step": 16900
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.6046696612390045e-05,
+      "loss": 0.043,
+      "step": 16902
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.604622871046229e-05,
+      "loss": 0.0709,
+      "step": 16904
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.604576080853453e-05,
+      "loss": 0.0474,
+      "step": 16906
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.604529290660678e-05,
+      "loss": 0.0657,
+      "step": 16908
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.604482500467902e-05,
+      "loss": 0.0666,
+      "step": 16910
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.604435710275127e-05,
+      "loss": 0.0794,
+      "step": 16912
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.604388920082351e-05,
+      "loss": 0.0522,
+      "step": 16914
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.604342129889575e-05,
+      "loss": 0.0546,
+      "step": 16916
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.6042953396968e-05,
+      "loss": 0.0876,
+      "step": 16918
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.6042485495040245e-05,
+      "loss": 0.0542,
+      "step": 16920
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.6042017593112484e-05,
+      "loss": 0.0445,
+      "step": 16922
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.604154969118473e-05,
+      "loss": 0.0845,
+      "step": 16924
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.6041081789256976e-05,
+      "loss": 0.0591,
+      "step": 16926
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.604061388732922e-05,
+      "loss": 0.0515,
+      "step": 16928
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.604014598540146e-05,
+      "loss": 0.0716,
+      "step": 16930
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.603967808347371e-05,
+      "loss": 0.045,
+      "step": 16932
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.603921018154595e-05,
+      "loss": 0.0721,
+      "step": 16934
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.60387422796182e-05,
+      "loss": 0.0594,
+      "step": 16936
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.603827437769044e-05,
+      "loss": 0.0614,
+      "step": 16938
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.6037806475762684e-05,
+      "loss": 0.0675,
+      "step": 16940
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.603733857383493e-05,
+      "loss": 0.0688,
+      "step": 16942
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.6036870671907175e-05,
+      "loss": 0.0542,
+      "step": 16944
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.6036402769979414e-05,
+      "loss": 0.0424,
+      "step": 16946
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.603593486805166e-05,
+      "loss": 0.0735,
+      "step": 16948
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.60354669661239e-05,
+      "loss": 0.0488,
+      "step": 16950
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.603499906419615e-05,
+      "loss": 0.0603,
+      "step": 16952
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.603453116226839e-05,
+      "loss": 0.0782,
+      "step": 16954
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.603406326034064e-05,
+      "loss": 0.0582,
+      "step": 16956
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.6033595358412876e-05,
+      "loss": 0.064,
+      "step": 16958
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.603312745648512e-05,
+      "loss": 0.0792,
+      "step": 16960
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.603265955455737e-05,
+      "loss": 0.0692,
+      "step": 16962
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.6032191652629614e-05,
+      "loss": 0.0634,
+      "step": 16964
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.603172375070185e-05,
+      "loss": 0.0441,
+      "step": 16966
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.60312558487741e-05,
+      "loss": 0.0729,
+      "step": 16968
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.6030787946846345e-05,
+      "loss": 0.0599,
+      "step": 16970
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.6030320044918584e-05,
+      "loss": 0.0608,
+      "step": 16972
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.602985214299083e-05,
+      "loss": 0.0811,
+      "step": 16974
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.6029384241063076e-05,
+      "loss": 0.0585,
+      "step": 16976
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.602891633913532e-05,
+      "loss": 0.0783,
+      "step": 16978
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.602844843720756e-05,
+      "loss": 0.066,
+      "step": 16980
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.602798053527981e-05,
+      "loss": 0.0548,
+      "step": 16982
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.6027512633352046e-05,
+      "loss": 0.0649,
+      "step": 16984
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.60270447314243e-05,
+      "loss": 0.0574,
+      "step": 16986
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.602657682949654e-05,
+      "loss": 0.0691,
+      "step": 16988
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.6026108927568784e-05,
+      "loss": 0.0485,
+      "step": 16990
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.602564102564102e-05,
+      "loss": 0.0437,
+      "step": 16992
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.602517312371327e-05,
+      "loss": 0.0648,
+      "step": 16994
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.6024705221785515e-05,
+      "loss": 0.062,
+      "step": 16996
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.602423731985776e-05,
+      "loss": 0.0586,
+      "step": 16998
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.602376941793e-05,
+      "loss": 0.0713,
+      "step": 17000
+    },
+    {
+      "epoch": 2.39,
+      "eval_gen_len": 29.3103,
+      "eval_loss": 1.077703595161438,
+      "eval_meteor": 0.0453,
+      "eval_runtime": 16.122,
+      "eval_samples_per_second": 3.598,
+      "eval_steps_per_second": 0.496,
+      "step": 17000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.6023301516002246e-05,
+      "loss": 0.0617,
+      "step": 17002
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.602283361407449e-05,
+      "loss": 0.0574,
+      "step": 17004
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.602236571214674e-05,
+      "loss": 0.0454,
+      "step": 17006
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.602189781021898e-05,
+      "loss": 0.0754,
+      "step": 17008
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.602142990829122e-05,
+      "loss": 0.076,
+      "step": 17010
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.602096200636347e-05,
+      "loss": 0.0908,
+      "step": 17012
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.6020494104435714e-05,
+      "loss": 0.0524,
+      "step": 17014
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.6020026202507954e-05,
+      "loss": 0.0693,
+      "step": 17016
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.60195583005802e-05,
+      "loss": 0.0546,
+      "step": 17018
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.6019090398652445e-05,
+      "loss": 0.0899,
+      "step": 17020
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.601862249672469e-05,
+      "loss": 0.0628,
+      "step": 17022
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.601815459479693e-05,
+      "loss": 0.0701,
+      "step": 17024
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.6017686692869176e-05,
+      "loss": 0.0752,
+      "step": 17026
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.6017218790941416e-05,
+      "loss": 0.0449,
+      "step": 17028
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.601675088901367e-05,
+      "loss": 0.066,
+      "step": 17030
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.601628298708591e-05,
+      "loss": 0.0833,
+      "step": 17032
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.601581508515815e-05,
+      "loss": 0.0474,
+      "step": 17034
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.601534718323039e-05,
+      "loss": 0.0568,
+      "step": 17036
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.6014879281302645e-05,
+      "loss": 0.0506,
+      "step": 17038
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.6014411379374884e-05,
+      "loss": 0.061,
+      "step": 17040
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.601394347744713e-05,
+      "loss": 0.0601,
+      "step": 17042
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.601347557551937e-05,
+      "loss": 0.0619,
+      "step": 17044
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.6013007673591615e-05,
+      "loss": 0.0704,
+      "step": 17046
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.601253977166386e-05,
+      "loss": 0.0464,
+      "step": 17048
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.601207186973611e-05,
+      "loss": 0.065,
+      "step": 17050
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.6011603967808346e-05,
+      "loss": 0.0587,
+      "step": 17052
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.601113606588059e-05,
+      "loss": 0.0758,
+      "step": 17054
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.601066816395284e-05,
+      "loss": 0.0538,
+      "step": 17056
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.6010200262025084e-05,
+      "loss": 0.0675,
+      "step": 17058
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.600973236009732e-05,
+      "loss": 0.0759,
+      "step": 17060
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.600926445816957e-05,
+      "loss": 0.0533,
+      "step": 17062
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.6008796556241815e-05,
+      "loss": 0.0525,
+      "step": 17064
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.600832865431406e-05,
+      "loss": 0.0541,
+      "step": 17066
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.60078607523863e-05,
+      "loss": 0.0637,
+      "step": 17068
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.6007392850458546e-05,
+      "loss": 0.0586,
+      "step": 17070
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.600692494853079e-05,
+      "loss": 0.072,
+      "step": 17072
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.600645704660304e-05,
+      "loss": 0.0754,
+      "step": 17074
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.600598914467528e-05,
+      "loss": 0.0752,
+      "step": 17076
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.600552124274752e-05,
+      "loss": 0.0705,
+      "step": 17078
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.600505334081976e-05,
+      "loss": 0.0503,
+      "step": 17080
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.6004585438892014e-05,
+      "loss": 0.0637,
+      "step": 17082
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.6004117536964254e-05,
+      "loss": 0.0715,
+      "step": 17084
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.60036496350365e-05,
+      "loss": 0.057,
+      "step": 17086
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.600318173310874e-05,
+      "loss": 0.0567,
+      "step": 17088
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.600271383118099e-05,
+      "loss": 0.0731,
+      "step": 17090
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.600224592925323e-05,
+      "loss": 0.0605,
+      "step": 17092
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.6001778027325476e-05,
+      "loss": 0.0607,
+      "step": 17094
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.6001310125397716e-05,
+      "loss": 0.0753,
+      "step": 17096
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.600084222346996e-05,
+      "loss": 0.0527,
+      "step": 17098
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.600037432154221e-05,
+      "loss": 0.0481,
+      "step": 17100
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.599990641961445e-05,
+      "loss": 0.0617,
+      "step": 17102
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.599943851768669e-05,
+      "loss": 0.0761,
+      "step": 17104
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.599897061575894e-05,
+      "loss": 0.0574,
+      "step": 17106
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.5998502713831184e-05,
+      "loss": 0.0524,
+      "step": 17108
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.599803481190343e-05,
+      "loss": 0.0573,
+      "step": 17110
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.599756690997567e-05,
+      "loss": 0.0556,
+      "step": 17112
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.5997099008047915e-05,
+      "loss": 0.0389,
+      "step": 17114
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.599663110612016e-05,
+      "loss": 0.0475,
+      "step": 17116
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.599616320419241e-05,
+      "loss": 0.0517,
+      "step": 17118
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.5995695302264646e-05,
+      "loss": 0.0611,
+      "step": 17120
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.599522740033689e-05,
+      "loss": 0.054,
+      "step": 17122
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.599475949840913e-05,
+      "loss": 0.0561,
+      "step": 17124
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.5994291596481384e-05,
+      "loss": 0.072,
+      "step": 17126
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.599382369455362e-05,
+      "loss": 0.054,
+      "step": 17128
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.599335579262587e-05,
+      "loss": 0.0766,
+      "step": 17130
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.599288789069811e-05,
+      "loss": 0.0794,
+      "step": 17132
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.599241998877036e-05,
+      "loss": 0.0664,
+      "step": 17134
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.59919520868426e-05,
+      "loss": 0.067,
+      "step": 17136
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.5991484184914846e-05,
+      "loss": 0.0645,
+      "step": 17138
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.5991016282987085e-05,
+      "loss": 0.0591,
+      "step": 17140
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.599054838105933e-05,
+      "loss": 0.0638,
+      "step": 17142
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.599008047913158e-05,
+      "loss": 0.0713,
+      "step": 17144
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.598961257720382e-05,
+      "loss": 0.061,
+      "step": 17146
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.598914467527606e-05,
+      "loss": 0.0592,
+      "step": 17148
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.598867677334831e-05,
+      "loss": 0.0666,
+      "step": 17150
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.5988208871420554e-05,
+      "loss": 0.0607,
+      "step": 17152
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.59877409694928e-05,
+      "loss": 0.0546,
+      "step": 17154
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.598727306756504e-05,
+      "loss": 0.0609,
+      "step": 17156
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.5986805165637285e-05,
+      "loss": 0.0678,
+      "step": 17158
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.598633726370953e-05,
+      "loss": 0.0629,
+      "step": 17160
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.5985869361781776e-05,
+      "loss": 0.0774,
+      "step": 17162
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.5985401459854016e-05,
+      "loss": 0.0653,
+      "step": 17164
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.598493355792626e-05,
+      "loss": 0.0508,
+      "step": 17166
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.598446565599851e-05,
+      "loss": 0.0537,
+      "step": 17168
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.598399775407075e-05,
+      "loss": 0.0452,
+      "step": 17170
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.598352985214299e-05,
+      "loss": 0.0739,
+      "step": 17172
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.598306195021524e-05,
+      "loss": 0.0569,
+      "step": 17174
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.598259404828748e-05,
+      "loss": 0.0488,
+      "step": 17176
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.598212614635973e-05,
+      "loss": 0.0352,
+      "step": 17178
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.598165824443197e-05,
+      "loss": 0.0631,
+      "step": 17180
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.5981190342504215e-05,
+      "loss": 0.0624,
+      "step": 17182
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.5980722440576454e-05,
+      "loss": 0.0418,
+      "step": 17184
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.598025453864871e-05,
+      "loss": 0.0597,
+      "step": 17186
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.5979786636720946e-05,
+      "loss": 0.0612,
+      "step": 17188
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.597931873479319e-05,
+      "loss": 0.0543,
+      "step": 17190
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.597885083286543e-05,
+      "loss": 0.0512,
+      "step": 17192
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.597838293093768e-05,
+      "loss": 0.0516,
+      "step": 17194
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.597791502900992e-05,
+      "loss": 0.0556,
+      "step": 17196
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.597744712708217e-05,
+      "loss": 0.0564,
+      "step": 17198
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.597697922515441e-05,
+      "loss": 0.044,
+      "step": 17200
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.5976511323226654e-05,
+      "loss": 0.078,
+      "step": 17202
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.59760434212989e-05,
+      "loss": 0.0606,
+      "step": 17204
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.5975575519371146e-05,
+      "loss": 0.0552,
+      "step": 17206
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.5975107617443385e-05,
+      "loss": 0.0616,
+      "step": 17208
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.597463971551563e-05,
+      "loss": 0.0567,
+      "step": 17210
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.597417181358788e-05,
+      "loss": 0.0598,
+      "step": 17212
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.597370391166012e-05,
+      "loss": 0.0723,
+      "step": 17214
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.597323600973236e-05,
+      "loss": 0.0473,
+      "step": 17216
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.597276810780461e-05,
+      "loss": 0.0779,
+      "step": 17218
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.5972300205876854e-05,
+      "loss": 0.0689,
+      "step": 17220
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.59718323039491e-05,
+      "loss": 0.0804,
+      "step": 17222
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.597136440202134e-05,
+      "loss": 0.0717,
+      "step": 17224
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.597089650009358e-05,
+      "loss": 0.0595,
+      "step": 17226
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.5970428598165824e-05,
+      "loss": 0.055,
+      "step": 17228
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.596996069623807e-05,
+      "loss": 0.0618,
+      "step": 17230
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.5969492794310316e-05,
+      "loss": 0.0858,
+      "step": 17232
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.5969024892382555e-05,
+      "loss": 0.0511,
+      "step": 17234
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.59685569904548e-05,
+      "loss": 0.0913,
+      "step": 17236
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.5968089088527047e-05,
+      "loss": 0.0728,
+      "step": 17238
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.596762118659929e-05,
+      "loss": 0.0665,
+      "step": 17240
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.596715328467153e-05,
+      "loss": 0.0573,
+      "step": 17242
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.596668538274378e-05,
+      "loss": 0.0545,
+      "step": 17244
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.5966217480816023e-05,
+      "loss": 0.0544,
+      "step": 17246
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.596574957888827e-05,
+      "loss": 0.062,
+      "step": 17248
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.596528167696051e-05,
+      "loss": 0.0514,
+      "step": 17250
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.5964813775032754e-05,
+      "loss": 0.061,
+      "step": 17252
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.5964345873105e-05,
+      "loss": 0.0646,
+      "step": 17254
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.5963877971177246e-05,
+      "loss": 0.0504,
+      "step": 17256
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.5963410069249485e-05,
+      "loss": 0.0726,
+      "step": 17258
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.596294216732173e-05,
+      "loss": 0.054,
+      "step": 17260
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.596247426539397e-05,
+      "loss": 0.0761,
+      "step": 17262
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.596200636346622e-05,
+      "loss": 0.0594,
+      "step": 17264
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.596153846153846e-05,
+      "loss": 0.0682,
+      "step": 17266
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.596107055961071e-05,
+      "loss": 0.0634,
+      "step": 17268
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.596060265768295e-05,
+      "loss": 0.0604,
+      "step": 17270
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.596013475575519e-05,
+      "loss": 0.0801,
+      "step": 17272
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.595966685382744e-05,
+      "loss": 0.0758,
+      "step": 17274
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.5959198951899685e-05,
+      "loss": 0.0713,
+      "step": 17276
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.5958731049971924e-05,
+      "loss": 0.0481,
+      "step": 17278
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.595826314804417e-05,
+      "loss": 0.0451,
+      "step": 17280
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.5957795246116416e-05,
+      "loss": 0.0611,
+      "step": 17282
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.595732734418866e-05,
+      "loss": 0.0627,
+      "step": 17284
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.59568594422609e-05,
+      "loss": 0.0542,
+      "step": 17286
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.595639154033315e-05,
+      "loss": 0.054,
+      "step": 17288
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.595592363840539e-05,
+      "loss": 0.0746,
+      "step": 17290
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.595545573647764e-05,
+      "loss": 0.0646,
+      "step": 17292
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.595498783454988e-05,
+      "loss": 0.0576,
+      "step": 17294
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.5954519932622124e-05,
+      "loss": 0.0434,
+      "step": 17296
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.595405203069437e-05,
+      "loss": 0.0592,
+      "step": 17298
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.5953584128766616e-05,
+      "loss": 0.0621,
+      "step": 17300
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.5953116226838855e-05,
+      "loss": 0.0599,
+      "step": 17302
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.59526483249111e-05,
+      "loss": 0.0931,
+      "step": 17304
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.595218042298334e-05,
+      "loss": 0.0755,
+      "step": 17306
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.595171252105559e-05,
+      "loss": 0.0653,
+      "step": 17308
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.595124461912783e-05,
+      "loss": 0.0642,
+      "step": 17310
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.595077671720008e-05,
+      "loss": 0.0703,
+      "step": 17312
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.595030881527232e-05,
+      "loss": 0.0717,
+      "step": 17314
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.594984091334457e-05,
+      "loss": 0.0565,
+      "step": 17316
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.594937301141681e-05,
+      "loss": 0.066,
+      "step": 17318
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.5948905109489054e-05,
+      "loss": 0.0815,
+      "step": 17320
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.5948437207561294e-05,
+      "loss": 0.0583,
+      "step": 17322
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.594796930563354e-05,
+      "loss": 0.0635,
+      "step": 17324
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.5947501403705785e-05,
+      "loss": 0.0643,
+      "step": 17326
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.594703350177803e-05,
+      "loss": 0.0534,
+      "step": 17328
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.594656559985027e-05,
+      "loss": 0.0778,
+      "step": 17330
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.5946097697922516e-05,
+      "loss": 0.0661,
+      "step": 17332
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.594562979599476e-05,
+      "loss": 0.0731,
+      "step": 17334
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.594516189406701e-05,
+      "loss": 0.0527,
+      "step": 17336
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.594469399213925e-05,
+      "loss": 0.0644,
+      "step": 17338
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.594422609021149e-05,
+      "loss": 0.0626,
+      "step": 17340
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.594375818828374e-05,
+      "loss": 0.0721,
+      "step": 17342
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.5943290286355985e-05,
+      "loss": 0.0726,
+      "step": 17344
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.5942822384428224e-05,
+      "loss": 0.0613,
+      "step": 17346
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.594235448250047e-05,
+      "loss": 0.0407,
+      "step": 17348
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.5941886580572716e-05,
+      "loss": 0.0511,
+      "step": 17350
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.594141867864496e-05,
+      "loss": 0.0644,
+      "step": 17352
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.59409507767172e-05,
+      "loss": 0.0679,
+      "step": 17354
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.594048287478945e-05,
+      "loss": 0.0597,
+      "step": 17356
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.5940014972861686e-05,
+      "loss": 0.0555,
+      "step": 17358
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.593954707093394e-05,
+      "loss": 0.0446,
+      "step": 17360
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.593907916900618e-05,
+      "loss": 0.0539,
+      "step": 17362
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.5938611267078424e-05,
+      "loss": 0.0805,
+      "step": 17364
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.593814336515066e-05,
+      "loss": 0.0582,
+      "step": 17366
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.5937675463222916e-05,
+      "loss": 0.0748,
+      "step": 17368
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.5937207561295155e-05,
+      "loss": 0.0731,
+      "step": 17370
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.59367396593674e-05,
+      "loss": 0.0769,
+      "step": 17372
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.593627175743964e-05,
+      "loss": 0.0725,
+      "step": 17374
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.5935803855511886e-05,
+      "loss": 0.0608,
+      "step": 17376
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.593533595358413e-05,
+      "loss": 0.0682,
+      "step": 17378
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.593486805165638e-05,
+      "loss": 0.0711,
+      "step": 17380
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.593440014972862e-05,
+      "loss": 0.05,
+      "step": 17382
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.593393224780086e-05,
+      "loss": 0.0613,
+      "step": 17384
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.593346434587311e-05,
+      "loss": 0.0728,
+      "step": 17386
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.5932996443945354e-05,
+      "loss": 0.0447,
+      "step": 17388
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.5932528542017594e-05,
+      "loss": 0.0402,
+      "step": 17390
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.593206064008984e-05,
+      "loss": 0.0498,
+      "step": 17392
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.5931592738162085e-05,
+      "loss": 0.0553,
+      "step": 17394
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.593112483623433e-05,
+      "loss": 0.0556,
+      "step": 17396
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.593065693430657e-05,
+      "loss": 0.0628,
+      "step": 17398
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.5930189032378816e-05,
+      "loss": 0.0647,
+      "step": 17400
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.5929721130451055e-05,
+      "loss": 0.0459,
+      "step": 17402
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.592925322852331e-05,
+      "loss": 0.0442,
+      "step": 17404
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.592878532659555e-05,
+      "loss": 0.0455,
+      "step": 17406
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.592831742466779e-05,
+      "loss": 0.0435,
+      "step": 17408
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.592784952274003e-05,
+      "loss": 0.0642,
+      "step": 17410
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.5927381620812285e-05,
+      "loss": 0.0517,
+      "step": 17412
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.5926913718884524e-05,
+      "loss": 0.0332,
+      "step": 17414
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.592644581695677e-05,
+      "loss": 0.0744,
+      "step": 17416
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.592597791502901e-05,
+      "loss": 0.0637,
+      "step": 17418
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.5925510013101255e-05,
+      "loss": 0.0749,
+      "step": 17420
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.59250421111735e-05,
+      "loss": 0.0784,
+      "step": 17422
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.592457420924575e-05,
+      "loss": 0.0613,
+      "step": 17424
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.5924106307317986e-05,
+      "loss": 0.0633,
+      "step": 17426
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.592363840539023e-05,
+      "loss": 0.0738,
+      "step": 17428
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.592317050346248e-05,
+      "loss": 0.0553,
+      "step": 17430
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.5922702601534724e-05,
+      "loss": 0.0489,
+      "step": 17432
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.592223469960696e-05,
+      "loss": 0.0674,
+      "step": 17434
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.592176679767921e-05,
+      "loss": 0.0708,
+      "step": 17436
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.5921298895751455e-05,
+      "loss": 0.0416,
+      "step": 17438
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.59208309938237e-05,
+      "loss": 0.0772,
+      "step": 17440
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.592036309189594e-05,
+      "loss": 0.0666,
+      "step": 17442
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.5919895189968186e-05,
+      "loss": 0.0476,
+      "step": 17444
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.591942728804043e-05,
+      "loss": 0.0762,
+      "step": 17446
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.591895938611268e-05,
+      "loss": 0.0708,
+      "step": 17448
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.591849148418492e-05,
+      "loss": 0.0836,
+      "step": 17450
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.591802358225716e-05,
+      "loss": 0.0539,
+      "step": 17452
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.59175556803294e-05,
+      "loss": 0.0663,
+      "step": 17454
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.5917087778401654e-05,
+      "loss": 0.0752,
+      "step": 17456
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.5916619876473894e-05,
+      "loss": 0.0744,
+      "step": 17458
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.591615197454614e-05,
+      "loss": 0.0574,
+      "step": 17460
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.591568407261838e-05,
+      "loss": 0.0496,
+      "step": 17462
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.591521617069063e-05,
+      "loss": 0.0725,
+      "step": 17464
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.591474826876287e-05,
+      "loss": 0.0516,
+      "step": 17466
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.5914280366835116e-05,
+      "loss": 0.0559,
+      "step": 17468
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.5913812464907355e-05,
+      "loss": 0.0705,
+      "step": 17470
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.59133445629796e-05,
+      "loss": 0.0542,
+      "step": 17472
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.591287666105185e-05,
+      "loss": 0.088,
+      "step": 17474
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.591240875912409e-05,
+      "loss": 0.0651,
+      "step": 17476
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.591194085719633e-05,
+      "loss": 0.0428,
+      "step": 17478
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.591147295526858e-05,
+      "loss": 0.0498,
+      "step": 17480
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.5911005053340824e-05,
+      "loss": 0.0568,
+      "step": 17482
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.591053715141306e-05,
+      "loss": 0.0718,
+      "step": 17484
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.591006924948531e-05,
+      "loss": 0.0672,
+      "step": 17486
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.590960134755755e-05,
+      "loss": 0.0563,
+      "step": 17488
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.59091334456298e-05,
+      "loss": 0.0661,
+      "step": 17490
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.590866554370204e-05,
+      "loss": 0.0686,
+      "step": 17492
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.5908197641774286e-05,
+      "loss": 0.0484,
+      "step": 17494
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.5907729739846525e-05,
+      "loss": 0.0558,
+      "step": 17496
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.590726183791878e-05,
+      "loss": 0.0628,
+      "step": 17498
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.590679393599102e-05,
+      "loss": 0.0566,
+      "step": 17500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.590632603406326e-05,
+      "loss": 0.0517,
+      "step": 17502
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.59058581321355e-05,
+      "loss": 0.0612,
+      "step": 17504
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.590539023020775e-05,
+      "loss": 0.0634,
+      "step": 17506
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.5904922328279994e-05,
+      "loss": 0.049,
+      "step": 17508
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.590445442635224e-05,
+      "loss": 0.0839,
+      "step": 17510
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.590398652442448e-05,
+      "loss": 0.0552,
+      "step": 17512
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.5903518622496725e-05,
+      "loss": 0.0704,
+      "step": 17514
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.590305072056897e-05,
+      "loss": 0.0565,
+      "step": 17516
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.590258281864122e-05,
+      "loss": 0.0716,
+      "step": 17518
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.5902114916713456e-05,
+      "loss": 0.0609,
+      "step": 17520
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.59016470147857e-05,
+      "loss": 0.0788,
+      "step": 17522
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.590117911285795e-05,
+      "loss": 0.058,
+      "step": 17524
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.5900711210930194e-05,
+      "loss": 0.0572,
+      "step": 17526
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.590024330900243e-05,
+      "loss": 0.0633,
+      "step": 17528
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.589977540707468e-05,
+      "loss": 0.086,
+      "step": 17530
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.5899307505146925e-05,
+      "loss": 0.0699,
+      "step": 17532
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.589883960321917e-05,
+      "loss": 0.1082,
+      "step": 17534
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.589837170129141e-05,
+      "loss": 0.0465,
+      "step": 17536
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.5897903799363655e-05,
+      "loss": 0.0599,
+      "step": 17538
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.5897435897435895e-05,
+      "loss": 0.0774,
+      "step": 17540
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.589696799550815e-05,
+      "loss": 0.0616,
+      "step": 17542
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.5896500093580386e-05,
+      "loss": 0.0889,
+      "step": 17544
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.589603219165263e-05,
+      "loss": 0.0605,
+      "step": 17546
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.589556428972487e-05,
+      "loss": 0.0632,
+      "step": 17548
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.589509638779712e-05,
+      "loss": 0.0577,
+      "step": 17550
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.589462848586936e-05,
+      "loss": 0.0548,
+      "step": 17552
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.589416058394161e-05,
+      "loss": 0.0549,
+      "step": 17554
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.589369268201385e-05,
+      "loss": 0.0402,
+      "step": 17556
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.5893224780086094e-05,
+      "loss": 0.0613,
+      "step": 17558
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.589275687815834e-05,
+      "loss": 0.076,
+      "step": 17560
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5892288976230586e-05,
+      "loss": 0.0689,
+      "step": 17562
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5891821074302825e-05,
+      "loss": 0.0634,
+      "step": 17564
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.589135317237507e-05,
+      "loss": 0.0766,
+      "step": 17566
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.589088527044732e-05,
+      "loss": 0.0475,
+      "step": 17568
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.589041736851956e-05,
+      "loss": 0.0632,
+      "step": 17570
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.58899494665918e-05,
+      "loss": 0.0821,
+      "step": 17572
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.588948156466405e-05,
+      "loss": 0.053,
+      "step": 17574
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5889013662736294e-05,
+      "loss": 0.0832,
+      "step": 17576
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.588854576080854e-05,
+      "loss": 0.0941,
+      "step": 17578
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.588807785888078e-05,
+      "loss": 0.0644,
+      "step": 17580
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5887609956953025e-05,
+      "loss": 0.0533,
+      "step": 17582
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5887142055025264e-05,
+      "loss": 0.0699,
+      "step": 17584
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.588667415309752e-05,
+      "loss": 0.064,
+      "step": 17586
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5886206251169756e-05,
+      "loss": 0.0677,
+      "step": 17588
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5885738349242e-05,
+      "loss": 0.0468,
+      "step": 17590
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.588527044731424e-05,
+      "loss": 0.0669,
+      "step": 17592
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5884802545386494e-05,
+      "loss": 0.0693,
+      "step": 17594
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.588433464345873e-05,
+      "loss": 0.0475,
+      "step": 17596
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.588386674153098e-05,
+      "loss": 0.0922,
+      "step": 17598
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.588339883960322e-05,
+      "loss": 0.0714,
+      "step": 17600
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5882930937675464e-05,
+      "loss": 0.0506,
+      "step": 17602
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.588246303574771e-05,
+      "loss": 0.056,
+      "step": 17604
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5881995133819956e-05,
+      "loss": 0.0527,
+      "step": 17606
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5881527231892195e-05,
+      "loss": 0.0761,
+      "step": 17608
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.588105932996444e-05,
+      "loss": 0.0497,
+      "step": 17610
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5880591428036686e-05,
+      "loss": 0.0687,
+      "step": 17612
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.588012352610893e-05,
+      "loss": 0.0524,
+      "step": 17614
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.587965562418117e-05,
+      "loss": 0.0633,
+      "step": 17616
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.587918772225342e-05,
+      "loss": 0.0666,
+      "step": 17618
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.587871982032566e-05,
+      "loss": 0.0597,
+      "step": 17620
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.587825191839791e-05,
+      "loss": 0.0489,
+      "step": 17622
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.587778401647015e-05,
+      "loss": 0.0527,
+      "step": 17624
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5877316114542394e-05,
+      "loss": 0.0649,
+      "step": 17626
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.587684821261464e-05,
+      "loss": 0.0676,
+      "step": 17628
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.5876380310686886e-05,
+      "loss": 0.0773,
+      "step": 17630
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.5875912408759125e-05,
+      "loss": 0.0742,
+      "step": 17632
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.587544450683137e-05,
+      "loss": 0.0586,
+      "step": 17634
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.587497660490361e-05,
+      "loss": 0.0665,
+      "step": 17636
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.587450870297586e-05,
+      "loss": 0.061,
+      "step": 17638
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.58740408010481e-05,
+      "loss": 0.0697,
+      "step": 17640
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.587357289912035e-05,
+      "loss": 0.0759,
+      "step": 17642
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.587310499719259e-05,
+      "loss": 0.078,
+      "step": 17644
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.587263709526484e-05,
+      "loss": 0.0636,
+      "step": 17646
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.587216919333708e-05,
+      "loss": 0.0553,
+      "step": 17648
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.5871701291409325e-05,
+      "loss": 0.0655,
+      "step": 17650
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.5871233389481564e-05,
+      "loss": 0.0632,
+      "step": 17652
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.587076548755381e-05,
+      "loss": 0.0642,
+      "step": 17654
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.5870297585626056e-05,
+      "loss": 0.0557,
+      "step": 17656
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.58698296836983e-05,
+      "loss": 0.0645,
+      "step": 17658
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.586936178177054e-05,
+      "loss": 0.0706,
+      "step": 17660
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.586889387984279e-05,
+      "loss": 0.0511,
+      "step": 17662
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.586842597791503e-05,
+      "loss": 0.059,
+      "step": 17664
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.586795807598728e-05,
+      "loss": 0.0614,
+      "step": 17666
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.586749017405952e-05,
+      "loss": 0.046,
+      "step": 17668
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.5867022272131764e-05,
+      "loss": 0.0649,
+      "step": 17670
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.586655437020401e-05,
+      "loss": 0.0633,
+      "step": 17672
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.5866086468276256e-05,
+      "loss": 0.0494,
+      "step": 17674
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.5865618566348495e-05,
+      "loss": 0.0504,
+      "step": 17676
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.586515066442074e-05,
+      "loss": 0.0646,
+      "step": 17678
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.5864682762492986e-05,
+      "loss": 0.0582,
+      "step": 17680
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.586421486056523e-05,
+      "loss": 0.0574,
+      "step": 17682
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.586374695863747e-05,
+      "loss": 0.0904,
+      "step": 17684
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.586327905670972e-05,
+      "loss": 0.0734,
+      "step": 17686
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.5862811154781957e-05,
+      "loss": 0.0562,
+      "step": 17688
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.586234325285421e-05,
+      "loss": 0.0643,
+      "step": 17690
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.586187535092645e-05,
+      "loss": 0.0668,
+      "step": 17692
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.5861407448998694e-05,
+      "loss": 0.0445,
+      "step": 17694
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.5860939547070933e-05,
+      "loss": 0.071,
+      "step": 17696
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.586047164514318e-05,
+      "loss": 0.0678,
+      "step": 17698
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.5860003743215425e-05,
+      "loss": 0.0446,
+      "step": 17700
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.585953584128767e-05,
+      "loss": 0.0642,
+      "step": 17702
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.585906793935991e-05,
+      "loss": 0.0502,
+      "step": 17704
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.5858600037432156e-05,
+      "loss": 0.0481,
+      "step": 17706
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.58581321355044e-05,
+      "loss": 0.0736,
+      "step": 17708
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.585766423357665e-05,
+      "loss": 0.0674,
+      "step": 17710
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.585719633164889e-05,
+      "loss": 0.0467,
+      "step": 17712
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.585672842972113e-05,
+      "loss": 0.0636,
+      "step": 17714
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.585626052779338e-05,
+      "loss": 0.0785,
+      "step": 17716
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.5855792625865625e-05,
+      "loss": 0.0584,
+      "step": 17718
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.5855324723937864e-05,
+      "loss": 0.0388,
+      "step": 17720
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.585485682201011e-05,
+      "loss": 0.0692,
+      "step": 17722
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.5854388920082356e-05,
+      "loss": 0.0619,
+      "step": 17724
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.58539210181546e-05,
+      "loss": 0.0624,
+      "step": 17726
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.585345311622684e-05,
+      "loss": 0.049,
+      "step": 17728
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.585298521429908e-05,
+      "loss": 0.0452,
+      "step": 17730
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.5852517312371326e-05,
+      "loss": 0.0644,
+      "step": 17732
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.585204941044357e-05,
+      "loss": 0.0575,
+      "step": 17734
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.585158150851582e-05,
+      "loss": 0.0643,
+      "step": 17736
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.585111360658806e-05,
+      "loss": 0.0657,
+      "step": 17738
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.58506457046603e-05,
+      "loss": 0.0603,
+      "step": 17740
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.585017780273255e-05,
+      "loss": 0.0641,
+      "step": 17742
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.5849709900804795e-05,
+      "loss": 0.0647,
+      "step": 17744
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.5849241998877034e-05,
+      "loss": 0.045,
+      "step": 17746
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.584877409694928e-05,
+      "loss": 0.0628,
+      "step": 17748
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.5848306195021526e-05,
+      "loss": 0.0619,
+      "step": 17750
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.584783829309377e-05,
+      "loss": 0.0657,
+      "step": 17752
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.584737039116601e-05,
+      "loss": 0.0645,
+      "step": 17754
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.5846902489238257e-05,
+      "loss": 0.0772,
+      "step": 17756
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.58464345873105e-05,
+      "loss": 0.0545,
+      "step": 17758
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.584596668538275e-05,
+      "loss": 0.0629,
+      "step": 17760
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.584549878345499e-05,
+      "loss": 0.088,
+      "step": 17762
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.5845030881527233e-05,
+      "loss": 0.0606,
+      "step": 17764
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.584456297959947e-05,
+      "loss": 0.0589,
+      "step": 17766
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.5844095077671725e-05,
+      "loss": 0.057,
+      "step": 17768
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.5843627175743964e-05,
+      "loss": 0.0668,
+      "step": 17770
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.584315927381621e-05,
+      "loss": 0.0618,
+      "step": 17772
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.584269137188845e-05,
+      "loss": 0.0636,
+      "step": 17774
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.58422234699607e-05,
+      "loss": 0.0617,
+      "step": 17776
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.584175556803294e-05,
+      "loss": 0.0615,
+      "step": 17778
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.584128766610519e-05,
+      "loss": 0.0678,
+      "step": 17780
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.5840819764177426e-05,
+      "loss": 0.0594,
+      "step": 17782
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.584035186224967e-05,
+      "loss": 0.051,
+      "step": 17784
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.583988396032192e-05,
+      "loss": 0.0807,
+      "step": 17786
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.5839416058394164e-05,
+      "loss": 0.0593,
+      "step": 17788
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.58389481564664e-05,
+      "loss": 0.0637,
+      "step": 17790
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.583848025453865e-05,
+      "loss": 0.0318,
+      "step": 17792
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.5838012352610895e-05,
+      "loss": 0.0564,
+      "step": 17794
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.583754445068314e-05,
+      "loss": 0.0575,
+      "step": 17796
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.583707654875538e-05,
+      "loss": 0.078,
+      "step": 17798
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.5836608646827626e-05,
+      "loss": 0.0557,
+      "step": 17800
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.583614074489987e-05,
+      "loss": 0.0446,
+      "step": 17802
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.583567284297212e-05,
+      "loss": 0.0582,
+      "step": 17804
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.583520494104436e-05,
+      "loss": 0.0591,
+      "step": 17806
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.58347370391166e-05,
+      "loss": 0.0552,
+      "step": 17808
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.583426913718885e-05,
+      "loss": 0.0943,
+      "step": 17810
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.5833801235261095e-05,
+      "loss": 0.0671,
+      "step": 17812
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 0.0914,
+      "step": 17814
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.583286543140558e-05,
+      "loss": 0.0672,
+      "step": 17816
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.583239752947782e-05,
+      "loss": 0.0605,
+      "step": 17818
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.583192962755007e-05,
+      "loss": 0.0664,
+      "step": 17820
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.583146172562231e-05,
+      "loss": 0.05,
+      "step": 17822
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.583099382369456e-05,
+      "loss": 0.0583,
+      "step": 17824
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.5830525921766796e-05,
+      "loss": 0.0592,
+      "step": 17826
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.583005801983904e-05,
+      "loss": 0.0468,
+      "step": 17828
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.582959011791129e-05,
+      "loss": 0.0662,
+      "step": 17830
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.5829122215983533e-05,
+      "loss": 0.0739,
+      "step": 17832
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.582865431405577e-05,
+      "loss": 0.0741,
+      "step": 17834
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.582818641212802e-05,
+      "loss": 0.0662,
+      "step": 17836
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.5827718510200264e-05,
+      "loss": 0.0741,
+      "step": 17838
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.582725060827251e-05,
+      "loss": 0.085,
+      "step": 17840
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.582678270634475e-05,
+      "loss": 0.0626,
+      "step": 17842
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.5826314804416995e-05,
+      "loss": 0.0599,
+      "step": 17844
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.582584690248924e-05,
+      "loss": 0.0551,
+      "step": 17846
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.582537900056149e-05,
+      "loss": 0.058,
+      "step": 17848
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.5824911098633726e-05,
+      "loss": 0.0534,
+      "step": 17850
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.582444319670597e-05,
+      "loss": 0.05,
+      "step": 17852
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.582397529477822e-05,
+      "loss": 0.0659,
+      "step": 17854
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.5823507392850464e-05,
+      "loss": 0.0593,
+      "step": 17856
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.58230394909227e-05,
+      "loss": 0.0534,
+      "step": 17858
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.582257158899495e-05,
+      "loss": 0.072,
+      "step": 17860
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.582210368706719e-05,
+      "loss": 0.0592,
+      "step": 17862
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.582163578513944e-05,
+      "loss": 0.0706,
+      "step": 17864
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.582116788321168e-05,
+      "loss": 0.0623,
+      "step": 17866
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.5820699981283926e-05,
+      "loss": 0.0639,
+      "step": 17868
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.5820232079356165e-05,
+      "loss": 0.0638,
+      "step": 17870
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.581976417742842e-05,
+      "loss": 0.0713,
+      "step": 17872
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.581929627550066e-05,
+      "loss": 0.0725,
+      "step": 17874
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.58188283735729e-05,
+      "loss": 0.0629,
+      "step": 17876
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.581836047164514e-05,
+      "loss": 0.0673,
+      "step": 17878
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.581789256971739e-05,
+      "loss": 0.063,
+      "step": 17880
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.5817424667789634e-05,
+      "loss": 0.0706,
+      "step": 17882
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.581695676586188e-05,
+      "loss": 0.0729,
+      "step": 17884
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.581648886393412e-05,
+      "loss": 0.0563,
+      "step": 17886
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.5816020962006365e-05,
+      "loss": 0.058,
+      "step": 17888
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.581555306007861e-05,
+      "loss": 0.0682,
+      "step": 17890
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.581508515815086e-05,
+      "loss": 0.0483,
+      "step": 17892
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.5814617256223096e-05,
+      "loss": 0.0651,
+      "step": 17894
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.581414935429534e-05,
+      "loss": 0.0691,
+      "step": 17896
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.581368145236759e-05,
+      "loss": 0.0664,
+      "step": 17898
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.5813213550439834e-05,
+      "loss": 0.0722,
+      "step": 17900
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.581274564851207e-05,
+      "loss": 0.0618,
+      "step": 17902
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.581227774658432e-05,
+      "loss": 0.0641,
+      "step": 17904
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.5811809844656564e-05,
+      "loss": 0.0549,
+      "step": 17906
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.581134194272881e-05,
+      "loss": 0.0581,
+      "step": 17908
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.581087404080105e-05,
+      "loss": 0.0466,
+      "step": 17910
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.5810406138873295e-05,
+      "loss": 0.0632,
+      "step": 17912
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.5809938236945535e-05,
+      "loss": 0.0846,
+      "step": 17914
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.580947033501779e-05,
+      "loss": 0.0557,
+      "step": 17916
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5809002433090026e-05,
+      "loss": 0.0645,
+      "step": 17918
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.580853453116227e-05,
+      "loss": 0.0547,
+      "step": 17920
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.580806662923451e-05,
+      "loss": 0.083,
+      "step": 17922
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5807598727306764e-05,
+      "loss": 0.0525,
+      "step": 17924
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5807130825379e-05,
+      "loss": 0.0499,
+      "step": 17926
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.580666292345125e-05,
+      "loss": 0.0718,
+      "step": 17928
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.580619502152349e-05,
+      "loss": 0.0704,
+      "step": 17930
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5805727119595734e-05,
+      "loss": 0.055,
+      "step": 17932
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.580525921766798e-05,
+      "loss": 0.1309,
+      "step": 17934
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5804791315740226e-05,
+      "loss": 0.099,
+      "step": 17936
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5804323413812465e-05,
+      "loss": 0.0566,
+      "step": 17938
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.580385551188471e-05,
+      "loss": 0.0975,
+      "step": 17940
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.580338760995696e-05,
+      "loss": 0.0441,
+      "step": 17942
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.58029197080292e-05,
+      "loss": 0.0631,
+      "step": 17944
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.580245180610144e-05,
+      "loss": 0.0603,
+      "step": 17946
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.580198390417369e-05,
+      "loss": 0.0821,
+      "step": 17948
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5801516002245934e-05,
+      "loss": 0.0696,
+      "step": 17950
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.580104810031818e-05,
+      "loss": 0.0596,
+      "step": 17952
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.580058019839042e-05,
+      "loss": 0.0837,
+      "step": 17954
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5800112296462665e-05,
+      "loss": 0.0728,
+      "step": 17956
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.579964439453491e-05,
+      "loss": 0.0573,
+      "step": 17958
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.579917649260716e-05,
+      "loss": 0.0813,
+      "step": 17960
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5798708590679396e-05,
+      "loss": 0.0614,
+      "step": 17962
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.579824068875164e-05,
+      "loss": 0.0495,
+      "step": 17964
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.579777278682388e-05,
+      "loss": 0.0763,
+      "step": 17966
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5797304884896134e-05,
+      "loss": 0.0801,
+      "step": 17968
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.579683698296837e-05,
+      "loss": 0.0909,
+      "step": 17970
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.579636908104062e-05,
+      "loss": 0.0803,
+      "step": 17972
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.579590117911286e-05,
+      "loss": 0.0755,
+      "step": 17974
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5795433277185104e-05,
+      "loss": 0.0455,
+      "step": 17976
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.579496537525735e-05,
+      "loss": 0.0625,
+      "step": 17978
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5794497473329595e-05,
+      "loss": 0.0573,
+      "step": 17980
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5794029571401835e-05,
+      "loss": 0.0725,
+      "step": 17982
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.579356166947408e-05,
+      "loss": 0.0406,
+      "step": 17984
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5793093767546326e-05,
+      "loss": 0.0562,
+      "step": 17986
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.5792625865618566e-05,
+      "loss": 0.0727,
+      "step": 17988
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.579215796369081e-05,
+      "loss": 0.074,
+      "step": 17990
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.579169006176306e-05,
+      "loss": 0.0805,
+      "step": 17992
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.57912221598353e-05,
+      "loss": 0.0646,
+      "step": 17994
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.579075425790754e-05,
+      "loss": 0.0542,
+      "step": 17996
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.579028635597979e-05,
+      "loss": 0.0552,
+      "step": 17998
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.578981845405203e-05,
+      "loss": 0.0685,
+      "step": 18000
+    },
+    {
+      "epoch": 2.53,
+      "eval_gen_len": 29.9828,
+      "eval_loss": 1.0630050897598267,
+      "eval_meteor": 0.0517,
+      "eval_runtime": 14.711,
+      "eval_samples_per_second": 3.943,
+      "eval_steps_per_second": 0.544,
+      "step": 18000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.578935055212428e-05,
+      "loss": 0.0574,
+      "step": 18002
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.578888265019652e-05,
+      "loss": 0.0585,
+      "step": 18004
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.5788414748268765e-05,
+      "loss": 0.0534,
+      "step": 18006
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.5787946846341004e-05,
+      "loss": 0.0853,
+      "step": 18008
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.578747894441325e-05,
+      "loss": 0.0485,
+      "step": 18010
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.5787011042485496e-05,
+      "loss": 0.0588,
+      "step": 18012
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.578654314055774e-05,
+      "loss": 0.0653,
+      "step": 18014
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.578607523862998e-05,
+      "loss": 0.0611,
+      "step": 18016
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.578560733670223e-05,
+      "loss": 0.0585,
+      "step": 18018
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.578513943477447e-05,
+      "loss": 0.0494,
+      "step": 18020
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.578467153284672e-05,
+      "loss": 0.0579,
+      "step": 18022
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.578420363091896e-05,
+      "loss": 0.0788,
+      "step": 18024
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.5783735728991204e-05,
+      "loss": 0.0835,
+      "step": 18026
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.578326782706345e-05,
+      "loss": 0.0799,
+      "step": 18028
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.5782799925135696e-05,
+      "loss": 0.0779,
+      "step": 18030
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.5782332023207935e-05,
+      "loss": 0.0508,
+      "step": 18032
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.578186412128018e-05,
+      "loss": 0.0542,
+      "step": 18034
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.578139621935243e-05,
+      "loss": 0.0572,
+      "step": 18036
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.578092831742467e-05,
+      "loss": 0.0846,
+      "step": 18038
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.578046041549691e-05,
+      "loss": 0.1055,
+      "step": 18040
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.577999251356916e-05,
+      "loss": 0.0624,
+      "step": 18042
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.57795246116414e-05,
+      "loss": 0.0574,
+      "step": 18044
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.577905670971365e-05,
+      "loss": 0.0669,
+      "step": 18046
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.577858880778589e-05,
+      "loss": 0.0626,
+      "step": 18048
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.5778120905858135e-05,
+      "loss": 0.0663,
+      "step": 18050
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.5777653003930374e-05,
+      "loss": 0.061,
+      "step": 18052
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.5777185102002626e-05,
+      "loss": 0.0967,
+      "step": 18054
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.5776717200074866e-05,
+      "loss": 0.076,
+      "step": 18056
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.577624929814711e-05,
+      "loss": 0.0503,
+      "step": 18058
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.577578139621935e-05,
+      "loss": 0.0614,
+      "step": 18060
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.5775313494291597e-05,
+      "loss": 0.0655,
+      "step": 18062
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.577484559236384e-05,
+      "loss": 0.0796,
+      "step": 18064
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.577437769043609e-05,
+      "loss": 0.0709,
+      "step": 18066
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.577390978850833e-05,
+      "loss": 0.0553,
+      "step": 18068
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.577344188658057e-05,
+      "loss": 0.064,
+      "step": 18070
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.577297398465282e-05,
+      "loss": 0.0773,
+      "step": 18072
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.5772506082725065e-05,
+      "loss": 0.0511,
+      "step": 18074
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.5772038180797304e-05,
+      "loss": 0.0555,
+      "step": 18076
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.577157027886955e-05,
+      "loss": 0.0628,
+      "step": 18078
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.5771102376941796e-05,
+      "loss": 0.0613,
+      "step": 18080
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.577063447501404e-05,
+      "loss": 0.0681,
+      "step": 18082
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.577016657308628e-05,
+      "loss": 0.0565,
+      "step": 18084
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.576969867115853e-05,
+      "loss": 0.0705,
+      "step": 18086
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.576923076923077e-05,
+      "loss": 0.0486,
+      "step": 18088
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.576876286730302e-05,
+      "loss": 0.0643,
+      "step": 18090
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.576829496537526e-05,
+      "loss": 0.0604,
+      "step": 18092
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.5767827063447504e-05,
+      "loss": 0.0679,
+      "step": 18094
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.576735916151974e-05,
+      "loss": 0.0723,
+      "step": 18096
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.5766891259591996e-05,
+      "loss": 0.0516,
+      "step": 18098
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.5766423357664235e-05,
+      "loss": 0.0815,
+      "step": 18100
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.576595545573648e-05,
+      "loss": 0.0627,
+      "step": 18102
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.576548755380872e-05,
+      "loss": 0.0669,
+      "step": 18104
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.5765019651880966e-05,
+      "loss": 0.0467,
+      "step": 18106
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.576455174995321e-05,
+      "loss": 0.0538,
+      "step": 18108
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.576408384802546e-05,
+      "loss": 0.0598,
+      "step": 18110
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.57636159460977e-05,
+      "loss": 0.0831,
+      "step": 18112
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.576314804416994e-05,
+      "loss": 0.0829,
+      "step": 18114
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.576268014224219e-05,
+      "loss": 0.0795,
+      "step": 18116
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.5762212240314435e-05,
+      "loss": 0.0695,
+      "step": 18118
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.5761744338386674e-05,
+      "loss": 0.0852,
+      "step": 18120
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.576127643645892e-05,
+      "loss": 0.0471,
+      "step": 18122
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.5760808534531166e-05,
+      "loss": 0.0622,
+      "step": 18124
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.576034063260341e-05,
+      "loss": 0.0434,
+      "step": 18126
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.575987273067565e-05,
+      "loss": 0.0489,
+      "step": 18128
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.5759404828747897e-05,
+      "loss": 0.0932,
+      "step": 18130
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.575893692682014e-05,
+      "loss": 0.0676,
+      "step": 18132
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.575846902489239e-05,
+      "loss": 0.0738,
+      "step": 18134
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.575800112296463e-05,
+      "loss": 0.0719,
+      "step": 18136
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5757533221036873e-05,
+      "loss": 0.0558,
+      "step": 18138
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.575706531910911e-05,
+      "loss": 0.0841,
+      "step": 18140
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5756597417181365e-05,
+      "loss": 0.0758,
+      "step": 18142
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5756129515253604e-05,
+      "loss": 0.0637,
+      "step": 18144
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.575566161332585e-05,
+      "loss": 0.0691,
+      "step": 18146
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.575519371139809e-05,
+      "loss": 0.0624,
+      "step": 18148
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.575472580947034e-05,
+      "loss": 0.0743,
+      "step": 18150
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.575425790754258e-05,
+      "loss": 0.0611,
+      "step": 18152
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.575379000561483e-05,
+      "loss": 0.0608,
+      "step": 18154
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5753322103687066e-05,
+      "loss": 0.0768,
+      "step": 18156
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.575285420175931e-05,
+      "loss": 0.0476,
+      "step": 18158
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.575238629983156e-05,
+      "loss": 0.0837,
+      "step": 18160
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5751918397903804e-05,
+      "loss": 0.0747,
+      "step": 18162
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.575145049597604e-05,
+      "loss": 0.0707,
+      "step": 18164
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.575098259404829e-05,
+      "loss": 0.0619,
+      "step": 18166
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5750514692120535e-05,
+      "loss": 0.0645,
+      "step": 18168
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.575004679019278e-05,
+      "loss": 0.0577,
+      "step": 18170
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.574957888826502e-05,
+      "loss": 0.0545,
+      "step": 18172
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5749110986337266e-05,
+      "loss": 0.0641,
+      "step": 18174
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.574864308440951e-05,
+      "loss": 0.0938,
+      "step": 18176
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.574817518248176e-05,
+      "loss": 0.0725,
+      "step": 18178
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5747707280554e-05,
+      "loss": 0.075,
+      "step": 18180
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.574723937862624e-05,
+      "loss": 0.0707,
+      "step": 18182
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.574677147669849e-05,
+      "loss": 0.0726,
+      "step": 18184
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5746303574770735e-05,
+      "loss": 0.0506,
+      "step": 18186
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5745835672842974e-05,
+      "loss": 0.0632,
+      "step": 18188
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.574536777091522e-05,
+      "loss": 0.0544,
+      "step": 18190
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.574489986898746e-05,
+      "loss": 0.0564,
+      "step": 18192
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.574443196705971e-05,
+      "loss": 0.0785,
+      "step": 18194
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.574396406513195e-05,
+      "loss": 0.074,
+      "step": 18196
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5743496163204197e-05,
+      "loss": 0.0464,
+      "step": 18198
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5743028261276436e-05,
+      "loss": 0.0614,
+      "step": 18200
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.574256035934869e-05,
+      "loss": 0.0843,
+      "step": 18202
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.574209245742093e-05,
+      "loss": 0.0725,
+      "step": 18204
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5741624555493173e-05,
+      "loss": 0.0684,
+      "step": 18206
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.574115665356541e-05,
+      "loss": 0.0989,
+      "step": 18208
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.574068875163766e-05,
+      "loss": 0.0952,
+      "step": 18210
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5740220849709904e-05,
+      "loss": 0.0607,
+      "step": 18212
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.573975294778215e-05,
+      "loss": 0.0572,
+      "step": 18214
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.573928504585439e-05,
+      "loss": 0.0673,
+      "step": 18216
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5738817143926635e-05,
+      "loss": 0.0595,
+      "step": 18218
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.573834924199888e-05,
+      "loss": 0.0681,
+      "step": 18220
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.573788134007113e-05,
+      "loss": 0.0632,
+      "step": 18222
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5737413438143366e-05,
+      "loss": 0.0569,
+      "step": 18224
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.573694553621561e-05,
+      "loss": 0.0609,
+      "step": 18226
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.573647763428786e-05,
+      "loss": 0.052,
+      "step": 18228
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5736009732360104e-05,
+      "loss": 0.0567,
+      "step": 18230
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.573554183043234e-05,
+      "loss": 0.0725,
+      "step": 18232
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.573507392850459e-05,
+      "loss": 0.0439,
+      "step": 18234
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5734606026576835e-05,
+      "loss": 0.0471,
+      "step": 18236
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5734138124649074e-05,
+      "loss": 0.0728,
+      "step": 18238
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.573367022272132e-05,
+      "loss": 0.0785,
+      "step": 18240
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.573320232079356e-05,
+      "loss": 0.0747,
+      "step": 18242
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5732734418865805e-05,
+      "loss": 0.0579,
+      "step": 18244
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.573226651693805e-05,
+      "loss": 0.0665,
+      "step": 18246
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.57317986150103e-05,
+      "loss": 0.0481,
+      "step": 18248
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5731330713082536e-05,
+      "loss": 0.0711,
+      "step": 18250
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.573086281115478e-05,
+      "loss": 0.0555,
+      "step": 18252
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.573039490922703e-05,
+      "loss": 0.0842,
+      "step": 18254
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5729927007299274e-05,
+      "loss": 0.0768,
+      "step": 18256
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.572945910537151e-05,
+      "loss": 0.0584,
+      "step": 18258
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.572899120344376e-05,
+      "loss": 0.0669,
+      "step": 18260
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5728523301516005e-05,
+      "loss": 0.0967,
+      "step": 18262
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.572805539958825e-05,
+      "loss": 0.0714,
+      "step": 18264
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.572758749766049e-05,
+      "loss": 0.0476,
+      "step": 18266
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5727119595732736e-05,
+      "loss": 0.0612,
+      "step": 18268
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.572665169380498e-05,
+      "loss": 0.0443,
+      "step": 18270
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.572618379187723e-05,
+      "loss": 0.0566,
+      "step": 18272
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.572571588994947e-05,
+      "loss": 0.0596,
+      "step": 18274
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.572524798802171e-05,
+      "loss": 0.0519,
+      "step": 18276
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.572478008609395e-05,
+      "loss": 0.0782,
+      "step": 18278
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.5724312184166204e-05,
+      "loss": 0.0565,
+      "step": 18280
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.5723844282238444e-05,
+      "loss": 0.0611,
+      "step": 18282
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.572337638031069e-05,
+      "loss": 0.0419,
+      "step": 18284
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.572290847838293e-05,
+      "loss": 0.0639,
+      "step": 18286
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.5722440576455174e-05,
+      "loss": 0.0831,
+      "step": 18288
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.572197267452742e-05,
+      "loss": 0.0537,
+      "step": 18290
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.5721504772599666e-05,
+      "loss": 0.076,
+      "step": 18292
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.5721036870671905e-05,
+      "loss": 0.0811,
+      "step": 18294
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.572056896874415e-05,
+      "loss": 0.0534,
+      "step": 18296
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.57201010668164e-05,
+      "loss": 0.0726,
+      "step": 18298
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.571963316488864e-05,
+      "loss": 0.0908,
+      "step": 18300
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.571916526296088e-05,
+      "loss": 0.0847,
+      "step": 18302
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.571869736103313e-05,
+      "loss": 0.065,
+      "step": 18304
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.5718229459105374e-05,
+      "loss": 0.0583,
+      "step": 18306
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.571776155717762e-05,
+      "loss": 0.059,
+      "step": 18308
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.571729365524986e-05,
+      "loss": 0.0935,
+      "step": 18310
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.5716825753322105e-05,
+      "loss": 0.0507,
+      "step": 18312
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.571635785139435e-05,
+      "loss": 0.0744,
+      "step": 18314
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.57158899494666e-05,
+      "loss": 0.0534,
+      "step": 18316
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.5715422047538836e-05,
+      "loss": 0.0578,
+      "step": 18318
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.571495414561108e-05,
+      "loss": 0.0678,
+      "step": 18320
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.571448624368332e-05,
+      "loss": 0.0631,
+      "step": 18322
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.5714018341755574e-05,
+      "loss": 0.0579,
+      "step": 18324
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.571355043982781e-05,
+      "loss": 0.0568,
+      "step": 18326
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.571308253790006e-05,
+      "loss": 0.0613,
+      "step": 18328
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.57126146359723e-05,
+      "loss": 0.0549,
+      "step": 18330
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.571214673404455e-05,
+      "loss": 0.0559,
+      "step": 18332
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.571167883211679e-05,
+      "loss": 0.0579,
+      "step": 18334
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.5711210930189036e-05,
+      "loss": 0.0581,
+      "step": 18336
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.5710743028261275e-05,
+      "loss": 0.0759,
+      "step": 18338
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.571027512633352e-05,
+      "loss": 0.0619,
+      "step": 18340
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.570980722440577e-05,
+      "loss": 0.0932,
+      "step": 18342
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.570933932247801e-05,
+      "loss": 0.0834,
+      "step": 18344
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.570887142055025e-05,
+      "loss": 0.0916,
+      "step": 18346
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.57084035186225e-05,
+      "loss": 0.0794,
+      "step": 18348
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.5707935616694744e-05,
+      "loss": 0.0707,
+      "step": 18350
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.570746771476699e-05,
+      "loss": 0.0715,
+      "step": 18352
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.570699981283923e-05,
+      "loss": 0.0645,
+      "step": 18354
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.5706531910911475e-05,
+      "loss": 0.0676,
+      "step": 18356
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.570606400898372e-05,
+      "loss": 0.0562,
+      "step": 18358
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.5705596107055966e-05,
+      "loss": 0.0768,
+      "step": 18360
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.5705128205128205e-05,
+      "loss": 0.0724,
+      "step": 18362
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.570466030320045e-05,
+      "loss": 0.0937,
+      "step": 18364
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.57041924012727e-05,
+      "loss": 0.0621,
+      "step": 18366
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.570372449934494e-05,
+      "loss": 0.0561,
+      "step": 18368
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.570325659741718e-05,
+      "loss": 0.0608,
+      "step": 18370
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.570278869548943e-05,
+      "loss": 0.0528,
+      "step": 18372
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.570232079356167e-05,
+      "loss": 0.0754,
+      "step": 18374
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.570185289163392e-05,
+      "loss": 0.0526,
+      "step": 18376
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.570138498970616e-05,
+      "loss": 0.0686,
+      "step": 18378
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.5700917087778405e-05,
+      "loss": 0.0589,
+      "step": 18380
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.5700449185850644e-05,
+      "loss": 0.0836,
+      "step": 18382
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.56999812839229e-05,
+      "loss": 0.0691,
+      "step": 18384
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.5699513381995136e-05,
+      "loss": 0.0577,
+      "step": 18386
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.569904548006738e-05,
+      "loss": 0.0598,
+      "step": 18388
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.569857757813962e-05,
+      "loss": 0.0531,
+      "step": 18390
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.569810967621187e-05,
+      "loss": 0.0687,
+      "step": 18392
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.569764177428411e-05,
+      "loss": 0.0573,
+      "step": 18394
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.569717387235636e-05,
+      "loss": 0.0716,
+      "step": 18396
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.56967059704286e-05,
+      "loss": 0.0508,
+      "step": 18398
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.5696238068500844e-05,
+      "loss": 0.0861,
+      "step": 18400
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.569577016657309e-05,
+      "loss": 0.0575,
+      "step": 18402
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.5695302264645336e-05,
+      "loss": 0.0549,
+      "step": 18404
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.5694834362717575e-05,
+      "loss": 0.0513,
+      "step": 18406
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.569436646078982e-05,
+      "loss": 0.066,
+      "step": 18408
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.569389855886207e-05,
+      "loss": 0.0694,
+      "step": 18410
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.569343065693431e-05,
+      "loss": 0.0617,
+      "step": 18412
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.569296275500655e-05,
+      "loss": 0.046,
+      "step": 18414
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.56924948530788e-05,
+      "loss": 0.0726,
+      "step": 18416
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.569202695115104e-05,
+      "loss": 0.0718,
+      "step": 18418
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.569155904922329e-05,
+      "loss": 0.0521,
+      "step": 18420
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.569109114729553e-05,
+      "loss": 0.07,
+      "step": 18422
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.5690623245367775e-05,
+      "loss": 0.0855,
+      "step": 18424
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.5690155343440014e-05,
+      "loss": 0.0811,
+      "step": 18426
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.5689687441512266e-05,
+      "loss": 0.0491,
+      "step": 18428
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.5689219539584505e-05,
+      "loss": 0.0697,
+      "step": 18430
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.568875163765675e-05,
+      "loss": 0.0671,
+      "step": 18432
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.568828373572899e-05,
+      "loss": 0.0611,
+      "step": 18434
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.5687815833801236e-05,
+      "loss": 0.0682,
+      "step": 18436
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.568734793187348e-05,
+      "loss": 0.0622,
+      "step": 18438
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.568688002994573e-05,
+      "loss": 0.0688,
+      "step": 18440
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.568641212801797e-05,
+      "loss": 0.0682,
+      "step": 18442
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.568594422609021e-05,
+      "loss": 0.0585,
+      "step": 18444
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.568547632416246e-05,
+      "loss": 0.0631,
+      "step": 18446
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.5685008422234705e-05,
+      "loss": 0.0519,
+      "step": 18448
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.5684540520306944e-05,
+      "loss": 0.0781,
+      "step": 18450
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.568407261837919e-05,
+      "loss": 0.0722,
+      "step": 18452
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.5683604716451436e-05,
+      "loss": 0.0735,
+      "step": 18454
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.568313681452368e-05,
+      "loss": 0.0465,
+      "step": 18456
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.568266891259592e-05,
+      "loss": 0.0464,
+      "step": 18458
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.568220101066817e-05,
+      "loss": 0.0834,
+      "step": 18460
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.568173310874041e-05,
+      "loss": 0.0868,
+      "step": 18462
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.568126520681266e-05,
+      "loss": 0.0675,
+      "step": 18464
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.56807973048849e-05,
+      "loss": 0.0623,
+      "step": 18466
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.5680329402957144e-05,
+      "loss": 0.0811,
+      "step": 18468
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.567986150102938e-05,
+      "loss": 0.1039,
+      "step": 18470
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.5679393599101636e-05,
+      "loss": 0.0557,
+      "step": 18472
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.5678925697173875e-05,
+      "loss": 0.048,
+      "step": 18474
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.567845779524612e-05,
+      "loss": 0.0504,
+      "step": 18476
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.567798989331836e-05,
+      "loss": 0.0646,
+      "step": 18478
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.567752199139061e-05,
+      "loss": 0.0745,
+      "step": 18480
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.567705408946285e-05,
+      "loss": 0.0639,
+      "step": 18482
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.56765861875351e-05,
+      "loss": 0.0683,
+      "step": 18484
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.567611828560734e-05,
+      "loss": 0.0596,
+      "step": 18486
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.567565038367958e-05,
+      "loss": 0.0624,
+      "step": 18488
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.567518248175183e-05,
+      "loss": 0.0536,
+      "step": 18490
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.567471457982407e-05,
+      "loss": 0.0704,
+      "step": 18492
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5674246677896314e-05,
+      "loss": 0.0923,
+      "step": 18494
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.567377877596856e-05,
+      "loss": 0.0783,
+      "step": 18496
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5673310874040806e-05,
+      "loss": 0.0792,
+      "step": 18498
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5672842972113045e-05,
+      "loss": 0.045,
+      "step": 18500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.567237507018529e-05,
+      "loss": 0.0592,
+      "step": 18502
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.567190716825753e-05,
+      "loss": 0.0582,
+      "step": 18504
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.567143926632978e-05,
+      "loss": 0.0808,
+      "step": 18506
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.567097136440202e-05,
+      "loss": 0.0711,
+      "step": 18508
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.567050346247427e-05,
+      "loss": 0.058,
+      "step": 18510
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5670035560546507e-05,
+      "loss": 0.0584,
+      "step": 18512
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.566956765861876e-05,
+      "loss": 0.0872,
+      "step": 18514
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5669099756691e-05,
+      "loss": 0.0625,
+      "step": 18516
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5668631854763244e-05,
+      "loss": 0.0688,
+      "step": 18518
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5668163952835483e-05,
+      "loss": 0.0666,
+      "step": 18520
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.566769605090773e-05,
+      "loss": 0.0567,
+      "step": 18522
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5667228148979975e-05,
+      "loss": 0.0653,
+      "step": 18524
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.566676024705222e-05,
+      "loss": 0.0905,
+      "step": 18526
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.566629234512446e-05,
+      "loss": 0.096,
+      "step": 18528
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5665824443196706e-05,
+      "loss": 0.0797,
+      "step": 18530
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.566535654126895e-05,
+      "loss": 0.0705,
+      "step": 18532
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.56648886393412e-05,
+      "loss": 0.0588,
+      "step": 18534
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.566442073741344e-05,
+      "loss": 0.0743,
+      "step": 18536
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.566395283548568e-05,
+      "loss": 0.0661,
+      "step": 18538
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.566348493355793e-05,
+      "loss": 0.0802,
+      "step": 18540
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5663017031630175e-05,
+      "loss": 0.0562,
+      "step": 18542
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5662549129702414e-05,
+      "loss": 0.0646,
+      "step": 18544
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.566208122777466e-05,
+      "loss": 0.0774,
+      "step": 18546
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5661613325846906e-05,
+      "loss": 0.0669,
+      "step": 18548
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.566114542391915e-05,
+      "loss": 0.0566,
+      "step": 18550
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.566067752199139e-05,
+      "loss": 0.0428,
+      "step": 18552
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.566020962006364e-05,
+      "loss": 0.0669,
+      "step": 18554
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5659741718135876e-05,
+      "loss": 0.0795,
+      "step": 18556
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.565927381620813e-05,
+      "loss": 0.0556,
+      "step": 18558
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.565880591428037e-05,
+      "loss": 0.0486,
+      "step": 18560
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.5658338012352614e-05,
+      "loss": 0.056,
+      "step": 18562
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.565787011042485e-05,
+      "loss": 0.0688,
+      "step": 18564
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.56574022084971e-05,
+      "loss": 0.0698,
+      "step": 18566
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.5656934306569345e-05,
+      "loss": 0.0619,
+      "step": 18568
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.565646640464159e-05,
+      "loss": 0.0527,
+      "step": 18570
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.565599850271383e-05,
+      "loss": 0.0389,
+      "step": 18572
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.5655530600786076e-05,
+      "loss": 0.0677,
+      "step": 18574
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.565506269885832e-05,
+      "loss": 0.0628,
+      "step": 18576
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.565459479693057e-05,
+      "loss": 0.0577,
+      "step": 18578
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.5654126895002807e-05,
+      "loss": 0.0564,
+      "step": 18580
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.565365899307505e-05,
+      "loss": 0.0572,
+      "step": 18582
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.56531910911473e-05,
+      "loss": 0.0669,
+      "step": 18584
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.5652723189219544e-05,
+      "loss": 0.0569,
+      "step": 18586
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.5652255287291783e-05,
+      "loss": 0.0608,
+      "step": 18588
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.565178738536403e-05,
+      "loss": 0.0616,
+      "step": 18590
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.5651319483436275e-05,
+      "loss": 0.101,
+      "step": 18592
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.565085158150852e-05,
+      "loss": 0.0599,
+      "step": 18594
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.565038367958076e-05,
+      "loss": 0.0807,
+      "step": 18596
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.5649915777653006e-05,
+      "loss": 0.051,
+      "step": 18598
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.5649447875725245e-05,
+      "loss": 0.076,
+      "step": 18600
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.56489799737975e-05,
+      "loss": 0.0665,
+      "step": 18602
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.564851207186974e-05,
+      "loss": 0.0898,
+      "step": 18604
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.564804416994198e-05,
+      "loss": 0.0615,
+      "step": 18606
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.564757626801422e-05,
+      "loss": 0.0567,
+      "step": 18608
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.5647108366086475e-05,
+      "loss": 0.0735,
+      "step": 18610
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.5646640464158714e-05,
+      "loss": 0.0742,
+      "step": 18612
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.564617256223096e-05,
+      "loss": 0.0531,
+      "step": 18614
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.56457046603032e-05,
+      "loss": 0.0808,
+      "step": 18616
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.5645236758375445e-05,
+      "loss": 0.0787,
+      "step": 18618
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.564476885644769e-05,
+      "loss": 0.0624,
+      "step": 18620
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.564430095451994e-05,
+      "loss": 0.0635,
+      "step": 18622
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.5643833052592176e-05,
+      "loss": 0.0672,
+      "step": 18624
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.564336515066442e-05,
+      "loss": 0.0847,
+      "step": 18626
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.564289724873667e-05,
+      "loss": 0.0717,
+      "step": 18628
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.5642429346808914e-05,
+      "loss": 0.0541,
+      "step": 18630
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.564196144488115e-05,
+      "loss": 0.0768,
+      "step": 18632
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.56414935429534e-05,
+      "loss": 0.0713,
+      "step": 18634
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.5641025641025645e-05,
+      "loss": 0.066,
+      "step": 18636
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.564055773909789e-05,
+      "loss": 0.0893,
+      "step": 18638
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.564008983717013e-05,
+      "loss": 0.0778,
+      "step": 18640
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.5639621935242376e-05,
+      "loss": 0.0615,
+      "step": 18642
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.563915403331462e-05,
+      "loss": 0.0626,
+      "step": 18644
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.563868613138687e-05,
+      "loss": 0.0516,
+      "step": 18646
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.5638218229459107e-05,
+      "loss": 0.0506,
+      "step": 18648
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.563775032753135e-05,
+      "loss": 0.0568,
+      "step": 18650
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.563728242560359e-05,
+      "loss": 0.057,
+      "step": 18652
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.5636814523675844e-05,
+      "loss": 0.062,
+      "step": 18654
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.5636346621748083e-05,
+      "loss": 0.0646,
+      "step": 18656
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.563587871982033e-05,
+      "loss": 0.0736,
+      "step": 18658
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.563541081789257e-05,
+      "loss": 0.0544,
+      "step": 18660
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.563494291596482e-05,
+      "loss": 0.072,
+      "step": 18662
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.563447501403706e-05,
+      "loss": 0.038,
+      "step": 18664
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.5634007112109306e-05,
+      "loss": 0.0732,
+      "step": 18666
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.5633539210181545e-05,
+      "loss": 0.0632,
+      "step": 18668
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.563307130825379e-05,
+      "loss": 0.0554,
+      "step": 18670
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.563260340632604e-05,
+      "loss": 0.0718,
+      "step": 18672
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.563213550439828e-05,
+      "loss": 0.0615,
+      "step": 18674
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.563166760247052e-05,
+      "loss": 0.0532,
+      "step": 18676
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.563119970054277e-05,
+      "loss": 0.0494,
+      "step": 18678
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.5630731798615014e-05,
+      "loss": 0.0624,
+      "step": 18680
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.563026389668726e-05,
+      "loss": 0.0544,
+      "step": 18682
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.56297959947595e-05,
+      "loss": 0.0613,
+      "step": 18684
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.5629328092831745e-05,
+      "loss": 0.0713,
+      "step": 18686
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.562886019090399e-05,
+      "loss": 0.0632,
+      "step": 18688
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.562839228897624e-05,
+      "loss": 0.0607,
+      "step": 18690
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.5627924387048476e-05,
+      "loss": 0.068,
+      "step": 18692
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.562745648512072e-05,
+      "loss": 0.0653,
+      "step": 18694
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.562698858319297e-05,
+      "loss": 0.0796,
+      "step": 18696
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.5626520681265214e-05,
+      "loss": 0.0606,
+      "step": 18698
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.562605277933745e-05,
+      "loss": 0.0678,
+      "step": 18700
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.56255848774097e-05,
+      "loss": 0.0536,
+      "step": 18702
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.562511697548194e-05,
+      "loss": 0.0632,
+      "step": 18704
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.562464907355419e-05,
+      "loss": 0.0571,
+      "step": 18706
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.562418117162643e-05,
+      "loss": 0.0728,
+      "step": 18708
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.5623713269698676e-05,
+      "loss": 0.0751,
+      "step": 18710
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.5623245367770915e-05,
+      "loss": 0.073,
+      "step": 18712
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.562277746584316e-05,
+      "loss": 0.0533,
+      "step": 18714
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.5622309563915407e-05,
+      "loss": 0.0686,
+      "step": 18716
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.562184166198765e-05,
+      "loss": 0.0733,
+      "step": 18718
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.562137376005989e-05,
+      "loss": 0.0781,
+      "step": 18720
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.562090585813214e-05,
+      "loss": 0.0726,
+      "step": 18722
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.5620437956204383e-05,
+      "loss": 0.072,
+      "step": 18724
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.561997005427663e-05,
+      "loss": 0.056,
+      "step": 18726
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.561950215234887e-05,
+      "loss": 0.0599,
+      "step": 18728
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.5619034250421114e-05,
+      "loss": 0.0684,
+      "step": 18730
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.561856634849336e-05,
+      "loss": 0.0667,
+      "step": 18732
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.5618098446565606e-05,
+      "loss": 0.083,
+      "step": 18734
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.5617630544637845e-05,
+      "loss": 0.0546,
+      "step": 18736
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.561716264271009e-05,
+      "loss": 0.0572,
+      "step": 18738
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.561669474078234e-05,
+      "loss": 0.0632,
+      "step": 18740
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.561622683885458e-05,
+      "loss": 0.0497,
+      "step": 18742
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.561575893692682e-05,
+      "loss": 0.0471,
+      "step": 18744
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.561529103499906e-05,
+      "loss": 0.062,
+      "step": 18746
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.561482313307131e-05,
+      "loss": 0.0707,
+      "step": 18748
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.561435523114355e-05,
+      "loss": 0.0727,
+      "step": 18750
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.56138873292158e-05,
+      "loss": 0.0546,
+      "step": 18752
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.561341942728804e-05,
+      "loss": 0.0725,
+      "step": 18754
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.5612951525360284e-05,
+      "loss": 0.0531,
+      "step": 18756
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.561248362343253e-05,
+      "loss": 0.0847,
+      "step": 18758
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.5612015721504776e-05,
+      "loss": 0.0603,
+      "step": 18760
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.5611547819577015e-05,
+      "loss": 0.0646,
+      "step": 18762
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.561107991764926e-05,
+      "loss": 0.0671,
+      "step": 18764
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.561061201572151e-05,
+      "loss": 0.0626,
+      "step": 18766
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.561014411379375e-05,
+      "loss": 0.0754,
+      "step": 18768
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.560967621186599e-05,
+      "loss": 0.0583,
+      "step": 18770
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.560920830993824e-05,
+      "loss": 0.0808,
+      "step": 18772
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.5608740408010484e-05,
+      "loss": 0.0559,
+      "step": 18774
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.560827250608273e-05,
+      "loss": 0.0659,
+      "step": 18776
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.560780460415497e-05,
+      "loss": 0.0571,
+      "step": 18778
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.5607336702227215e-05,
+      "loss": 0.0578,
+      "step": 18780
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.5606868800299454e-05,
+      "loss": 0.0604,
+      "step": 18782
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.560640089837171e-05,
+      "loss": 0.0682,
+      "step": 18784
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.5605932996443946e-05,
+      "loss": 0.0456,
+      "step": 18786
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.560546509451619e-05,
+      "loss": 0.0641,
+      "step": 18788
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.560499719258843e-05,
+      "loss": 0.0702,
+      "step": 18790
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.5604529290660683e-05,
+      "loss": 0.0541,
+      "step": 18792
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.560406138873292e-05,
+      "loss": 0.0545,
+      "step": 18794
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.560359348680517e-05,
+      "loss": 0.0776,
+      "step": 18796
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.560312558487741e-05,
+      "loss": 0.0748,
+      "step": 18798
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.5602657682949654e-05,
+      "loss": 0.062,
+      "step": 18800
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.56021897810219e-05,
+      "loss": 0.0597,
+      "step": 18802
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.5601721879094145e-05,
+      "loss": 0.0576,
+      "step": 18804
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.5601253977166385e-05,
+      "loss": 0.0621,
+      "step": 18806
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.560078607523863e-05,
+      "loss": 0.065,
+      "step": 18808
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.5600318173310876e-05,
+      "loss": 0.0754,
+      "step": 18810
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.559985027138312e-05,
+      "loss": 0.068,
+      "step": 18812
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.559938236945536e-05,
+      "loss": 0.0525,
+      "step": 18814
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.559891446752761e-05,
+      "loss": 0.0633,
+      "step": 18816
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.559844656559985e-05,
+      "loss": 0.0751,
+      "step": 18818
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.55979786636721e-05,
+      "loss": 0.0747,
+      "step": 18820
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.559751076174434e-05,
+      "loss": 0.0561,
+      "step": 18822
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.5597042859816584e-05,
+      "loss": 0.0641,
+      "step": 18824
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.559657495788883e-05,
+      "loss": 0.0682,
+      "step": 18826
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.5596107055961076e-05,
+      "loss": 0.0755,
+      "step": 18828
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.5595639154033315e-05,
+      "loss": 0.0675,
+      "step": 18830
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.559517125210556e-05,
+      "loss": 0.0602,
+      "step": 18832
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.55947033501778e-05,
+      "loss": 0.0609,
+      "step": 18834
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.559423544825005e-05,
+      "loss": 0.0635,
+      "step": 18836
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.559376754632229e-05,
+      "loss": 0.0697,
+      "step": 18838
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.559329964439454e-05,
+      "loss": 0.0515,
+      "step": 18840
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.559283174246678e-05,
+      "loss": 0.0562,
+      "step": 18842
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.559236384053902e-05,
+      "loss": 0.0694,
+      "step": 18844
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.559189593861127e-05,
+      "loss": 0.0796,
+      "step": 18846
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.5591428036683515e-05,
+      "loss": 0.0685,
+      "step": 18848
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.5590960134755754e-05,
+      "loss": 0.0648,
+      "step": 18850
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.5590492232828e-05,
+      "loss": 0.0689,
+      "step": 18852
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.5590024330900246e-05,
+      "loss": 0.0586,
+      "step": 18854
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.558955642897249e-05,
+      "loss": 0.0408,
+      "step": 18856
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.558908852704473e-05,
+      "loss": 0.0653,
+      "step": 18858
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.558862062511698e-05,
+      "loss": 0.0431,
+      "step": 18860
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.558815272318922e-05,
+      "loss": 0.0681,
+      "step": 18862
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.558768482126147e-05,
+      "loss": 0.0815,
+      "step": 18864
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.558721691933371e-05,
+      "loss": 0.0746,
+      "step": 18866
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.5586749017405954e-05,
+      "loss": 0.0648,
+      "step": 18868
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.55862811154782e-05,
+      "loss": 0.0511,
+      "step": 18870
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.5585813213550445e-05,
+      "loss": 0.0745,
+      "step": 18872
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.5585345311622685e-05,
+      "loss": 0.0765,
+      "step": 18874
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.558487740969493e-05,
+      "loss": 0.0779,
+      "step": 18876
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.558440950776717e-05,
+      "loss": 0.0693,
+      "step": 18878
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.558394160583942e-05,
+      "loss": 0.0779,
+      "step": 18880
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.558347370391166e-05,
+      "loss": 0.0706,
+      "step": 18882
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.558300580198391e-05,
+      "loss": 0.0539,
+      "step": 18884
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.5582537900056146e-05,
+      "loss": 0.0695,
+      "step": 18886
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.55820699981284e-05,
+      "loss": 0.0581,
+      "step": 18888
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.558160209620064e-05,
+      "loss": 0.0647,
+      "step": 18890
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.5581134194272884e-05,
+      "loss": 0.0665,
+      "step": 18892
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.558066629234512e-05,
+      "loss": 0.063,
+      "step": 18894
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.558019839041737e-05,
+      "loss": 0.06,
+      "step": 18896
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.5579730488489615e-05,
+      "loss": 0.0868,
+      "step": 18898
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.557926258656186e-05,
+      "loss": 0.059,
+      "step": 18900
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.55787946846341e-05,
+      "loss": 0.0526,
+      "step": 18902
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.5578326782706346e-05,
+      "loss": 0.0748,
+      "step": 18904
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.557785888077859e-05,
+      "loss": 0.069,
+      "step": 18906
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.557739097885084e-05,
+      "loss": 0.0612,
+      "step": 18908
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.557692307692308e-05,
+      "loss": 0.049,
+      "step": 18910
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.557645517499532e-05,
+      "loss": 0.0793,
+      "step": 18912
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.557598727306757e-05,
+      "loss": 0.0985,
+      "step": 18914
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.5575519371139815e-05,
+      "loss": 0.0623,
+      "step": 18916
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.5575051469212054e-05,
+      "loss": 0.0652,
+      "step": 18918
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.55745835672843e-05,
+      "loss": 0.0655,
+      "step": 18920
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.5574115665356546e-05,
+      "loss": 0.0721,
+      "step": 18922
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.557364776342879e-05,
+      "loss": 0.0596,
+      "step": 18924
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.557317986150103e-05,
+      "loss": 0.0819,
+      "step": 18926
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.557271195957328e-05,
+      "loss": 0.0507,
+      "step": 18928
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.5572244057645516e-05,
+      "loss": 0.0519,
+      "step": 18930
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.557177615571777e-05,
+      "loss": 0.0694,
+      "step": 18932
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.557130825379001e-05,
+      "loss": 0.0723,
+      "step": 18934
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.5570840351862254e-05,
+      "loss": 0.0775,
+      "step": 18936
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.557037244993449e-05,
+      "loss": 0.0823,
+      "step": 18938
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.5569904548006745e-05,
+      "loss": 0.063,
+      "step": 18940
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.5569436646078985e-05,
+      "loss": 0.0498,
+      "step": 18942
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.556896874415123e-05,
+      "loss": 0.0713,
+      "step": 18944
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.556850084222347e-05,
+      "loss": 0.0698,
+      "step": 18946
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.5568032940295716e-05,
+      "loss": 0.065,
+      "step": 18948
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.556756503836796e-05,
+      "loss": 0.0562,
+      "step": 18950
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.556709713644021e-05,
+      "loss": 0.0549,
+      "step": 18952
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.5566629234512447e-05,
+      "loss": 0.058,
+      "step": 18954
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.556616133258469e-05,
+      "loss": 0.0707,
+      "step": 18956
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.556569343065694e-05,
+      "loss": 0.0784,
+      "step": 18958
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.5565225528729184e-05,
+      "loss": 0.057,
+      "step": 18960
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.556475762680142e-05,
+      "loss": 0.0569,
+      "step": 18962
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.556428972487367e-05,
+      "loss": 0.0696,
+      "step": 18964
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.5563821822945915e-05,
+      "loss": 0.0619,
+      "step": 18966
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.556335392101816e-05,
+      "loss": 0.0745,
+      "step": 18968
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.55628860190904e-05,
+      "loss": 0.0691,
+      "step": 18970
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.5562418117162646e-05,
+      "loss": 0.084,
+      "step": 18972
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.556195021523489e-05,
+      "loss": 0.0603,
+      "step": 18974
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.556148231330714e-05,
+      "loss": 0.0585,
+      "step": 18976
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.556101441137938e-05,
+      "loss": 0.0505,
+      "step": 18978
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.556054650945162e-05,
+      "loss": 0.0765,
+      "step": 18980
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.556007860752386e-05,
+      "loss": 0.0823,
+      "step": 18982
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.5559610705596115e-05,
+      "loss": 0.0715,
+      "step": 18984
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.5559142803668354e-05,
+      "loss": 0.0713,
+      "step": 18986
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.55586749017406e-05,
+      "loss": 0.078,
+      "step": 18988
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.555820699981284e-05,
+      "loss": 0.0643,
+      "step": 18990
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.5557739097885085e-05,
+      "loss": 0.057,
+      "step": 18992
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.555727119595733e-05,
+      "loss": 0.067,
+      "step": 18994
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.555680329402957e-05,
+      "loss": 0.0474,
+      "step": 18996
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.5556335392101816e-05,
+      "loss": 0.0659,
+      "step": 18998
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.555586749017406e-05,
+      "loss": 0.0504,
+      "step": 19000
+    },
+    {
+      "epoch": 2.67,
+      "eval_gen_len": 30.431,
+      "eval_loss": 1.0705924034118652,
+      "eval_meteor": 0.0441,
+      "eval_runtime": 16.562,
+      "eval_samples_per_second": 3.502,
+      "eval_steps_per_second": 0.483,
+      "step": 19000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.555539958824631e-05,
+      "loss": 0.0504,
+      "step": 19002
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.555493168631855e-05,
+      "loss": 0.0731,
+      "step": 19004
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.555446378439079e-05,
+      "loss": 0.0599,
+      "step": 19006
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.555399588246303e-05,
+      "loss": 0.0748,
+      "step": 19008
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.5553527980535285e-05,
+      "loss": 0.06,
+      "step": 19010
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.5553060078607524e-05,
+      "loss": 0.0765,
+      "step": 19012
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.555259217667977e-05,
+      "loss": 0.069,
+      "step": 19014
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.555212427475201e-05,
+      "loss": 0.0649,
+      "step": 19016
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.555165637282426e-05,
+      "loss": 0.0666,
+      "step": 19018
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.55511884708965e-05,
+      "loss": 0.0585,
+      "step": 19020
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.5550720568968747e-05,
+      "loss": 0.0764,
+      "step": 19022
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.5550252667040986e-05,
+      "loss": 0.0548,
+      "step": 19024
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.554978476511323e-05,
+      "loss": 0.0594,
+      "step": 19026
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.554931686318548e-05,
+      "loss": 0.0696,
+      "step": 19028
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.5548848961257723e-05,
+      "loss": 0.0481,
+      "step": 19030
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.554838105932996e-05,
+      "loss": 0.0692,
+      "step": 19032
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.554791315740221e-05,
+      "loss": 0.0682,
+      "step": 19034
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.5547445255474454e-05,
+      "loss": 0.0632,
+      "step": 19036
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.55469773535467e-05,
+      "loss": 0.0829,
+      "step": 19038
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.554650945161894e-05,
+      "loss": 0.0775,
+      "step": 19040
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.5546041549691185e-05,
+      "loss": 0.0518,
+      "step": 19042
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.554557364776343e-05,
+      "loss": 0.0752,
+      "step": 19044
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.554510574583568e-05,
+      "loss": 0.0581,
+      "step": 19046
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.5544637843907916e-05,
+      "loss": 0.0826,
+      "step": 19048
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.554416994198016e-05,
+      "loss": 0.0852,
+      "step": 19050
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.554370204005241e-05,
+      "loss": 0.0631,
+      "step": 19052
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.5543234138124654e-05,
+      "loss": 0.0701,
+      "step": 19054
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.554276623619689e-05,
+      "loss": 0.0688,
+      "step": 19056
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.554229833426914e-05,
+      "loss": 0.0678,
+      "step": 19058
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.554183043234138e-05,
+      "loss": 0.0652,
+      "step": 19060
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.554136253041363e-05,
+      "loss": 0.0785,
+      "step": 19062
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.554089462848587e-05,
+      "loss": 0.0489,
+      "step": 19064
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.5540426726558116e-05,
+      "loss": 0.0755,
+      "step": 19066
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.5539958824630355e-05,
+      "loss": 0.0735,
+      "step": 19068
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.553949092270261e-05,
+      "loss": 0.044,
+      "step": 19070
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.553902302077485e-05,
+      "loss": 0.0729,
+      "step": 19072
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.553855511884709e-05,
+      "loss": 0.0615,
+      "step": 19074
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.553808721691933e-05,
+      "loss": 0.0526,
+      "step": 19076
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.553761931499158e-05,
+      "loss": 0.0603,
+      "step": 19078
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.5537151413063824e-05,
+      "loss": 0.087,
+      "step": 19080
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.553668351113607e-05,
+      "loss": 0.0739,
+      "step": 19082
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.553621560920831e-05,
+      "loss": 0.0643,
+      "step": 19084
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.5535747707280555e-05,
+      "loss": 0.0575,
+      "step": 19086
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.55352798053528e-05,
+      "loss": 0.0624,
+      "step": 19088
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.5534811903425047e-05,
+      "loss": 0.0632,
+      "step": 19090
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.5534344001497286e-05,
+      "loss": 0.0607,
+      "step": 19092
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.553387609956953e-05,
+      "loss": 0.0689,
+      "step": 19094
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.553340819764178e-05,
+      "loss": 0.0604,
+      "step": 19096
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.5532940295714023e-05,
+      "loss": 0.0714,
+      "step": 19098
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.553247239378626e-05,
+      "loss": 0.0531,
+      "step": 19100
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.553200449185851e-05,
+      "loss": 0.0674,
+      "step": 19102
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.5531536589930754e-05,
+      "loss": 0.052,
+      "step": 19104
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.5531068688003e-05,
+      "loss": 0.0647,
+      "step": 19106
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.553060078607524e-05,
+      "loss": 0.0566,
+      "step": 19108
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.5530132884147485e-05,
+      "loss": 0.0649,
+      "step": 19110
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.5529664982219724e-05,
+      "loss": 0.0778,
+      "step": 19112
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.552919708029198e-05,
+      "loss": 0.0608,
+      "step": 19114
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.5528729178364216e-05,
+      "loss": 0.0413,
+      "step": 19116
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.552826127643646e-05,
+      "loss": 0.0594,
+      "step": 19118
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.55277933745087e-05,
+      "loss": 0.0597,
+      "step": 19120
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.552732547258095e-05,
+      "loss": 0.0633,
+      "step": 19122
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.552685757065319e-05,
+      "loss": 0.05,
+      "step": 19124
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.552638966872544e-05,
+      "loss": 0.073,
+      "step": 19126
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.552592176679768e-05,
+      "loss": 0.0781,
+      "step": 19128
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.5525453864869924e-05,
+      "loss": 0.0681,
+      "step": 19130
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.552498596294217e-05,
+      "loss": 0.0609,
+      "step": 19132
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.5524518061014416e-05,
+      "loss": 0.0785,
+      "step": 19134
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.5524050159086655e-05,
+      "loss": 0.061,
+      "step": 19136
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.55235822571589e-05,
+      "loss": 0.0505,
+      "step": 19138
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.552311435523115e-05,
+      "loss": 0.0731,
+      "step": 19140
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.552264645330339e-05,
+      "loss": 0.0712,
+      "step": 19142
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.552217855137563e-05,
+      "loss": 0.0726,
+      "step": 19144
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.552171064944788e-05,
+      "loss": 0.0575,
+      "step": 19146
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.5521242747520124e-05,
+      "loss": 0.0604,
+      "step": 19148
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.552077484559237e-05,
+      "loss": 0.0475,
+      "step": 19150
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.552030694366461e-05,
+      "loss": 0.0734,
+      "step": 19152
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.5519839041736855e-05,
+      "loss": 0.0659,
+      "step": 19154
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.5519371139809094e-05,
+      "loss": 0.0629,
+      "step": 19156
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.5518903237881347e-05,
+      "loss": 0.0676,
+      "step": 19158
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.5518435335953586e-05,
+      "loss": 0.0714,
+      "step": 19160
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.551796743402583e-05,
+      "loss": 0.0646,
+      "step": 19162
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.551749953209807e-05,
+      "loss": 0.0534,
+      "step": 19164
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.5517031630170323e-05,
+      "loss": 0.0719,
+      "step": 19166
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.551656372824256e-05,
+      "loss": 0.0642,
+      "step": 19168
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.551609582631481e-05,
+      "loss": 0.066,
+      "step": 19170
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.551562792438705e-05,
+      "loss": 0.0562,
+      "step": 19172
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.5515160022459294e-05,
+      "loss": 0.0592,
+      "step": 19174
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.551469212053154e-05,
+      "loss": 0.0606,
+      "step": 19176
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.5514224218603785e-05,
+      "loss": 0.0603,
+      "step": 19178
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.5513756316676024e-05,
+      "loss": 0.068,
+      "step": 19180
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.551328841474827e-05,
+      "loss": 0.0748,
+      "step": 19182
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.5512820512820516e-05,
+      "loss": 0.0691,
+      "step": 19184
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.551235261089276e-05,
+      "loss": 0.0742,
+      "step": 19186
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.5511884708965e-05,
+      "loss": 0.0575,
+      "step": 19188
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.551141680703725e-05,
+      "loss": 0.0545,
+      "step": 19190
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.551094890510949e-05,
+      "loss": 0.0569,
+      "step": 19192
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.551048100318174e-05,
+      "loss": 0.0844,
+      "step": 19194
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.551001310125398e-05,
+      "loss": 0.0789,
+      "step": 19196
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.5509545199326224e-05,
+      "loss": 0.0648,
+      "step": 19198
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.550907729739847e-05,
+      "loss": 0.0764,
+      "step": 19200
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5508609395470716e-05,
+      "loss": 0.0917,
+      "step": 19202
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5508141493542955e-05,
+      "loss": 0.0543,
+      "step": 19204
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.55076735916152e-05,
+      "loss": 0.0609,
+      "step": 19206
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.550720568968744e-05,
+      "loss": 0.0483,
+      "step": 19208
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.550673778775969e-05,
+      "loss": 0.0676,
+      "step": 19210
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.550626988583193e-05,
+      "loss": 0.0832,
+      "step": 19212
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.550580198390418e-05,
+      "loss": 0.0605,
+      "step": 19214
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.550533408197642e-05,
+      "loss": 0.068,
+      "step": 19216
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.550486618004867e-05,
+      "loss": 0.066,
+      "step": 19218
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.550439827812091e-05,
+      "loss": 0.0515,
+      "step": 19220
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5503930376193155e-05,
+      "loss": 0.086,
+      "step": 19222
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5503462474265394e-05,
+      "loss": 0.0766,
+      "step": 19224
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.550299457233764e-05,
+      "loss": 0.0666,
+      "step": 19226
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5502526670409886e-05,
+      "loss": 0.0722,
+      "step": 19228
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.550205876848213e-05,
+      "loss": 0.0682,
+      "step": 19230
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.550159086655437e-05,
+      "loss": 0.0736,
+      "step": 19232
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.550112296462662e-05,
+      "loss": 0.0439,
+      "step": 19234
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.550065506269886e-05,
+      "loss": 0.0654,
+      "step": 19236
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.550018716077111e-05,
+      "loss": 0.0754,
+      "step": 19238
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.549971925884335e-05,
+      "loss": 0.0553,
+      "step": 19240
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5499251356915594e-05,
+      "loss": 0.0719,
+      "step": 19242
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.549878345498784e-05,
+      "loss": 0.0593,
+      "step": 19244
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5498315553060085e-05,
+      "loss": 0.0577,
+      "step": 19246
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5497847651132324e-05,
+      "loss": 0.0555,
+      "step": 19248
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5497379749204564e-05,
+      "loss": 0.0575,
+      "step": 19250
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5496911847276816e-05,
+      "loss": 0.0824,
+      "step": 19252
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5496443945349055e-05,
+      "loss": 0.0802,
+      "step": 19254
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.54959760434213e-05,
+      "loss": 0.0755,
+      "step": 19256
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.549550814149354e-05,
+      "loss": 0.0592,
+      "step": 19258
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5495040239565786e-05,
+      "loss": 0.0548,
+      "step": 19260
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.549457233763803e-05,
+      "loss": 0.0653,
+      "step": 19262
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.549410443571028e-05,
+      "loss": 0.055,
+      "step": 19264
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.549363653378252e-05,
+      "loss": 0.0724,
+      "step": 19266
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.549316863185476e-05,
+      "loss": 0.0656,
+      "step": 19268
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.549270072992701e-05,
+      "loss": 0.0491,
+      "step": 19270
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5492232827999255e-05,
+      "loss": 0.0565,
+      "step": 19272
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5491764926071494e-05,
+      "loss": 0.0735,
+      "step": 19274
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.549129702414374e-05,
+      "loss": 0.0564,
+      "step": 19276
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5490829122215986e-05,
+      "loss": 0.074,
+      "step": 19278
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.549036122028823e-05,
+      "loss": 0.068,
+      "step": 19280
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.548989331836047e-05,
+      "loss": 0.067,
+      "step": 19282
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.548942541643272e-05,
+      "loss": 0.0689,
+      "step": 19284
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.548895751450496e-05,
+      "loss": 0.0556,
+      "step": 19286
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.548848961257721e-05,
+      "loss": 0.0582,
+      "step": 19288
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.548802171064945e-05,
+      "loss": 0.0523,
+      "step": 19290
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5487553808721694e-05,
+      "loss": 0.0598,
+      "step": 19292
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.548708590679393e-05,
+      "loss": 0.0523,
+      "step": 19294
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5486618004866186e-05,
+      "loss": 0.0419,
+      "step": 19296
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5486150102938425e-05,
+      "loss": 0.0432,
+      "step": 19298
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.548568220101067e-05,
+      "loss": 0.059,
+      "step": 19300
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.548521429908291e-05,
+      "loss": 0.077,
+      "step": 19302
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5484746397155156e-05,
+      "loss": 0.0592,
+      "step": 19304
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.54842784952274e-05,
+      "loss": 0.0649,
+      "step": 19306
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.548381059329965e-05,
+      "loss": 0.0603,
+      "step": 19308
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.548334269137189e-05,
+      "loss": 0.074,
+      "step": 19310
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.548287478944413e-05,
+      "loss": 0.0663,
+      "step": 19312
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.548240688751638e-05,
+      "loss": 0.0613,
+      "step": 19314
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5481938985588625e-05,
+      "loss": 0.0723,
+      "step": 19316
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5481471083660864e-05,
+      "loss": 0.0665,
+      "step": 19318
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.548100318173311e-05,
+      "loss": 0.08,
+      "step": 19320
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5480535279805355e-05,
+      "loss": 0.0713,
+      "step": 19322
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.54800673778776e-05,
+      "loss": 0.0742,
+      "step": 19324
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.547959947594984e-05,
+      "loss": 0.0652,
+      "step": 19326
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5479131574022086e-05,
+      "loss": 0.0818,
+      "step": 19328
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.547866367209433e-05,
+      "loss": 0.0762,
+      "step": 19330
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.547819577016658e-05,
+      "loss": 0.0653,
+      "step": 19332
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.547772786823882e-05,
+      "loss": 0.0663,
+      "step": 19334
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.547725996631106e-05,
+      "loss": 0.0596,
+      "step": 19336
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.54767920643833e-05,
+      "loss": 0.0731,
+      "step": 19338
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5476324162455555e-05,
+      "loss": 0.0628,
+      "step": 19340
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5475856260527794e-05,
+      "loss": 0.0415,
+      "step": 19342
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.547538835860004e-05,
+      "loss": 0.0531,
+      "step": 19344
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.547492045667228e-05,
+      "loss": 0.088,
+      "step": 19346
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.547445255474453e-05,
+      "loss": 0.084,
+      "step": 19348
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.547398465281677e-05,
+      "loss": 0.0515,
+      "step": 19350
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.547351675088902e-05,
+      "loss": 0.0747,
+      "step": 19352
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5473048848961256e-05,
+      "loss": 0.0598,
+      "step": 19354
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.54725809470335e-05,
+      "loss": 0.0594,
+      "step": 19356
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.547211304510575e-05,
+      "loss": 0.0726,
+      "step": 19358
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5471645143177994e-05,
+      "loss": 0.0763,
+      "step": 19360
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.547117724125023e-05,
+      "loss": 0.0562,
+      "step": 19362
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.547070933932248e-05,
+      "loss": 0.0957,
+      "step": 19364
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5470241437394725e-05,
+      "loss": 0.0545,
+      "step": 19366
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.546977353546697e-05,
+      "loss": 0.0679,
+      "step": 19368
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.546930563353921e-05,
+      "loss": 0.0555,
+      "step": 19370
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5468837731611456e-05,
+      "loss": 0.0656,
+      "step": 19372
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.54683698296837e-05,
+      "loss": 0.0852,
+      "step": 19374
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.546790192775595e-05,
+      "loss": 0.0689,
+      "step": 19376
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.546743402582819e-05,
+      "loss": 0.068,
+      "step": 19378
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.546696612390043e-05,
+      "loss": 0.0604,
+      "step": 19380
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.546649822197268e-05,
+      "loss": 0.0525,
+      "step": 19382
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5466030320044925e-05,
+      "loss": 0.0712,
+      "step": 19384
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5465562418117164e-05,
+      "loss": 0.0672,
+      "step": 19386
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.546509451618941e-05,
+      "loss": 0.0593,
+      "step": 19388
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.546462661426165e-05,
+      "loss": 0.0772,
+      "step": 19390
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.54641587123339e-05,
+      "loss": 0.0518,
+      "step": 19392
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.546369081040614e-05,
+      "loss": 0.0641,
+      "step": 19394
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5463222908478386e-05,
+      "loss": 0.0677,
+      "step": 19396
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5462755006550626e-05,
+      "loss": 0.0702,
+      "step": 19398
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.546228710462287e-05,
+      "loss": 0.067,
+      "step": 19400
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.546181920269512e-05,
+      "loss": 0.0771,
+      "step": 19402
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.546135130076736e-05,
+      "loss": 0.0658,
+      "step": 19404
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.54608833988396e-05,
+      "loss": 0.0665,
+      "step": 19406
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.546041549691185e-05,
+      "loss": 0.0538,
+      "step": 19408
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5459947594984094e-05,
+      "loss": 0.0468,
+      "step": 19410
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.545947969305634e-05,
+      "loss": 0.0793,
+      "step": 19412
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.545901179112858e-05,
+      "loss": 0.0563,
+      "step": 19414
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5458543889200825e-05,
+      "loss": 0.0623,
+      "step": 19416
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.545807598727307e-05,
+      "loss": 0.0568,
+      "step": 19418
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.545760808534532e-05,
+      "loss": 0.0633,
+      "step": 19420
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5457140183417556e-05,
+      "loss": 0.0618,
+      "step": 19422
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.54566722814898e-05,
+      "loss": 0.0648,
+      "step": 19424
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.545620437956205e-05,
+      "loss": 0.0549,
+      "step": 19426
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5455736477634294e-05,
+      "loss": 0.0495,
+      "step": 19428
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.545526857570653e-05,
+      "loss": 0.0602,
+      "step": 19430
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.545480067377878e-05,
+      "loss": 0.0654,
+      "step": 19432
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.545433277185102e-05,
+      "loss": 0.0697,
+      "step": 19434
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.545386486992327e-05,
+      "loss": 0.0556,
+      "step": 19436
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.545339696799551e-05,
+      "loss": 0.0901,
+      "step": 19438
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5452929066067756e-05,
+      "loss": 0.0668,
+      "step": 19440
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5452461164139995e-05,
+      "loss": 0.053,
+      "step": 19442
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.545199326221225e-05,
+      "loss": 0.0725,
+      "step": 19444
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.545152536028449e-05,
+      "loss": 0.0628,
+      "step": 19446
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.545105745835673e-05,
+      "loss": 0.0901,
+      "step": 19448
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.545058955642897e-05,
+      "loss": 0.0553,
+      "step": 19450
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.545012165450122e-05,
+      "loss": 0.0812,
+      "step": 19452
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5449653752573464e-05,
+      "loss": 0.0603,
+      "step": 19454
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.544918585064571e-05,
+      "loss": 0.0691,
+      "step": 19456
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.544871794871795e-05,
+      "loss": 0.0566,
+      "step": 19458
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5448250046790195e-05,
+      "loss": 0.0724,
+      "step": 19460
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.544778214486244e-05,
+      "loss": 0.0753,
+      "step": 19462
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5447314242934686e-05,
+      "loss": 0.0713,
+      "step": 19464
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5446846341006926e-05,
+      "loss": 0.0683,
+      "step": 19466
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.544637843907917e-05,
+      "loss": 0.0846,
+      "step": 19468
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.544591053715142e-05,
+      "loss": 0.0595,
+      "step": 19470
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.544544263522366e-05,
+      "loss": 0.0471,
+      "step": 19472
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.54449747332959e-05,
+      "loss": 0.0723,
+      "step": 19474
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.544450683136815e-05,
+      "loss": 0.0627,
+      "step": 19476
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5444038929440394e-05,
+      "loss": 0.0688,
+      "step": 19478
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.544357102751264e-05,
+      "loss": 0.069,
+      "step": 19480
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.544310312558488e-05,
+      "loss": 0.0623,
+      "step": 19482
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5442635223657125e-05,
+      "loss": 0.0502,
+      "step": 19484
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5442167321729364e-05,
+      "loss": 0.0588,
+      "step": 19486
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.544169941980162e-05,
+      "loss": 0.0726,
+      "step": 19488
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5441231517873856e-05,
+      "loss": 0.0762,
+      "step": 19490
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.54407636159461e-05,
+      "loss": 0.0669,
+      "step": 19492
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.544029571401834e-05,
+      "loss": 0.0543,
+      "step": 19494
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5439827812090594e-05,
+      "loss": 0.0528,
+      "step": 19496
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.543935991016283e-05,
+      "loss": 0.0633,
+      "step": 19498
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.543889200823508e-05,
+      "loss": 0.0552,
+      "step": 19500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.543842410630732e-05,
+      "loss": 0.0763,
+      "step": 19502
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5437956204379564e-05,
+      "loss": 0.0545,
+      "step": 19504
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.543748830245181e-05,
+      "loss": 0.0695,
+      "step": 19506
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.543702040052405e-05,
+      "loss": 0.069,
+      "step": 19508
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5436552498596295e-05,
+      "loss": 0.0894,
+      "step": 19510
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.543608459666854e-05,
+      "loss": 0.0437,
+      "step": 19512
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.543561669474079e-05,
+      "loss": 0.0792,
+      "step": 19514
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5435148792813026e-05,
+      "loss": 0.0734,
+      "step": 19516
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.543468089088527e-05,
+      "loss": 0.0431,
+      "step": 19518
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.543421298895751e-05,
+      "loss": 0.0727,
+      "step": 19520
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5433745087029764e-05,
+      "loss": 0.0816,
+      "step": 19522
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5433277185102e-05,
+      "loss": 0.0608,
+      "step": 19524
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.543280928317425e-05,
+      "loss": 0.0633,
+      "step": 19526
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.543234138124649e-05,
+      "loss": 0.0703,
+      "step": 19528
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.543187347931874e-05,
+      "loss": 0.0581,
+      "step": 19530
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.543140557739098e-05,
+      "loss": 0.0641,
+      "step": 19532
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5430937675463226e-05,
+      "loss": 0.0694,
+      "step": 19534
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5430469773535465e-05,
+      "loss": 0.0527,
+      "step": 19536
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.543000187160771e-05,
+      "loss": 0.0524,
+      "step": 19538
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5429533969679957e-05,
+      "loss": 0.0767,
+      "step": 19540
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.54290660677522e-05,
+      "loss": 0.0624,
+      "step": 19542
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.542859816582444e-05,
+      "loss": 0.0769,
+      "step": 19544
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.542813026389669e-05,
+      "loss": 0.0615,
+      "step": 19546
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5427662361968933e-05,
+      "loss": 0.0767,
+      "step": 19548
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.542719446004118e-05,
+      "loss": 0.0764,
+      "step": 19550
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.542672655811342e-05,
+      "loss": 0.0618,
+      "step": 19552
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5426258656185664e-05,
+      "loss": 0.0635,
+      "step": 19554
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.542579075425791e-05,
+      "loss": 0.078,
+      "step": 19556
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.5425322852330156e-05,
+      "loss": 0.0589,
+      "step": 19558
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.5424854950402395e-05,
+      "loss": 0.0724,
+      "step": 19560
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.542438704847464e-05,
+      "loss": 0.0552,
+      "step": 19562
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.542391914654689e-05,
+      "loss": 0.0413,
+      "step": 19564
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.542345124461913e-05,
+      "loss": 0.0676,
+      "step": 19566
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.542298334269137e-05,
+      "loss": 0.0735,
+      "step": 19568
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.542251544076362e-05,
+      "loss": 0.0909,
+      "step": 19570
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.542204753883586e-05,
+      "loss": 0.0715,
+      "step": 19572
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.542157963690811e-05,
+      "loss": 0.0639,
+      "step": 19574
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.542111173498035e-05,
+      "loss": 0.0744,
+      "step": 19576
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.5420643833052595e-05,
+      "loss": 0.0783,
+      "step": 19578
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.5420175931124834e-05,
+      "loss": 0.0552,
+      "step": 19580
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.541970802919708e-05,
+      "loss": 0.0751,
+      "step": 19582
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.5419240127269326e-05,
+      "loss": 0.0764,
+      "step": 19584
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.541877222534157e-05,
+      "loss": 0.0739,
+      "step": 19586
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.541830432341381e-05,
+      "loss": 0.0482,
+      "step": 19588
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.541783642148606e-05,
+      "loss": 0.0758,
+      "step": 19590
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.54173685195583e-05,
+      "loss": 0.0751,
+      "step": 19592
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.541690061763055e-05,
+      "loss": 0.0566,
+      "step": 19594
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.541643271570279e-05,
+      "loss": 0.0504,
+      "step": 19596
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.5415964813775034e-05,
+      "loss": 0.0535,
+      "step": 19598
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.541549691184728e-05,
+      "loss": 0.0662,
+      "step": 19600
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.5415029009919526e-05,
+      "loss": 0.0714,
+      "step": 19602
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.5414561107991765e-05,
+      "loss": 0.0624,
+      "step": 19604
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.541409320606401e-05,
+      "loss": 0.0571,
+      "step": 19606
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.5413625304136257e-05,
+      "loss": 0.054,
+      "step": 19608
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.54131574022085e-05,
+      "loss": 0.0696,
+      "step": 19610
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.541268950028074e-05,
+      "loss": 0.0785,
+      "step": 19612
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.541222159835299e-05,
+      "loss": 0.0435,
+      "step": 19614
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.541175369642523e-05,
+      "loss": 0.0757,
+      "step": 19616
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.541128579449748e-05,
+      "loss": 0.0575,
+      "step": 19618
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.541081789256972e-05,
+      "loss": 0.089,
+      "step": 19620
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.5410349990641964e-05,
+      "loss": 0.0588,
+      "step": 19622
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.5409882088714204e-05,
+      "loss": 0.0677,
+      "step": 19624
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.5409414186786456e-05,
+      "loss": 0.0845,
+      "step": 19626
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.5408946284858695e-05,
+      "loss": 0.0549,
+      "step": 19628
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.540847838293094e-05,
+      "loss": 0.0841,
+      "step": 19630
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.540801048100318e-05,
+      "loss": 0.0744,
+      "step": 19632
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.5407542579075426e-05,
+      "loss": 0.0634,
+      "step": 19634
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.540707467714767e-05,
+      "loss": 0.0807,
+      "step": 19636
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.540660677521992e-05,
+      "loss": 0.0731,
+      "step": 19638
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.540613887329216e-05,
+      "loss": 0.0739,
+      "step": 19640
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.54056709713644e-05,
+      "loss": 0.0634,
+      "step": 19642
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.540520306943665e-05,
+      "loss": 0.0718,
+      "step": 19644
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.5404735167508895e-05,
+      "loss": 0.0679,
+      "step": 19646
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.5404267265581134e-05,
+      "loss": 0.065,
+      "step": 19648
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.540379936365338e-05,
+      "loss": 0.0652,
+      "step": 19650
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.5403331461725626e-05,
+      "loss": 0.0818,
+      "step": 19652
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.540286355979787e-05,
+      "loss": 0.0758,
+      "step": 19654
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.540239565787011e-05,
+      "loss": 0.0519,
+      "step": 19656
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.540192775594236e-05,
+      "loss": 0.061,
+      "step": 19658
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.54014598540146e-05,
+      "loss": 0.0553,
+      "step": 19660
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.540099195208685e-05,
+      "loss": 0.0802,
+      "step": 19662
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.540052405015909e-05,
+      "loss": 0.0617,
+      "step": 19664
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.5400056148231334e-05,
+      "loss": 0.0555,
+      "step": 19666
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.539958824630357e-05,
+      "loss": 0.0882,
+      "step": 19668
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.5399120344375826e-05,
+      "loss": 0.0765,
+      "step": 19670
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.5398652442448065e-05,
+      "loss": 0.0644,
+      "step": 19672
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.539818454052031e-05,
+      "loss": 0.0619,
+      "step": 19674
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.539771663859255e-05,
+      "loss": 0.0749,
+      "step": 19676
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.53972487366648e-05,
+      "loss": 0.0661,
+      "step": 19678
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.539678083473704e-05,
+      "loss": 0.0654,
+      "step": 19680
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.539631293280929e-05,
+      "loss": 0.0752,
+      "step": 19682
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.539584503088153e-05,
+      "loss": 0.0627,
+      "step": 19684
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.539537712895377e-05,
+      "loss": 0.0608,
+      "step": 19686
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.539490922702602e-05,
+      "loss": 0.0715,
+      "step": 19688
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.5394441325098264e-05,
+      "loss": 0.0614,
+      "step": 19690
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.5393973423170504e-05,
+      "loss": 0.071,
+      "step": 19692
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.539350552124275e-05,
+      "loss": 0.0462,
+      "step": 19694
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.5393037619314995e-05,
+      "loss": 0.0656,
+      "step": 19696
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.539256971738724e-05,
+      "loss": 0.0698,
+      "step": 19698
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.539210181545948e-05,
+      "loss": 0.0709,
+      "step": 19700
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.5391633913531726e-05,
+      "loss": 0.0644,
+      "step": 19702
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.539116601160397e-05,
+      "loss": 0.0823,
+      "step": 19704
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.539069810967622e-05,
+      "loss": 0.0657,
+      "step": 19706
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.539023020774846e-05,
+      "loss": 0.0682,
+      "step": 19708
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.53897623058207e-05,
+      "loss": 0.0574,
+      "step": 19710
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.538929440389294e-05,
+      "loss": 0.0621,
+      "step": 19712
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.5388826501965195e-05,
+      "loss": 0.0777,
+      "step": 19714
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.5388358600037434e-05,
+      "loss": 0.0759,
+      "step": 19716
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.538789069810968e-05,
+      "loss": 0.0735,
+      "step": 19718
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.538742279618192e-05,
+      "loss": 0.0611,
+      "step": 19720
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.538695489425417e-05,
+      "loss": 0.0881,
+      "step": 19722
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.538648699232641e-05,
+      "loss": 0.0563,
+      "step": 19724
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.538601909039866e-05,
+      "loss": 0.0561,
+      "step": 19726
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.5385551188470896e-05,
+      "loss": 0.0569,
+      "step": 19728
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.538508328654314e-05,
+      "loss": 0.077,
+      "step": 19730
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.538461538461539e-05,
+      "loss": 0.0643,
+      "step": 19732
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.5384147482687634e-05,
+      "loss": 0.0555,
+      "step": 19734
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.538367958075987e-05,
+      "loss": 0.0595,
+      "step": 19736
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.538321167883212e-05,
+      "loss": 0.0582,
+      "step": 19738
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.5382743776904365e-05,
+      "loss": 0.0724,
+      "step": 19740
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.538227587497661e-05,
+      "loss": 0.0678,
+      "step": 19742
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.538180797304885e-05,
+      "loss": 0.0564,
+      "step": 19744
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.5381340071121096e-05,
+      "loss": 0.0886,
+      "step": 19746
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.538087216919334e-05,
+      "loss": 0.0591,
+      "step": 19748
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.538040426726559e-05,
+      "loss": 0.0716,
+      "step": 19750
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.537993636533783e-05,
+      "loss": 0.0814,
+      "step": 19752
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.537946846341007e-05,
+      "loss": 0.0917,
+      "step": 19754
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.537900056148232e-05,
+      "loss": 0.0526,
+      "step": 19756
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.537853265955456e-05,
+      "loss": 0.067,
+      "step": 19758
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.5378064757626804e-05,
+      "loss": 0.0629,
+      "step": 19760
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.537759685569904e-05,
+      "loss": 0.0868,
+      "step": 19762
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.537712895377129e-05,
+      "loss": 0.077,
+      "step": 19764
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.5376661051843535e-05,
+      "loss": 0.071,
+      "step": 19766
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.537619314991578e-05,
+      "loss": 0.0748,
+      "step": 19768
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.537572524798802e-05,
+      "loss": 0.0775,
+      "step": 19770
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5375257346060266e-05,
+      "loss": 0.0652,
+      "step": 19772
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.537478944413251e-05,
+      "loss": 0.07,
+      "step": 19774
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.537432154220476e-05,
+      "loss": 0.0767,
+      "step": 19776
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5373853640276996e-05,
+      "loss": 0.0756,
+      "step": 19778
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.537338573834924e-05,
+      "loss": 0.0569,
+      "step": 19780
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.537291783642149e-05,
+      "loss": 0.0516,
+      "step": 19782
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5372449934493734e-05,
+      "loss": 0.0617,
+      "step": 19784
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.537198203256597e-05,
+      "loss": 0.0691,
+      "step": 19786
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.537151413063822e-05,
+      "loss": 0.0813,
+      "step": 19788
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5371046228710465e-05,
+      "loss": 0.0938,
+      "step": 19790
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.537057832678271e-05,
+      "loss": 0.0675,
+      "step": 19792
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.537011042485495e-05,
+      "loss": 0.0616,
+      "step": 19794
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5369642522927196e-05,
+      "loss": 0.0643,
+      "step": 19796
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5369174620999435e-05,
+      "loss": 0.0681,
+      "step": 19798
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.536870671907169e-05,
+      "loss": 0.0562,
+      "step": 19800
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.536823881714393e-05,
+      "loss": 0.0901,
+      "step": 19802
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.536777091521617e-05,
+      "loss": 0.0784,
+      "step": 19804
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.536730301328841e-05,
+      "loss": 0.0787,
+      "step": 19806
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5366835111360665e-05,
+      "loss": 0.0533,
+      "step": 19808
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5366367209432904e-05,
+      "loss": 0.0456,
+      "step": 19810
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.536589930750515e-05,
+      "loss": 0.0672,
+      "step": 19812
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.536543140557739e-05,
+      "loss": 0.0406,
+      "step": 19814
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5364963503649635e-05,
+      "loss": 0.0633,
+      "step": 19816
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.536449560172188e-05,
+      "loss": 0.0619,
+      "step": 19818
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.536402769979413e-05,
+      "loss": 0.0761,
+      "step": 19820
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5363559797866366e-05,
+      "loss": 0.0581,
+      "step": 19822
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.536309189593861e-05,
+      "loss": 0.0584,
+      "step": 19824
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.536262399401086e-05,
+      "loss": 0.064,
+      "step": 19826
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5362156092083104e-05,
+      "loss": 0.0677,
+      "step": 19828
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.536168819015534e-05,
+      "loss": 0.0465,
+      "step": 19830
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.536122028822759e-05,
+      "loss": 0.0571,
+      "step": 19832
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5360752386299835e-05,
+      "loss": 0.0684,
+      "step": 19834
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.536028448437208e-05,
+      "loss": 0.0612,
+      "step": 19836
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.535981658244432e-05,
+      "loss": 0.054,
+      "step": 19838
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5359348680516566e-05,
+      "loss": 0.0587,
+      "step": 19840
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.535888077858881e-05,
+      "loss": 0.0657,
+      "step": 19842
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.535841287666106e-05,
+      "loss": 0.0591,
+      "step": 19844
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.5357944974733296e-05,
+      "loss": 0.0682,
+      "step": 19846
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.535747707280554e-05,
+      "loss": 0.0813,
+      "step": 19848
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.535700917087778e-05,
+      "loss": 0.0878,
+      "step": 19850
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.5356541268950034e-05,
+      "loss": 0.0768,
+      "step": 19852
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.535607336702227e-05,
+      "loss": 0.0755,
+      "step": 19854
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.535560546509452e-05,
+      "loss": 0.0582,
+      "step": 19856
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.535513756316676e-05,
+      "loss": 0.064,
+      "step": 19858
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.5354669661239004e-05,
+      "loss": 0.0593,
+      "step": 19860
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.535420175931125e-05,
+      "loss": 0.0766,
+      "step": 19862
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.5353733857383496e-05,
+      "loss": 0.088,
+      "step": 19864
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.5353265955455735e-05,
+      "loss": 0.079,
+      "step": 19866
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.535279805352798e-05,
+      "loss": 0.0584,
+      "step": 19868
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.535233015160023e-05,
+      "loss": 0.0563,
+      "step": 19870
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.535186224967247e-05,
+      "loss": 0.0555,
+      "step": 19872
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.535139434774471e-05,
+      "loss": 0.0936,
+      "step": 19874
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.535092644581696e-05,
+      "loss": 0.0635,
+      "step": 19876
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.5350458543889204e-05,
+      "loss": 0.0664,
+      "step": 19878
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.534999064196145e-05,
+      "loss": 0.0554,
+      "step": 19880
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.534952274003369e-05,
+      "loss": 0.0917,
+      "step": 19882
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.5349054838105935e-05,
+      "loss": 0.0565,
+      "step": 19884
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.534858693617818e-05,
+      "loss": 0.0716,
+      "step": 19886
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.534811903425043e-05,
+      "loss": 0.0707,
+      "step": 19888
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.5347651132322666e-05,
+      "loss": 0.0745,
+      "step": 19890
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.534718323039491e-05,
+      "loss": 0.0768,
+      "step": 19892
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.534671532846715e-05,
+      "loss": 0.0776,
+      "step": 19894
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.5346247426539404e-05,
+      "loss": 0.0834,
+      "step": 19896
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.534577952461164e-05,
+      "loss": 0.0666,
+      "step": 19898
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.534531162268389e-05,
+      "loss": 0.0813,
+      "step": 19900
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.534484372075613e-05,
+      "loss": 0.077,
+      "step": 19902
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.534437581882838e-05,
+      "loss": 0.0506,
+      "step": 19904
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.534390791690062e-05,
+      "loss": 0.0766,
+      "step": 19906
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.5343440014972866e-05,
+      "loss": 0.0815,
+      "step": 19908
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.5342972113045105e-05,
+      "loss": 0.0618,
+      "step": 19910
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.534250421111735e-05,
+      "loss": 0.0726,
+      "step": 19912
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.5342036309189597e-05,
+      "loss": 0.0688,
+      "step": 19914
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.534156840726184e-05,
+      "loss": 0.0624,
+      "step": 19916
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.534110050533408e-05,
+      "loss": 0.0857,
+      "step": 19918
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.534063260340633e-05,
+      "loss": 0.083,
+      "step": 19920
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.534016470147857e-05,
+      "loss": 0.063,
+      "step": 19922
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.533969679955082e-05,
+      "loss": 0.0587,
+      "step": 19924
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.533922889762306e-05,
+      "loss": 0.0435,
+      "step": 19926
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.5338760995695304e-05,
+      "loss": 0.0605,
+      "step": 19928
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.533829309376755e-05,
+      "loss": 0.0856,
+      "step": 19930
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.5337825191839796e-05,
+      "loss": 0.0854,
+      "step": 19932
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.5337357289912035e-05,
+      "loss": 0.0674,
+      "step": 19934
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.533688938798428e-05,
+      "loss": 0.0453,
+      "step": 19936
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.533642148605653e-05,
+      "loss": 0.0656,
+      "step": 19938
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.533595358412877e-05,
+      "loss": 0.077,
+      "step": 19940
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.533548568220101e-05,
+      "loss": 0.0655,
+      "step": 19942
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.533501778027326e-05,
+      "loss": 0.0688,
+      "step": 19944
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.53345498783455e-05,
+      "loss": 0.0745,
+      "step": 19946
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.533408197641775e-05,
+      "loss": 0.0589,
+      "step": 19948
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.533361407448999e-05,
+      "loss": 0.0559,
+      "step": 19950
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.5333146172562235e-05,
+      "loss": 0.0798,
+      "step": 19952
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.5332678270634474e-05,
+      "loss": 0.0677,
+      "step": 19954
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.533221036870673e-05,
+      "loss": 0.0581,
+      "step": 19956
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.5331742466778966e-05,
+      "loss": 0.073,
+      "step": 19958
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.533127456485121e-05,
+      "loss": 0.0769,
+      "step": 19960
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.533080666292345e-05,
+      "loss": 0.0616,
+      "step": 19962
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.53303387609957e-05,
+      "loss": 0.069,
+      "step": 19964
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.532987085906794e-05,
+      "loss": 0.0505,
+      "step": 19966
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.532940295714019e-05,
+      "loss": 0.0569,
+      "step": 19968
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.532893505521243e-05,
+      "loss": 0.0925,
+      "step": 19970
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.5328467153284674e-05,
+      "loss": 0.0643,
+      "step": 19972
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.532799925135692e-05,
+      "loss": 0.0617,
+      "step": 19974
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.5327531349429166e-05,
+      "loss": 0.0697,
+      "step": 19976
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.5327063447501405e-05,
+      "loss": 0.0583,
+      "step": 19978
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.532659554557365e-05,
+      "loss": 0.0478,
+      "step": 19980
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.5326127643645897e-05,
+      "loss": 0.0592,
+      "step": 19982
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.532565974171814e-05,
+      "loss": 0.0671,
+      "step": 19984
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.532519183979038e-05,
+      "loss": 0.0581,
+      "step": 19986
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.532472393786263e-05,
+      "loss": 0.0655,
+      "step": 19988
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.5324256035934873e-05,
+      "loss": 0.0543,
+      "step": 19990
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.532378813400712e-05,
+      "loss": 0.0843,
+      "step": 19992
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.532332023207936e-05,
+      "loss": 0.0496,
+      "step": 19994
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.5322852330151604e-05,
+      "loss": 0.0775,
+      "step": 19996
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.5322384428223843e-05,
+      "loss": 0.071,
+      "step": 19998
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.5321916526296096e-05,
+      "loss": 0.0646,
+      "step": 20000
+    },
+    {
+      "epoch": 2.81,
+      "eval_gen_len": 30.9655,
+      "eval_loss": 1.0668187141418457,
+      "eval_meteor": 0.0542,
+      "eval_runtime": 15.772,
+      "eval_samples_per_second": 3.677,
+      "eval_steps_per_second": 0.507,
+      "step": 20000
+    }
+  ],
+  "max_steps": 213720,
+  "num_train_epochs": 30,
+  "total_flos": 2.878106005251529e+19,
+  "trial_name": null,
+  "trial_params": null
+}