diff --git "a/LORAs/gpt4-tinyllama-lora/checkpoint-52002/trainer_state.json" "b/LORAs/gpt4-tinyllama-lora/checkpoint-52002/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/LORAs/gpt4-tinyllama-lora/checkpoint-52002/trainer_state.json"
@@ -0,0 +1,31219 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 52002,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999999771892448e-05,
+      "loss": 1.875,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999999087569809e-05,
+      "loss": 2.0139,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999997947032149e-05,
+      "loss": 1.7626,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99999635027957e-05,
+      "loss": 1.8976,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999994297312218e-05,
+      "loss": 1.7202,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999991788130281e-05,
+      "loss": 1.5012,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999988822733987e-05,
+      "loss": 1.3992,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999985401123607e-05,
+      "loss": 1.4993,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999981523299453e-05,
+      "loss": 1.4088,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99997718926188e-05,
+      "loss": 1.2877,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999972399011282e-05,
+      "loss": 1.3944,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999967152548095e-05,
+      "loss": 1.6015,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999961449872801e-05,
+      "loss": 1.6425,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999955290985919e-05,
+      "loss": 1.4663,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999948675888009e-05,
+      "loss": 1.2942,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999941604579676e-05,
+      "loss": 1.3777,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999934077061568e-05,
+      "loss": 1.4109,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999926093334366e-05,
+      "loss": 1.36,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999917653398803e-05,
+      "loss": 1.2688,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999908757255648e-05,
+      "loss": 1.4992,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999899404905713e-05,
+      "loss": 1.35,
+      "step": 210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99988959634985e-05,
+      "loss": 1.477,
+      "step": 220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999879331588954e-05,
+      "loss": 1.3487,
+      "step": 230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999868610623963e-05,
+      "loss": 1.3792,
+      "step": 240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999857433455855e-05,
+      "loss": 1.5009,
+      "step": 250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999845800085648e-05,
+      "loss": 1.4627,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999833710514405e-05,
+      "loss": 1.2347,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99982116474323e-05,
+      "loss": 1.2666,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999808162773266e-05,
+      "loss": 1.2938,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999794704605701e-05,
+      "loss": 1.3973,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999780790241761e-05,
+      "loss": 1.4668,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999766419682716e-05,
+      "loss": 1.3756,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999751592929878e-05,
+      "loss": 1.2096,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999736309984599e-05,
+      "loss": 1.1644,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999720570848277e-05,
+      "loss": 1.0741,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999704375522343e-05,
+      "loss": 1.2642,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999687724008276e-05,
+      "loss": 1.2261,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999670616307597e-05,
+      "loss": 1.4182,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999653052421867e-05,
+      "loss": 1.3194,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999635032352687e-05,
+      "loss": 1.35,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999616556101701e-05,
+      "loss": 1.14,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999597623670597e-05,
+      "loss": 1.4579,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9995782350611e-05,
+      "loss": 1.3264,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999558390274981e-05,
+      "loss": 1.528,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999538089314051e-05,
+      "loss": 1.2982,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99951733218016e-05,
+      "loss": 1.3311,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999496118875204e-05,
+      "loss": 1.2632,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999474449401116e-05,
+      "loss": 1.4559,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999452323759878e-05,
+      "loss": 1.4609,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999429741953504e-05,
+      "loss": 1.3248,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999406703984055e-05,
+      "loss": 1.2921,
+      "step": 510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999383209853637e-05,
+      "loss": 1.2958,
+      "step": 520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999359259564389e-05,
+      "loss": 1.4268,
+      "step": 530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999334853118498e-05,
+      "loss": 1.6085,
+      "step": 540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999309990518194e-05,
+      "loss": 1.4809,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999284671765739e-05,
+      "loss": 1.4938,
+      "step": 560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999258896863448e-05,
+      "loss": 1.5519,
+      "step": 570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999232665813674e-05,
+      "loss": 1.3456,
+      "step": 580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999205978618806e-05,
+      "loss": 1.124,
+      "step": 590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99917883528128e-05,
+      "loss": 1.4275,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999151235803576e-05,
+      "loss": 1.2882,
+      "step": 610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999123180188208e-05,
+      "loss": 1.7446,
+      "step": 620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99909466843774e-05,
+      "loss": 1.3899,
+      "step": 630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99906570055477e-05,
+      "loss": 1.2669,
+      "step": 640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999036276541943e-05,
+      "loss": 1.0545,
+      "step": 650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999006396401942e-05,
+      "loss": 1.3325,
+      "step": 660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998976060137495e-05,
+      "loss": 1.3019,
+      "step": 670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99894526775137e-05,
+      "loss": 1.3425,
+      "step": 680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998914019246377e-05,
+      "loss": 1.5241,
+      "step": 690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998882314625365e-05,
+      "loss": 1.3705,
+      "step": 700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998850153891228e-05,
+      "loss": 1.402,
+      "step": 710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998817537046902e-05,
+      "loss": 1.5151,
+      "step": 720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99878446409536e-05,
+      "loss": 1.1965,
+      "step": 730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998750935039623e-05,
+      "loss": 1.5337,
+      "step": 740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998716949882747e-05,
+      "loss": 1.3151,
+      "step": 750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998682508627835e-05,
+      "loss": 1.3851,
+      "step": 760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99864761127803e-05,
+      "loss": 1.4297,
+      "step": 770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998612257836514e-05,
+      "loss": 1.5431,
+      "step": 780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998576448306515e-05,
+      "loss": 1.5623,
+      "step": 790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998540182691298e-05,
+      "loss": 1.4463,
+      "step": 800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998503460994176e-05,
+      "loss": 1.685,
+      "step": 810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998466283218494e-05,
+      "loss": 1.0598,
+      "step": 820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998428649367648e-05,
+      "loss": 1.3097,
+      "step": 830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998390559445071e-05,
+      "loss": 1.2586,
+      "step": 840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998352013454239e-05,
+      "loss": 1.3021,
+      "step": 850
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998313011398668e-05,
+      "loss": 1.3545,
+      "step": 860
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998273553281916e-05,
+      "loss": 1.2164,
+      "step": 870
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998233639107586e-05,
+      "loss": 1.3504,
+      "step": 880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998193268879317e-05,
+      "loss": 1.5319,
+      "step": 890
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998152442600794e-05,
+      "loss": 1.1939,
+      "step": 900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998111160275742e-05,
+      "loss": 1.5217,
+      "step": 910
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998069421907929e-05,
+      "loss": 1.3646,
+      "step": 920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998027227501161e-05,
+      "loss": 1.1651,
+      "step": 930
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997984577059288e-05,
+      "loss": 1.5059,
+      "step": 940
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997941470586203e-05,
+      "loss": 1.1902,
+      "step": 950
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997897908085839e-05,
+      "loss": 1.2637,
+      "step": 960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99785388956217e-05,
+      "loss": 1.3935,
+      "step": 970
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997809415019213e-05,
+      "loss": 1.3405,
+      "step": 980
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997764484461025e-05,
+      "loss": 1.2142,
+      "step": 990
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997719097891707e-05,
+      "loss": 1.2017,
+      "step": 1000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997673255315401e-05,
+      "loss": 1.5894,
+      "step": 1010
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997626956736286e-05,
+      "loss": 1.2648,
+      "step": 1020
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99758020215859e-05,
+      "loss": 1.3068,
+      "step": 1030
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997532991586578e-05,
+      "loss": 1.3405,
+      "step": 1040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997485325024555e-05,
+      "loss": 1.4155,
+      "step": 1050
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997437202476875e-05,
+      "loss": 1.3779,
+      "step": 1060
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997388623947928e-05,
+      "loss": 1.3744,
+      "step": 1070
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997339589442143e-05,
+      "loss": 1.4678,
+      "step": 1080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997290098963996e-05,
+      "loss": 1.2851,
+      "step": 1090
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997240152518003e-05,
+      "loss": 1.4561,
+      "step": 1100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997189750108721e-05,
+      "loss": 1.3897,
+      "step": 1110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997138891740748e-05,
+      "loss": 1.136,
+      "step": 1120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997087577418727e-05,
+      "loss": 1.143,
+      "step": 1130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997035807147337e-05,
+      "loss": 1.3829,
+      "step": 1140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996983580931304e-05,
+      "loss": 1.5637,
+      "step": 1150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996930898775391e-05,
+      "loss": 1.2417,
+      "step": 1160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996877760684407e-05,
+      "loss": 1.3978,
+      "step": 1170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996824166663201e-05,
+      "loss": 1.1616,
+      "step": 1180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99677011671666e-05,
+      "loss": 1.339,
+      "step": 1190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996715610849718e-05,
+      "loss": 1.651,
+      "step": 1200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996660649067349e-05,
+      "loss": 1.1262,
+      "step": 1210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996605231374565e-05,
+      "loss": 1.3203,
+      "step": 1220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996549357776425e-05,
+      "loss": 1.3868,
+      "step": 1230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996493028278026e-05,
+      "loss": 1.2582,
+      "step": 1240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996436242884508e-05,
+      "loss": 1.4204,
+      "step": 1250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99637900160105e-05,
+      "loss": 1.2503,
+      "step": 1260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99632130443288e-05,
+      "loss": 1.2795,
+      "step": 1270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99626315138526e-05,
+      "loss": 1.3889,
+      "step": 1280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996204542463492e-05,
+      "loss": 1.6772,
+      "step": 1290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99614547767293e-05,
+      "loss": 1.347,
+      "step": 1300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.996085957018959e-05,
+      "loss": 1.3079,
+      "step": 1310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.996025980507012e-05,
+      "loss": 1.4347,
+      "step": 1320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.995965548142559e-05,
+      "loss": 1.2224,
+      "step": 1330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.995904659931118e-05,
+      "loss": 1.2926,
+      "step": 1340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.995843315878241e-05,
+      "loss": 1.2501,
+      "step": 1350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.995781515989528e-05,
+      "loss": 1.3179,
+      "step": 1360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.995719260270616e-05,
+      "loss": 1.1125,
+      "step": 1370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.995656548727184e-05,
+      "loss": 1.3954,
+      "step": 1380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.995593381364958e-05,
+      "loss": 1.3111,
+      "step": 1390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.995529758189696e-05,
+      "loss": 1.2314,
+      "step": 1400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99546567920721e-05,
+      "loss": 1.2019,
+      "step": 1410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.995401144423342e-05,
+      "loss": 1.3759,
+      "step": 1420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.995336153843982e-05,
+      "loss": 1.5097,
+      "step": 1430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.995270707475059e-05,
+      "loss": 1.2159,
+      "step": 1440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.995204805322547e-05,
+      "loss": 1.5312,
+      "step": 1450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.995138447392455e-05,
+      "loss": 1.6136,
+      "step": 1460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99507163369084e-05,
+      "loss": 1.3264,
+      "step": 1470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9950043642238e-05,
+      "loss": 1.409,
+      "step": 1480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99493663899747e-05,
+      "loss": 1.2682,
+      "step": 1490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994868458018029e-05,
+      "loss": 1.2743,
+      "step": 1500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994799821291701e-05,
+      "loss": 1.3114,
+      "step": 1510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994730728824748e-05,
+      "loss": 1.5562,
+      "step": 1520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994661180623472e-05,
+      "loss": 1.3428,
+      "step": 1530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994591176694221e-05,
+      "loss": 1.3074,
+      "step": 1540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994520717043381e-05,
+      "loss": 1.1452,
+      "step": 1550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99444980167738e-05,
+      "loss": 1.1624,
+      "step": 1560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994378430602693e-05,
+      "loss": 1.4339,
+      "step": 1570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994306603825828e-05,
+      "loss": 1.2881,
+      "step": 1580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99423432135334e-05,
+      "loss": 1.3104,
+      "step": 1590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994161583191824e-05,
+      "loss": 1.2123,
+      "step": 1600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994088389347917e-05,
+      "loss": 1.3504,
+      "step": 1610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994014739828297e-05,
+      "loss": 1.4634,
+      "step": 1620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993940634639684e-05,
+      "loss": 1.2906,
+      "step": 1630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99386607378884e-05,
+      "loss": 1.285,
+      "step": 1640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99379105728257e-05,
+      "loss": 1.4598,
+      "step": 1650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993715585127714e-05,
+      "loss": 1.456,
+      "step": 1660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993639657331162e-05,
+      "loss": 1.5713,
+      "step": 1670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993563273899841e-05,
+      "loss": 1.3016,
+      "step": 1680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993486434840723e-05,
+      "loss": 1.1334,
+      "step": 1690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993409140160815e-05,
+      "loss": 1.0233,
+      "step": 1700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993331389867171e-05,
+      "loss": 1.2469,
+      "step": 1710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993253183966884e-05,
+      "loss": 1.1256,
+      "step": 1720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993174522467093e-05,
+      "loss": 1.3883,
+      "step": 1730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993095405374973e-05,
+      "loss": 1.4294,
+      "step": 1740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993015832697742e-05,
+      "loss": 1.33,
+      "step": 1750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.992935804442662e-05,
+      "loss": 1.2521,
+      "step": 1760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.992855320617035e-05,
+      "loss": 1.2279,
+      "step": 1770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.992774381228205e-05,
+      "loss": 1.2524,
+      "step": 1780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.992692986283557e-05,
+      "loss": 1.4408,
+      "step": 1790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.992611135790515e-05,
+      "loss": 1.513,
+      "step": 1800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99252882975655e-05,
+      "loss": 1.3177,
+      "step": 1810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.992446068189172e-05,
+      "loss": 1.2595,
+      "step": 1820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.992362851095931e-05,
+      "loss": 1.3211,
+      "step": 1830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99227917848442e-05,
+      "loss": 1.3533,
+      "step": 1840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.992195050362276e-05,
+      "loss": 1.195,
+      "step": 1850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.992110466737173e-05,
+      "loss": 1.3458,
+      "step": 1860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99202542761683e-05,
+      "loss": 1.3658,
+      "step": 1870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991939933009002e-05,
+      "loss": 1.2936,
+      "step": 1880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991853982921497e-05,
+      "loss": 1.3322,
+      "step": 1890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99176757736215e-05,
+      "loss": 1.2131,
+      "step": 1900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99168071633885e-05,
+      "loss": 1.2902,
+      "step": 1910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99159339985952e-05,
+      "loss": 1.5979,
+      "step": 1920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99150562793213e-05,
+      "loss": 1.3416,
+      "step": 1930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991417400564684e-05,
+      "loss": 1.2639,
+      "step": 1940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991328717765234e-05,
+      "loss": 1.495,
+      "step": 1950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991239579541874e-05,
+      "loss": 1.3425,
+      "step": 1960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991149985902735e-05,
+      "loss": 1.1307,
+      "step": 1970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991059936855991e-05,
+      "loss": 1.5018,
+      "step": 1980
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99096943240986e-05,
+      "loss": 1.1457,
+      "step": 1990
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9908784725726e-05,
+      "loss": 1.3289,
+      "step": 2000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.990787057352508e-05,
+      "loss": 1.3292,
+      "step": 2010
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.990695186757931e-05,
+      "loss": 1.6662,
+      "step": 2020
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.990602860797244e-05,
+      "loss": 1.0469,
+      "step": 2030
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.990510079478875e-05,
+      "loss": 1.2761,
+      "step": 2040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99041684281129e-05,
+      "loss": 1.2587,
+      "step": 2050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.990323150802993e-05,
+      "loss": 1.1929,
+      "step": 2060
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.990229003462539e-05,
+      "loss": 1.6202,
+      "step": 2070
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.990134400798512e-05,
+      "loss": 1.6077,
+      "step": 2080
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.990039342819546e-05,
+      "loss": 1.3055,
+      "step": 2090
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.989943829534316e-05,
+      "loss": 1.4407,
+      "step": 2100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.989847860951535e-05,
+      "loss": 1.5117,
+      "step": 2110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.98975143707996e-05,
+      "loss": 1.4667,
+      "step": 2120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.989654557928388e-05,
+      "loss": 1.5847,
+      "step": 2130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.989557223505661e-05,
+      "loss": 1.2733,
+      "step": 2140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.989459433820658e-05,
+      "loss": 1.321,
+      "step": 2150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.989361188882303e-05,
+      "loss": 1.2065,
+      "step": 2160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.989262488699558e-05,
+      "loss": 1.3144,
+      "step": 2170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.98916333328143e-05,
+      "loss": 1.2705,
+      "step": 2180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.989063722636969e-05,
+      "loss": 1.4665,
+      "step": 2190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.988963656775259e-05,
+      "loss": 1.2998,
+      "step": 2200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.988863135705433e-05,
+      "loss": 1.4893,
+      "step": 2210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.988762159436662e-05,
+      "loss": 1.2859,
+      "step": 2220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.988660727978158e-05,
+      "loss": 1.2458,
+      "step": 2230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.98855884133918e-05,
+      "loss": 1.0838,
+      "step": 2240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.988456499529023e-05,
+      "loss": 1.328,
+      "step": 2250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.988353702557022e-05,
+      "loss": 1.3803,
+      "step": 2260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.98825045043256e-05,
+      "loss": 1.315,
+      "step": 2270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.988146743165054e-05,
+      "loss": 1.2964,
+      "step": 2280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.988042580763972e-05,
+      "loss": 1.3524,
+      "step": 2290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.987937963238813e-05,
+      "loss": 1.3787,
+      "step": 2300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.987832890599126e-05,
+      "loss": 1.284,
+      "step": 2310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.987727362854496e-05,
+      "loss": 1.4093,
+      "step": 2320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.987621380014553e-05,
+      "loss": 1.3782,
+      "step": 2330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.987514942088968e-05,
+      "loss": 1.1652,
+      "step": 2340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987408049087452e-05,
+      "loss": 1.0603,
+      "step": 2350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987300701019755e-05,
+      "loss": 1.4174,
+      "step": 2360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987192897895677e-05,
+      "loss": 1.4376,
+      "step": 2370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987084639725051e-05,
+      "loss": 1.5728,
+      "step": 2380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986975926517755e-05,
+      "loss": 1.3401,
+      "step": 2390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98686675828371e-05,
+      "loss": 1.2169,
+      "step": 2400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986757135032875e-05,
+      "loss": 1.2284,
+      "step": 2410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986647056775254e-05,
+      "loss": 1.2892,
+      "step": 2420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986536523520889e-05,
+      "loss": 1.2494,
+      "step": 2430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986425535279868e-05,
+      "loss": 1.2383,
+      "step": 2440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986314092062315e-05,
+      "loss": 1.7485,
+      "step": 2450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9862021938784e-05,
+      "loss": 1.162,
+      "step": 2460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986089840738333e-05,
+      "loss": 1.3814,
+      "step": 2470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985977032652365e-05,
+      "loss": 1.3171,
+      "step": 2480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98586376963079e-05,
+      "loss": 1.1927,
+      "step": 2490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98575005168394e-05,
+      "loss": 1.5072,
+      "step": 2500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985635878822193e-05,
+      "loss": 1.3116,
+      "step": 2510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985521251055966e-05,
+      "loss": 1.3766,
+      "step": 2520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985406168395719e-05,
+      "loss": 1.36,
+      "step": 2530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985290630851949e-05,
+      "loss": 1.1809,
+      "step": 2540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985174638435202e-05,
+      "loss": 1.4045,
+      "step": 2550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98505819115606e-05,
+      "loss": 1.2913,
+      "step": 2560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984941289025149e-05,
+      "loss": 1.3008,
+      "step": 2570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984823932053133e-05,
+      "loss": 1.1802,
+      "step": 2580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984706120250721e-05,
+      "loss": 1.3826,
+      "step": 2590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984587853628664e-05,
+      "loss": 1.3033,
+      "step": 2600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984469132197752e-05,
+      "loss": 1.4091,
+      "step": 2610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984349955968817e-05,
+      "loss": 1.3903,
+      "step": 2620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984230324952732e-05,
+      "loss": 1.3725,
+      "step": 2630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984110239160416e-05,
+      "loss": 1.4251,
+      "step": 2640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983989698602824e-05,
+      "loss": 1.28,
+      "step": 2650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983868703290953e-05,
+      "loss": 1.4113,
+      "step": 2660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983747253235846e-05,
+      "loss": 1.3604,
+      "step": 2670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983625348448581e-05,
+      "loss": 1.0976,
+      "step": 2680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983502988940285e-05,
+      "loss": 1.3496,
+      "step": 2690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983380174722119e-05,
+      "loss": 1.2368,
+      "step": 2700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98325690580529e-05,
+      "loss": 1.217,
+      "step": 2710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983133182201046e-05,
+      "loss": 1.3363,
+      "step": 2720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983009003920676e-05,
+      "loss": 1.2462,
+      "step": 2730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.982884370975509e-05,
+      "loss": 1.1793,
+      "step": 2740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98275928337692e-05,
+      "loss": 1.3108,
+      "step": 2750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.982633741136317e-05,
+      "loss": 1.4057,
+      "step": 2760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.982507744265162e-05,
+      "loss": 1.2219,
+      "step": 2770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.982381292774945e-05,
+      "loss": 1.2523,
+      "step": 2780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.982254386677206e-05,
+      "loss": 1.2229,
+      "step": 2790
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.982127025983525e-05,
+      "loss": 1.2954,
+      "step": 2800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.981999210705523e-05,
+      "loss": 1.2734,
+      "step": 2810
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98187094085486e-05,
+      "loss": 1.3261,
+      "step": 2820
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.981742216443242e-05,
+      "loss": 1.3873,
+      "step": 2830
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.981613037482414e-05,
+      "loss": 1.2543,
+      "step": 2840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.981483403984162e-05,
+      "loss": 1.2404,
+      "step": 2850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.981353315960315e-05,
+      "loss": 1.2803,
+      "step": 2860
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98122277342274e-05,
+      "loss": 1.3491,
+      "step": 2870
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981091776383351e-05,
+      "loss": 1.4603,
+      "step": 2880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980960324854101e-05,
+      "loss": 1.2107,
+      "step": 2890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980828418846981e-05,
+      "loss": 1.1329,
+      "step": 2900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980696058374029e-05,
+      "loss": 1.4284,
+      "step": 2910
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980563243447321e-05,
+      "loss": 1.1568,
+      "step": 2920
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980429974078975e-05,
+      "loss": 1.5218,
+      "step": 2930
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980296250281152e-05,
+      "loss": 1.232,
+      "step": 2940
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980162072066053e-05,
+      "loss": 1.4554,
+      "step": 2950
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98002743944592e-05,
+      "loss": 1.3907,
+      "step": 2960
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979892352433038e-05,
+      "loss": 1.212,
+      "step": 2970
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979756811039733e-05,
+      "loss": 1.3243,
+      "step": 2980
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97962081527837e-05,
+      "loss": 1.4059,
+      "step": 2990
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979484365161362e-05,
+      "loss": 1.5286,
+      "step": 3000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979347460701157e-05,
+      "loss": 1.3507,
+      "step": 3010
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979210101910244e-05,
+      "loss": 1.3491,
+      "step": 3020
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97907228880116e-05,
+      "loss": 1.2906,
+      "step": 3030
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978934021386477e-05,
+      "loss": 1.3218,
+      "step": 3040
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978795299678812e-05,
+      "loss": 1.4068,
+      "step": 3050
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978656123690822e-05,
+      "loss": 1.14,
+      "step": 3060
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978516493435205e-05,
+      "loss": 1.5555,
+      "step": 3070
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978376408924704e-05,
+      "loss": 1.4966,
+      "step": 3080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978235870172097e-05,
+      "loss": 1.2175,
+      "step": 3090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978094877190209e-05,
+      "loss": 1.4057,
+      "step": 3100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977953429991907e-05,
+      "loss": 1.18,
+      "step": 3110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977811528590094e-05,
+      "loss": 1.1956,
+      "step": 3120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977669172997717e-05,
+      "loss": 1.4967,
+      "step": 3130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977526363227767e-05,
+      "loss": 1.159,
+      "step": 3140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977383099293273e-05,
+      "loss": 1.5311,
+      "step": 3150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97723938120731e-05,
+      "loss": 1.2848,
+      "step": 3160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977095208982987e-05,
+      "loss": 1.207,
+      "step": 3170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976950582633461e-05,
+      "loss": 1.3055,
+      "step": 3180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976805502171926e-05,
+      "loss": 1.325,
+      "step": 3190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976659967611624e-05,
+      "loss": 1.3816,
+      "step": 3200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976513978965829e-05,
+      "loss": 1.1429,
+      "step": 3210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976367536247864e-05,
+      "loss": 1.281,
+      "step": 3220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976220639471092e-05,
+      "loss": 1.1709,
+      "step": 3230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976073288648913e-05,
+      "loss": 1.5605,
+      "step": 3240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975925483794775e-05,
+      "loss": 1.2462,
+      "step": 3250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975777224922163e-05,
+      "loss": 1.5913,
+      "step": 3260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975628512044604e-05,
+      "loss": 1.5767,
+      "step": 3270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975479345175667e-05,
+      "loss": 1.3546,
+      "step": 3280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975329724328961e-05,
+      "loss": 1.3576,
+      "step": 3290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975179649518143e-05,
+      "loss": 1.2719,
+      "step": 3300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9750291207569e-05,
+      "loss": 1.2336,
+      "step": 3310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974878138058972e-05,
+      "loss": 1.3509,
+      "step": 3320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97472670143813e-05,
+      "loss": 1.2767,
+      "step": 3330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974574810908197e-05,
+      "loss": 1.2492,
+      "step": 3340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974422466483029e-05,
+      "loss": 1.2595,
+      "step": 3350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974269668176525e-05,
+      "loss": 1.4714,
+      "step": 3360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974116416002629e-05,
+      "loss": 1.1935,
+      "step": 3370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973962709975324e-05,
+      "loss": 1.2193,
+      "step": 3380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.973808550108634e-05,
+      "loss": 1.4745,
+      "step": 3390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.973653936416625e-05,
+      "loss": 1.2986,
+      "step": 3400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.973498868913404e-05,
+      "loss": 1.488,
+      "step": 3410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.973343347613121e-05,
+      "loss": 1.3544,
+      "step": 3420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.973187372529967e-05,
+      "loss": 1.3733,
+      "step": 3430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.97303094367817e-05,
+      "loss": 1.3409,
+      "step": 3440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.972874061072007e-05,
+      "loss": 1.4664,
+      "step": 3450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.972716724725789e-05,
+      "loss": 1.4377,
+      "step": 3460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.972558934653875e-05,
+      "loss": 1.4165,
+      "step": 3470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.972400690870661e-05,
+      "loss": 1.2062,
+      "step": 3480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.972241993390585e-05,
+      "loss": 1.2318,
+      "step": 3490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.972082842228128e-05,
+      "loss": 1.2213,
+      "step": 3500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.97192323739781e-05,
+      "loss": 1.1936,
+      "step": 3510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.971763178914195e-05,
+      "loss": 1.3197,
+      "step": 3520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.971602666791888e-05,
+      "loss": 1.2308,
+      "step": 3530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.971441701045533e-05,
+      "loss": 1.3912,
+      "step": 3540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.971280281689818e-05,
+      "loss": 1.2397,
+      "step": 3550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.971118408739471e-05,
+      "loss": 1.359,
+      "step": 3560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.970956082209261e-05,
+      "loss": 1.2904,
+      "step": 3570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.970793302114002e-05,
+      "loss": 1.1786,
+      "step": 3580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.970630068468541e-05,
+      "loss": 1.1694,
+      "step": 3590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.970466381287779e-05,
+      "loss": 1.4849,
+      "step": 3600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.970302240586646e-05,
+      "loss": 1.2668,
+      "step": 3610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.97013764638012e-05,
+      "loss": 1.6899,
+      "step": 3620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.969972598683221e-05,
+      "loss": 1.4086,
+      "step": 3630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.969807097511006e-05,
+      "loss": 1.2225,
+      "step": 3640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.969641142878576e-05,
+      "loss": 1.2884,
+      "step": 3650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.969474734801076e-05,
+      "loss": 1.2065,
+      "step": 3660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.969307873293686e-05,
+      "loss": 1.1839,
+      "step": 3670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.969140558371633e-05,
+      "loss": 1.4486,
+      "step": 3680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.968972790050184e-05,
+      "loss": 1.4194,
+      "step": 3690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.968804568344644e-05,
+      "loss": 1.3068,
+      "step": 3700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.968635893270363e-05,
+      "loss": 1.2503,
+      "step": 3710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.968466764842734e-05,
+      "loss": 1.2139,
+      "step": 3720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.968297183077186e-05,
+      "loss": 1.4462,
+      "step": 3730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.968127147989192e-05,
+      "loss": 1.0493,
+      "step": 3740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967956659594268e-05,
+      "loss": 1.7619,
+      "step": 3750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967785717907971e-05,
+      "loss": 1.4466,
+      "step": 3760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967614322945895e-05,
+      "loss": 1.4525,
+      "step": 3770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967442474723681e-05,
+      "loss": 1.3149,
+      "step": 3780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967270173257007e-05,
+      "loss": 1.3513,
+      "step": 3790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967097418561596e-05,
+      "loss": 1.1702,
+      "step": 3800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96692421065321e-05,
+      "loss": 1.3409,
+      "step": 3810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966750549547654e-05,
+      "loss": 1.1612,
+      "step": 3820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966576435260771e-05,
+      "loss": 1.1527,
+      "step": 3830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96640186780845e-05,
+      "loss": 1.5699,
+      "step": 3840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966226847206619e-05,
+      "loss": 1.3352,
+      "step": 3850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966051373471246e-05,
+      "loss": 1.5389,
+      "step": 3860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965875446618342e-05,
+      "loss": 1.05,
+      "step": 3870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965699066663959e-05,
+      "loss": 1.1565,
+      "step": 3880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965522233624191e-05,
+      "loss": 1.1598,
+      "step": 3890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965344947515174e-05,
+      "loss": 1.5574,
+      "step": 3900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.965167208353082e-05,
+      "loss": 1.4266,
+      "step": 3910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.964989016154134e-05,
+      "loss": 1.2152,
+      "step": 3920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.964810370934587e-05,
+      "loss": 1.3679,
+      "step": 3930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.964631272710742e-05,
+      "loss": 1.3647,
+      "step": 3940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.964451721498943e-05,
+      "loss": 1.4123,
+      "step": 3950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.96427171731557e-05,
+      "loss": 1.2552,
+      "step": 3960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.964091260177047e-05,
+      "loss": 1.5544,
+      "step": 3970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.96391035009984e-05,
+      "loss": 1.3068,
+      "step": 3980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.963728987100454e-05,
+      "loss": 1.3945,
+      "step": 3990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.963547171195441e-05,
+      "loss": 1.3676,
+      "step": 4000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.963364902401388e-05,
+      "loss": 1.3087,
+      "step": 4010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.963182180734926e-05,
+      "loss": 1.3994,
+      "step": 4020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.962999006212727e-05,
+      "loss": 1.4359,
+      "step": 4030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.962815378851505e-05,
+      "loss": 1.2962,
+      "step": 4040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.962631298668015e-05,
+      "loss": 1.3718,
+      "step": 4050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.962446765679051e-05,
+      "loss": 1.186,
+      "step": 4060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.962261779901452e-05,
+      "loss": 1.4768,
+      "step": 4070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.962076341352097e-05,
+      "loss": 1.2492,
+      "step": 4080
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.961890450047905e-05,
+      "loss": 1.2654,
+      "step": 4090
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.961704106005838e-05,
+      "loss": 1.2193,
+      "step": 4100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.961517309242898e-05,
+      "loss": 1.0234,
+      "step": 4110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.961330059776128e-05,
+      "loss": 1.466,
+      "step": 4120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.961142357622616e-05,
+      "loss": 1.3056,
+      "step": 4130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.960954202799486e-05,
+      "loss": 1.1035,
+      "step": 4140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.960765595323907e-05,
+      "loss": 1.0571,
+      "step": 4150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.960576535213088e-05,
+      "loss": 1.6929,
+      "step": 4160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.960387022484279e-05,
+      "loss": 1.2753,
+      "step": 4170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.960197057154772e-05,
+      "loss": 1.5613,
+      "step": 4180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.960006639241898e-05,
+      "loss": 1.4367,
+      "step": 4190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.959815768763037e-05,
+      "loss": 1.3649,
+      "step": 4200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.959624445735598e-05,
+      "loss": 1.2436,
+      "step": 4210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.959432670177041e-05,
+      "loss": 1.3484,
+      "step": 4220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.959240442104864e-05,
+      "loss": 1.346,
+      "step": 4230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.959047761536607e-05,
+      "loss": 1.4762,
+      "step": 4240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.958854628489851e-05,
+      "loss": 1.3351,
+      "step": 4250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.958661042982216e-05,
+      "loss": 1.3207,
+      "step": 4260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.958467005031366e-05,
+      "loss": 1.3722,
+      "step": 4270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.958272514655006e-05,
+      "loss": 1.6121,
+      "step": 4280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.958077571870882e-05,
+      "loss": 1.5082,
+      "step": 4290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.957882176696783e-05,
+      "loss": 1.2211,
+      "step": 4300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.957686329150534e-05,
+      "loss": 1.5882,
+      "step": 4310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.957490029250007e-05,
+      "loss": 1.3154,
+      "step": 4320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.957293277013111e-05,
+      "loss": 1.5031,
+      "step": 4330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.957096072457801e-05,
+      "loss": 1.2494,
+      "step": 4340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.95689841560207e-05,
+      "loss": 1.2569,
+      "step": 4350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.95670030646395e-05,
+      "loss": 1.3808,
+      "step": 4360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.95650174506152e-05,
+      "loss": 1.3327,
+      "step": 4370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.956302731412895e-05,
+      "loss": 1.3175,
+      "step": 4380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.956103265536236e-05,
+      "loss": 1.3404,
+      "step": 4390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.955903347449743e-05,
+      "loss": 1.1529,
+      "step": 4400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.955702977171654e-05,
+      "loss": 1.4676,
+      "step": 4410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.955502154720253e-05,
+      "loss": 1.3654,
+      "step": 4420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.955300880113866e-05,
+      "loss": 1.307,
+      "step": 4430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.955099153370855e-05,
+      "loss": 1.1598,
+      "step": 4440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.954896974509627e-05,
+      "loss": 1.1874,
+      "step": 4450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.95469434354863e-05,
+      "loss": 1.386,
+      "step": 4460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.954491260506351e-05,
+      "loss": 1.147,
+      "step": 4470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.954287725401322e-05,
+      "loss": 1.3012,
+      "step": 4480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.954083738252114e-05,
+      "loss": 1.1995,
+      "step": 4490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.953879299077337e-05,
+      "loss": 1.1864,
+      "step": 4500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.953674407895647e-05,
+      "loss": 1.2765,
+      "step": 4510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.953469064725737e-05,
+      "loss": 1.3308,
+      "step": 4520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.953263269586346e-05,
+      "loss": 1.4213,
+      "step": 4530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.953057022496248e-05,
+      "loss": 1.3585,
+      "step": 4540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.952850323474265e-05,
+      "loss": 1.4576,
+      "step": 4550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.952643172539253e-05,
+      "loss": 1.4012,
+      "step": 4560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.952435569710117e-05,
+      "loss": 1.2296,
+      "step": 4570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.952227515005797e-05,
+      "loss": 1.5211,
+      "step": 4580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.952019008445277e-05,
+      "loss": 1.4293,
+      "step": 4590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.951810050047581e-05,
+      "loss": 1.1504,
+      "step": 4600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.951600639831778e-05,
+      "loss": 1.3165,
+      "step": 4610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.951390777816971e-05,
+      "loss": 1.3421,
+      "step": 4620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.95118046402231e-05,
+      "loss": 1.233,
+      "step": 4630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.950969698466986e-05,
+      "loss": 1.264,
+      "step": 4640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.950758481170229e-05,
+      "loss": 1.3277,
+      "step": 4650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.950546812151311e-05,
+      "loss": 1.3612,
+      "step": 4660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.950334691429545e-05,
+      "loss": 1.2901,
+      "step": 4670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.950122119024287e-05,
+      "loss": 1.3418,
+      "step": 4680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.949909094954933e-05,
+      "loss": 1.0178,
+      "step": 4690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.949695619240916e-05,
+      "loss": 1.2333,
+      "step": 4700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.949481691901718e-05,
+      "loss": 1.5076,
+      "step": 4710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.949267312956855e-05,
+      "loss": 1.3271,
+      "step": 4720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.949052482425893e-05,
+      "loss": 1.448,
+      "step": 4730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.948837200328429e-05,
+      "loss": 1.3328,
+      "step": 4740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.948621466684108e-05,
+      "loss": 1.3179,
+      "step": 4750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.948405281512614e-05,
+      "loss": 1.3409,
+      "step": 4760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.948188644833671e-05,
+      "loss": 1.3366,
+      "step": 4770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.947971556667048e-05,
+      "loss": 1.3604,
+      "step": 4780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.947754017032551e-05,
+      "loss": 1.4013,
+      "step": 4790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.947536025950031e-05,
+      "loss": 1.5929,
+      "step": 4800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.947317583439375e-05,
+      "loss": 1.3983,
+      "step": 4810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.947098689520516e-05,
+      "loss": 1.3858,
+      "step": 4820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.946879344213426e-05,
+      "loss": 1.3243,
+      "step": 4830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.94665954753812e-05,
+      "loss": 1.0476,
+      "step": 4840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.946439299514653e-05,
+      "loss": 1.2685,
+      "step": 4850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.94621860016312e-05,
+      "loss": 1.4931,
+      "step": 4860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.945997449503657e-05,
+      "loss": 1.2334,
+      "step": 4870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.945775847556445e-05,
+      "loss": 1.2166,
+      "step": 4880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.945553794341703e-05,
+      "loss": 1.2837,
+      "step": 4890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.945331289879692e-05,
+      "loss": 1.3663,
+      "step": 4900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.945108334190713e-05,
+      "loss": 1.3059,
+      "step": 4910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.944884927295108e-05,
+      "loss": 1.3455,
+      "step": 4920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.944661069213264e-05,
+      "loss": 1.3991,
+      "step": 4930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.944436759965607e-05,
+      "loss": 1.3451,
+      "step": 4940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.9442119995726e-05,
+      "loss": 1.4169,
+      "step": 4950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.943986788054753e-05,
+      "loss": 1.2441,
+      "step": 4960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.943761125432616e-05,
+      "loss": 1.4679,
+      "step": 4970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.943535011726776e-05,
+      "loss": 1.2354,
+      "step": 4980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.943308446957868e-05,
+      "loss": 1.1651,
+      "step": 4990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.943081431146563e-05,
+      "loss": 1.0765,
+      "step": 5000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.942853964313574e-05,
+      "loss": 1.2185,
+      "step": 5010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.942626046479655e-05,
+      "loss": 1.3779,
+      "step": 5020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.942397677665605e-05,
+      "loss": 1.3761,
+      "step": 5030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.942168857892258e-05,
+      "loss": 1.1939,
+      "step": 5040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.941939587180492e-05,
+      "loss": 1.4915,
+      "step": 5050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94170986555123e-05,
+      "loss": 1.1359,
+      "step": 5060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.941479693025428e-05,
+      "loss": 1.3325,
+      "step": 5070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94124906962409e-05,
+      "loss": 1.2584,
+      "step": 5080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94101799536826e-05,
+      "loss": 1.1938,
+      "step": 5090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94078647027902e-05,
+      "loss": 1.1098,
+      "step": 5100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.940554494377495e-05,
+      "loss": 1.1939,
+      "step": 5110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.940322067684852e-05,
+      "loss": 1.2726,
+      "step": 5120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.940089190222297e-05,
+      "loss": 1.2925,
+      "step": 5130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.939855862011081e-05,
+      "loss": 1.3837,
+      "step": 5140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.93962208307249e-05,
+      "loss": 1.436,
+      "step": 5150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.93938785342786e-05,
+      "loss": 1.2485,
+      "step": 5160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.939153173098556e-05,
+      "loss": 1.3532,
+      "step": 5170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.938918042105996e-05,
+      "loss": 1.2223,
+      "step": 5180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.938682460471634e-05,
+      "loss": 1.4696,
+      "step": 5190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.938446428216962e-05,
+      "loss": 1.443,
+      "step": 5200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.938209945363518e-05,
+      "loss": 1.2612,
+      "step": 5210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.937973011932879e-05,
+      "loss": 1.3298,
+      "step": 5220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.937735627946664e-05,
+      "loss": 1.3296,
+      "step": 5230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.937497793426536e-05,
+      "loss": 1.2662,
+      "step": 5240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.937259508394189e-05,
+      "loss": 1.3806,
+      "step": 5250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.937020772871368e-05,
+      "loss": 1.3534,
+      "step": 5260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.936781586879857e-05,
+      "loss": 1.1739,
+      "step": 5270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.936541950441479e-05,
+      "loss": 1.3313,
+      "step": 5280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.936301863578099e-05,
+      "loss": 1.1424,
+      "step": 5290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.936061326311626e-05,
+      "loss": 1.4169,
+      "step": 5300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.935820338664002e-05,
+      "loss": 1.3815,
+      "step": 5310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.935578900657218e-05,
+      "loss": 1.0693,
+      "step": 5320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.935337012313307e-05,
+      "loss": 1.2608,
+      "step": 5330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.935094673654333e-05,
+      "loss": 1.2212,
+      "step": 5340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.934851884702414e-05,
+      "loss": 1.2499,
+      "step": 5350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.934608645479698e-05,
+      "loss": 1.1823,
+      "step": 5360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.934364956008382e-05,
+      "loss": 1.2197,
+      "step": 5370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.934120816310699e-05,
+      "loss": 1.2803,
+      "step": 5380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.933876226408927e-05,
+      "loss": 1.0522,
+      "step": 5390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.933631186325382e-05,
+      "loss": 1.1957,
+      "step": 5400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.933385696082422e-05,
+      "loss": 1.2611,
+      "step": 5410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.933139755702445e-05,
+      "loss": 1.3267,
+      "step": 5420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.932893365207896e-05,
+      "loss": 1.2291,
+      "step": 5430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.93264652462125e-05,
+      "loss": 1.3878,
+      "step": 5440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.932399233965035e-05,
+      "loss": 1.3184,
+      "step": 5450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.932151493261812e-05,
+      "loss": 1.437,
+      "step": 5460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.931903302534185e-05,
+      "loss": 1.5066,
+      "step": 5470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.931654661804801e-05,
+      "loss": 1.2534,
+      "step": 5480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.931405571096346e-05,
+      "loss": 1.1359,
+      "step": 5490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.931156030431548e-05,
+      "loss": 1.3241,
+      "step": 5500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.930906039833177e-05,
+      "loss": 1.5322,
+      "step": 5510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.930655599324041e-05,
+      "loss": 1.3395,
+      "step": 5520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.930404708926991e-05,
+      "loss": 1.3896,
+      "step": 5530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.930153368664921e-05,
+      "loss": 1.2139,
+      "step": 5540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.929901578560763e-05,
+      "loss": 1.2718,
+      "step": 5550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.929649338637491e-05,
+      "loss": 1.2648,
+      "step": 5560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.92939664891812e-05,
+      "loss": 1.2574,
+      "step": 5570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.929143509425705e-05,
+      "loss": 1.3138,
+      "step": 5580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.928889920183344e-05,
+      "loss": 1.2763,
+      "step": 5590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.928635881214177e-05,
+      "loss": 1.2074,
+      "step": 5600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.928381392541382e-05,
+      "loss": 1.414,
+      "step": 5610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.928126454188178e-05,
+      "loss": 1.3734,
+      "step": 5620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.927871066177829e-05,
+      "loss": 1.1972,
+      "step": 5630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.927615228533635e-05,
+      "loss": 1.1967,
+      "step": 5640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.927358941278942e-05,
+      "loss": 1.6523,
+      "step": 5650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.927102204437133e-05,
+      "loss": 1.1186,
+      "step": 5660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.926845018031631e-05,
+      "loss": 1.1298,
+      "step": 5670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.926587382085906e-05,
+      "loss": 1.2431,
+      "step": 5680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.926329296623465e-05,
+      "loss": 1.2669,
+      "step": 5690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.926070761667854e-05,
+      "loss": 1.1762,
+      "step": 5700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.925811777242664e-05,
+      "loss": 1.4493,
+      "step": 5710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.925552343371527e-05,
+      "loss": 1.1156,
+      "step": 5720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.925292460078113e-05,
+      "loss": 1.3443,
+      "step": 5730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.925032127386136e-05,
+      "loss": 1.2967,
+      "step": 5740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.924771345319346e-05,
+      "loss": 1.5565,
+      "step": 5750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.924510113901542e-05,
+      "loss": 1.3646,
+      "step": 5760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.924248433156556e-05,
+      "loss": 1.0625,
+      "step": 5770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.923986303108266e-05,
+      "loss": 1.1386,
+      "step": 5780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.92372372378059e-05,
+      "loss": 1.4176,
+      "step": 5790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.923460695197486e-05,
+      "loss": 1.4019,
+      "step": 5800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.923197217382953e-05,
+      "loss": 1.2456,
+      "step": 5810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.922933290361034e-05,
+      "loss": 1.7389,
+      "step": 5820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.922668914155806e-05,
+      "loss": 1.2519,
+      "step": 5830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.922404088791396e-05,
+      "loss": 1.3408,
+      "step": 5840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.922138814291965e-05,
+      "loss": 1.3465,
+      "step": 5850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.921873090681716e-05,
+      "loss": 1.2421,
+      "step": 5860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.9216069179849e-05,
+      "loss": 1.2956,
+      "step": 5870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.921340296225798e-05,
+      "loss": 1.1981,
+      "step": 5880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.921073225428739e-05,
+      "loss": 1.2752,
+      "step": 5890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.920805705618091e-05,
+      "loss": 1.1301,
+      "step": 5900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.920537736818265e-05,
+      "loss": 1.3806,
+      "step": 5910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.92026931905371e-05,
+      "loss": 1.3252,
+      "step": 5920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.920000452348917e-05,
+      "loss": 1.3651,
+      "step": 5930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.919731136728419e-05,
+      "loss": 1.3466,
+      "step": 5940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.919461372216788e-05,
+      "loss": 1.1675,
+      "step": 5950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.91919115883864e-05,
+      "loss": 1.2723,
+      "step": 5960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.918920496618626e-05,
+      "loss": 1.2592,
+      "step": 5970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.918649385581449e-05,
+      "loss": 1.2989,
+      "step": 5980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.91837782575184e-05,
+      "loss": 1.5011,
+      "step": 5990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.91810581715458e-05,
+      "loss": 1.2073,
+      "step": 6000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.917833359814484e-05,
+      "loss": 1.3457,
+      "step": 6010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.917560453756419e-05,
+      "loss": 1.2271,
+      "step": 6020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.917287099005276e-05,
+      "loss": 1.4907,
+      "step": 6030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.917013295586007e-05,
+      "loss": 1.4253,
+      "step": 6040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.916739043523588e-05,
+      "loss": 1.4048,
+      "step": 6050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.916464342843044e-05,
+      "loss": 1.3438,
+      "step": 6060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.91618919356944e-05,
+      "loss": 1.3314,
+      "step": 6070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.915913595727882e-05,
+      "loss": 1.5093,
+      "step": 6080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.915637549343514e-05,
+      "loss": 1.3973,
+      "step": 6090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.915361054441528e-05,
+      "loss": 1.253,
+      "step": 6100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.915084111047147e-05,
+      "loss": 1.4922,
+      "step": 6110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.914806719185643e-05,
+      "loss": 1.1819,
+      "step": 6120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.914528878882326e-05,
+      "loss": 1.2614,
+      "step": 6130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.914250590162546e-05,
+      "loss": 1.3158,
+      "step": 6140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.913971853051696e-05,
+      "loss": 1.5722,
+      "step": 6150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.913692667575208e-05,
+      "loss": 1.2991,
+      "step": 6160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.913413033758554e-05,
+      "loss": 1.2018,
+      "step": 6170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.913132951627251e-05,
+      "loss": 1.185,
+      "step": 6180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.912852421206856e-05,
+      "loss": 1.3143,
+      "step": 6190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.912571442522963e-05,
+      "loss": 1.432,
+      "step": 6200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.912290015601208e-05,
+      "loss": 1.228,
+      "step": 6210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.912008140467274e-05,
+      "loss": 1.3895,
+      "step": 6220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.911725817146875e-05,
+      "loss": 1.2642,
+      "step": 6230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.911443045665776e-05,
+      "loss": 1.273,
+      "step": 6240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.911159826049774e-05,
+      "loss": 1.4531,
+      "step": 6250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.91087615832471e-05,
+      "loss": 1.193,
+      "step": 6260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.910592042516471e-05,
+      "loss": 1.4373,
+      "step": 6270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.910307478650978e-05,
+      "loss": 1.4789,
+      "step": 6280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.910022466754195e-05,
+      "loss": 1.2663,
+      "step": 6290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.909737006852128e-05,
+      "loss": 1.2694,
+      "step": 6300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.909451098970824e-05,
+      "loss": 1.4176,
+      "step": 6310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.909164743136369e-05,
+      "loss": 1.3513,
+      "step": 6320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.908877939374892e-05,
+      "loss": 1.2542,
+      "step": 6330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.90859068771256e-05,
+      "loss": 1.3829,
+      "step": 6340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.908302988175586e-05,
+      "loss": 1.1073,
+      "step": 6350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.908014840790216e-05,
+      "loss": 1.282,
+      "step": 6360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.907726245582747e-05,
+      "loss": 1.1775,
+      "step": 6370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.907437202579507e-05,
+      "loss": 1.5318,
+      "step": 6380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.907147711806869e-05,
+      "loss": 1.124,
+      "step": 6390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.906857773291249e-05,
+      "loss": 1.3886,
+      "step": 6400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.906567387059103e-05,
+      "loss": 1.3465,
+      "step": 6410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.906276553136923e-05,
+      "loss": 1.3887,
+      "step": 6420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.90598527155125e-05,
+      "loss": 1.2334,
+      "step": 6430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.905693542328658e-05,
+      "loss": 1.2251,
+      "step": 6440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.905401365495766e-05,
+      "loss": 1.3017,
+      "step": 6450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.905108741079232e-05,
+      "loss": 1.3361,
+      "step": 6460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.90481566910576e-05,
+      "loss": 1.4113,
+      "step": 6470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.904522149602086e-05,
+      "loss": 1.2214,
+      "step": 6480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.904228182594995e-05,
+      "loss": 1.4428,
+      "step": 6490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.903933768111306e-05,
+      "loss": 1.2321,
+      "step": 6500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.903638906177887e-05,
+      "loss": 1.1375,
+      "step": 6510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.903343596821637e-05,
+      "loss": 1.3733,
+      "step": 6520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.903047840069505e-05,
+      "loss": 1.2392,
+      "step": 6530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.902751635948475e-05,
+      "loss": 1.7574,
+      "step": 6540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.902454984485574e-05,
+      "loss": 1.3911,
+      "step": 6550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.902157885707868e-05,
+      "loss": 0.9818,
+      "step": 6560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.901860339642468e-05,
+      "loss": 1.4248,
+      "step": 6570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.90156234631652e-05,
+      "loss": 1.3005,
+      "step": 6580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.901263905757215e-05,
+      "loss": 1.2265,
+      "step": 6590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.900965017991784e-05,
+      "loss": 1.2636,
+      "step": 6600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.900665683047499e-05,
+      "loss": 1.3453,
+      "step": 6610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.900365900951669e-05,
+      "loss": 1.4115,
+      "step": 6620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.900065671731652e-05,
+      "loss": 1.2485,
+      "step": 6630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.899764995414838e-05,
+      "loss": 1.1799,
+      "step": 6640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.899463872028663e-05,
+      "loss": 1.295,
+      "step": 6650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.899162301600601e-05,
+      "loss": 1.3853,
+      "step": 6660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.898860284158172e-05,
+      "loss": 1.2228,
+      "step": 6670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.898557819728928e-05,
+      "loss": 1.3557,
+      "step": 6680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.89825490834047e-05,
+      "loss": 1.319,
+      "step": 6690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.897951550020436e-05,
+      "loss": 1.5404,
+      "step": 6700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.897647744796504e-05,
+      "loss": 1.2451,
+      "step": 6710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.897343492696395e-05,
+      "loss": 1.1714,
+      "step": 6720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.897038793747873e-05,
+      "loss": 1.3124,
+      "step": 6730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.896733647978734e-05,
+      "loss": 1.3044,
+      "step": 6740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.896428055416825e-05,
+      "loss": 1.4287,
+      "step": 6750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.896122016090024e-05,
+      "loss": 1.5254,
+      "step": 6760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.895815530026261e-05,
+      "loss": 1.4553,
+      "step": 6770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.895508597253498e-05,
+      "loss": 1.1696,
+      "step": 6780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.895201217799739e-05,
+      "loss": 1.4748,
+      "step": 6790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.894893391693032e-05,
+      "loss": 1.5152,
+      "step": 6800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.894585118961464e-05,
+      "loss": 1.2252,
+      "step": 6810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.894276399633163e-05,
+      "loss": 1.4116,
+      "step": 6820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.893967233736297e-05,
+      "loss": 1.2627,
+      "step": 6830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.893657621299072e-05,
+      "loss": 1.3812,
+      "step": 6840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.893347562349744e-05,
+      "loss": 1.404,
+      "step": 6850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.893037056916601e-05,
+      "loss": 1.1933,
+      "step": 6860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.892726105027974e-05,
+      "loss": 1.5206,
+      "step": 6870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.892414706712235e-05,
+      "loss": 1.1466,
+      "step": 6880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.892102861997797e-05,
+      "loss": 1.2785,
+      "step": 6890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.891790570913114e-05,
+      "loss": 1.2178,
+      "step": 6900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.89147783348668e-05,
+      "loss": 1.3832,
+      "step": 6910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.89116464974703e-05,
+      "loss": 1.2707,
+      "step": 6920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.890851019722742e-05,
+      "loss": 1.5353,
+      "step": 6930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.890536943442431e-05,
+      "loss": 1.4406,
+      "step": 6940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.890222420934754e-05,
+      "loss": 1.23,
+      "step": 6950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.889907452228409e-05,
+      "loss": 1.1093,
+      "step": 6960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.889592037352133e-05,
+      "loss": 1.3189,
+      "step": 6970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.889276176334709e-05,
+      "loss": 1.1306,
+      "step": 6980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.888959869204954e-05,
+      "loss": 1.1474,
+      "step": 6990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.888643115991733e-05,
+      "loss": 1.2099,
+      "step": 7000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.888325916723942e-05,
+      "loss": 1.3138,
+      "step": 7010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.888008271430527e-05,
+      "loss": 1.0532,
+      "step": 7020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.88769018014047e-05,
+      "loss": 1.4223,
+      "step": 7030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.887371642882794e-05,
+      "loss": 1.3858,
+      "step": 7040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.887052659686563e-05,
+      "loss": 1.0461,
+      "step": 7050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.886733230580883e-05,
+      "loss": 1.6259,
+      "step": 7060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.886413355594901e-05,
+      "loss": 1.4475,
+      "step": 7070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.8860930347578e-05,
+      "loss": 1.1167,
+      "step": 7080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.88577226809881e-05,
+      "loss": 1.1784,
+      "step": 7090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.885451055647196e-05,
+      "loss": 1.4284,
+      "step": 7100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.885129397432269e-05,
+      "loss": 1.3067,
+      "step": 7110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.884807293483378e-05,
+      "loss": 1.3647,
+      "step": 7120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.884484743829909e-05,
+      "loss": 1.4115,
+      "step": 7130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.884161748501298e-05,
+      "loss": 1.1733,
+      "step": 7140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.883838307527012e-05,
+      "loss": 1.3816,
+      "step": 7150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.883514420936564e-05,
+      "loss": 1.3084,
+      "step": 7160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.883190088759506e-05,
+      "loss": 1.5996,
+      "step": 7170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.882865311025431e-05,
+      "loss": 1.3527,
+      "step": 7180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.882540087763973e-05,
+      "loss": 1.4004,
+      "step": 7190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.882214419004807e-05,
+      "loss": 1.3129,
+      "step": 7200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.881888304777648e-05,
+      "loss": 1.3845,
+      "step": 7210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.88156174511225e-05,
+      "loss": 1.2199,
+      "step": 7220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.881234740038412e-05,
+      "loss": 1.3709,
+      "step": 7230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.880907289585967e-05,
+      "loss": 1.1905,
+      "step": 7240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.880579393784797e-05,
+      "loss": 1.287,
+      "step": 7250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.880251052664816e-05,
+      "loss": 1.1743,
+      "step": 7260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.879922266255986e-05,
+      "loss": 1.1883,
+      "step": 7270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.879593034588306e-05,
+      "loss": 1.4818,
+      "step": 7280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.879263357691814e-05,
+      "loss": 1.1763,
+      "step": 7290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.878933235596594e-05,
+      "loss": 1.3497,
+      "step": 7300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.878602668332764e-05,
+      "loss": 1.3765,
+      "step": 7310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.878271655930488e-05,
+      "loss": 1.2395,
+      "step": 7320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.877940198419968e-05,
+      "loss": 1.3532,
+      "step": 7330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.877608295831447e-05,
+      "loss": 1.2763,
+      "step": 7340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.87727594819521e-05,
+      "loss": 1.3,
+      "step": 7350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.876943155541578e-05,
+      "loss": 1.2306,
+      "step": 7360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.876609917900921e-05,
+      "loss": 1.4453,
+      "step": 7370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.876276235303641e-05,
+      "loss": 1.2722,
+      "step": 7380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.875942107780185e-05,
+      "loss": 1.135,
+      "step": 7390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.87560753536104e-05,
+      "loss": 1.4157,
+      "step": 7400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.875272518076734e-05,
+      "loss": 1.52,
+      "step": 7410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.874937055957834e-05,
+      "loss": 1.3416,
+      "step": 7420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.87460114903495e-05,
+      "loss": 1.2881,
+      "step": 7430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.874264797338729e-05,
+      "loss": 1.5264,
+      "step": 7440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.873928000899863e-05,
+      "loss": 1.3366,
+      "step": 7450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.873590759749081e-05,
+      "loss": 1.3968,
+      "step": 7460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.873253073917153e-05,
+      "loss": 1.1943,
+      "step": 7470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.872914943434892e-05,
+      "loss": 1.33,
+      "step": 7480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.872576368333151e-05,
+      "loss": 1.1105,
+      "step": 7490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.87223734864282e-05,
+      "loss": 1.2994,
+      "step": 7500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.871897884394833e-05,
+      "loss": 1.5565,
+      "step": 7510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.871557975620164e-05,
+      "loss": 1.2055,
+      "step": 7520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.871217622349829e-05,
+      "loss": 1.4947,
+      "step": 7530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.87087682461488e-05,
+      "loss": 1.2159,
+      "step": 7540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.870535582446416e-05,
+      "loss": 1.3459,
+      "step": 7550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.87019389587557e-05,
+      "loss": 1.3822,
+      "step": 7560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.869851764933518e-05,
+      "loss": 1.3279,
+      "step": 7570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.869509189651477e-05,
+      "loss": 1.2908,
+      "step": 7580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.86916617006071e-05,
+      "loss": 1.2709,
+      "step": 7590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.868822706192509e-05,
+      "loss": 1.4026,
+      "step": 7600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.868478798078214e-05,
+      "loss": 1.4043,
+      "step": 7610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.868134445749207e-05,
+      "loss": 1.3301,
+      "step": 7620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.867789649236903e-05,
+      "loss": 1.3424,
+      "step": 7630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.867444408572767e-05,
+      "loss": 1.3771,
+      "step": 7640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.867098723788297e-05,
+      "loss": 1.3783,
+      "step": 7650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.866752594915036e-05,
+      "loss": 1.1698,
+      "step": 7660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.866406021984565e-05,
+      "loss": 1.3314,
+      "step": 7670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.866059005028505e-05,
+      "loss": 1.2497,
+      "step": 7680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.86571154407852e-05,
+      "loss": 1.172,
+      "step": 7690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.865363639166314e-05,
+      "loss": 1.1296,
+      "step": 7700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.865015290323631e-05,
+      "loss": 1.4019,
+      "step": 7710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.864666497582255e-05,
+      "loss": 1.3886,
+      "step": 7720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.864317260974011e-05,
+      "loss": 1.2295,
+      "step": 7730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.863967580530765e-05,
+      "loss": 1.4068,
+      "step": 7740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.86361745628442e-05,
+      "loss": 1.3704,
+      "step": 7750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.863266888266924e-05,
+      "loss": 1.1668,
+      "step": 7760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.862915876510266e-05,
+      "loss": 1.3474,
+      "step": 7770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.862564421046472e-05,
+      "loss": 1.1253,
+      "step": 7780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.86221252190761e-05,
+      "loss": 1.4082,
+      "step": 7790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.861860179125785e-05,
+      "loss": 1.1985,
+      "step": 7800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.861507392733152e-05,
+      "loss": 1.2986,
+      "step": 7810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.861154162761896e-05,
+      "loss": 1.1855,
+      "step": 7820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.860800489244246e-05,
+      "loss": 1.3365,
+      "step": 7830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.860446372212478e-05,
+      "loss": 1.3017,
+      "step": 7840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.860091811698895e-05,
+      "loss": 1.1642,
+      "step": 7850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.859736807735853e-05,
+      "loss": 1.2197,
+      "step": 7860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.859381360355743e-05,
+      "loss": 1.3722,
+      "step": 7870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.859025469590997e-05,
+      "loss": 1.3946,
+      "step": 7880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.858669135474088e-05,
+      "loss": 1.4931,
+      "step": 7890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.858312358037526e-05,
+      "loss": 1.5107,
+      "step": 7900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.857955137313868e-05,
+      "loss": 1.3741,
+      "step": 7910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.857597473335707e-05,
+      "loss": 1.4331,
+      "step": 7920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.857239366135676e-05,
+      "loss": 1.2076,
+      "step": 7930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.856880815746452e-05,
+      "loss": 1.2076,
+      "step": 7940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.856521822200746e-05,
+      "loss": 1.2114,
+      "step": 7950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.856162385531319e-05,
+      "loss": 1.1001,
+      "step": 7960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.855802505770964e-05,
+      "loss": 1.2581,
+      "step": 7970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.855442182952519e-05,
+      "loss": 1.0122,
+      "step": 7980
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.855081417108859e-05,
+      "loss": 1.0096,
+      "step": 7990
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.854720208272903e-05,
+      "loss": 1.3871,
+      "step": 8000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.854358556477608e-05,
+      "loss": 1.2386,
+      "step": 8010
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.853996461755972e-05,
+      "loss": 1.2004,
+      "step": 8020
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.853633924141036e-05,
+      "loss": 1.281,
+      "step": 8030
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.853270943665875e-05,
+      "loss": 1.1245,
+      "step": 8040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.852907520363613e-05,
+      "loss": 1.2192,
+      "step": 8050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.852543654267405e-05,
+      "loss": 1.4254,
+      "step": 8060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.852179345410454e-05,
+      "loss": 1.3942,
+      "step": 8070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.851814593826e-05,
+      "loss": 1.4358,
+      "step": 8080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.851449399547326e-05,
+      "loss": 1.1081,
+      "step": 8090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.851083762607751e-05,
+      "loss": 1.4539,
+      "step": 8100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.850717683040638e-05,
+      "loss": 1.2445,
+      "step": 8110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.850351160879388e-05,
+      "loss": 1.2617,
+      "step": 8120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.849984196157444e-05,
+      "loss": 1.2604,
+      "step": 8130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.849616788908289e-05,
+      "loss": 1.3513,
+      "step": 8140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.849248939165447e-05,
+      "loss": 1.3139,
+      "step": 8150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.848880646962483e-05,
+      "loss": 1.2969,
+      "step": 8160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.848511912332998e-05,
+      "loss": 1.3979,
+      "step": 8170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.848142735310639e-05,
+      "loss": 1.0536,
+      "step": 8180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.847773115929088e-05,
+      "loss": 1.5034,
+      "step": 8190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.847403054222074e-05,
+      "loss": 1.3399,
+      "step": 8200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.847032550223359e-05,
+      "loss": 1.3307,
+      "step": 8210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.846661603966751e-05,
+      "loss": 1.3468,
+      "step": 8220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.846290215486095e-05,
+      "loss": 1.1599,
+      "step": 8230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.845918384815281e-05,
+      "loss": 1.1876,
+      "step": 8240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.845546111988231e-05,
+      "loss": 1.3491,
+      "step": 8250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.845173397038914e-05,
+      "loss": 1.2454,
+      "step": 8260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.844800240001339e-05,
+      "loss": 1.356,
+      "step": 8270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.844426640909553e-05,
+      "loss": 1.0493,
+      "step": 8280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.844052599797644e-05,
+      "loss": 1.3075,
+      "step": 8290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.843678116699741e-05,
+      "loss": 1.2018,
+      "step": 8300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.843303191650015e-05,
+      "loss": 1.338,
+      "step": 8310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.842927824682672e-05,
+      "loss": 1.3271,
+      "step": 8320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.842552015831962e-05,
+      "loss": 1.2457,
+      "step": 8330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.842175765132178e-05,
+      "loss": 1.2274,
+      "step": 8340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.841799072617646e-05,
+      "loss": 1.5004,
+      "step": 8350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.841421938322741e-05,
+      "loss": 1.3789,
+      "step": 8360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.84104436228187e-05,
+      "loss": 1.2079,
+      "step": 8370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.840666344529486e-05,
+      "loss": 1.4544,
+      "step": 8380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.840287885100081e-05,
+      "loss": 1.5508,
+      "step": 8390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.839908984028185e-05,
+      "loss": 1.1513,
+      "step": 8400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.839529641348372e-05,
+      "loss": 1.1086,
+      "step": 8410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.839149857095254e-05,
+      "loss": 1.2385,
+      "step": 8420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.838769631303481e-05,
+      "loss": 1.3531,
+      "step": 8430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.83838896400775e-05,
+      "loss": 1.4215,
+      "step": 8440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.838007855242791e-05,
+      "loss": 1.2625,
+      "step": 8450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.837626305043379e-05,
+      "loss": 1.2192,
+      "step": 8460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.83724431344433e-05,
+      "loss": 1.2347,
+      "step": 8470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.836861880480493e-05,
+      "loss": 1.4267,
+      "step": 8480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.836479006186765e-05,
+      "loss": 1.2304,
+      "step": 8490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.836095690598083e-05,
+      "loss": 1.3287,
+      "step": 8500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.835711933749419e-05,
+      "loss": 1.2256,
+      "step": 8510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.835327735675787e-05,
+      "loss": 1.4997,
+      "step": 8520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.834943096412245e-05,
+      "loss": 1.0479,
+      "step": 8530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.834558015993887e-05,
+      "loss": 1.2734,
+      "step": 8540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.834172494455851e-05,
+      "loss": 1.3274,
+      "step": 8550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.83378653183331e-05,
+      "loss": 1.1544,
+      "step": 8560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.833400128161483e-05,
+      "loss": 1.2328,
+      "step": 8570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.833013283475627e-05,
+      "loss": 1.1407,
+      "step": 8580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.832625997811036e-05,
+      "loss": 1.1987,
+      "step": 8590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.83223827120305e-05,
+      "loss": 1.3577,
+      "step": 8600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.831850103687044e-05,
+      "loss": 1.2481,
+      "step": 8610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.831461495298438e-05,
+      "loss": 1.4829,
+      "step": 8620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.831072446072687e-05,
+      "loss": 1.1032,
+      "step": 8630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.830682956045291e-05,
+      "loss": 1.2056,
+      "step": 8640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.830293025251789e-05,
+      "loss": 0.9819,
+      "step": 8650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.829902653727756e-05,
+      "loss": 1.6266,
+      "step": 8660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.829511841508815e-05,
+      "loss": 1.1374,
+      "step": 8670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.829120588630622e-05,
+      "loss": 1.3165,
+      "step": 8680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.828728895128877e-05,
+      "loss": 1.2896,
+      "step": 8690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.828336761039318e-05,
+      "loss": 1.4283,
+      "step": 8700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.827944186397728e-05,
+      "loss": 1.4235,
+      "step": 8710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.827551171239924e-05,
+      "loss": 1.3255,
+      "step": 8720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.827157715601765e-05,
+      "loss": 1.2647,
+      "step": 8730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.826763819519153e-05,
+      "loss": 1.3183,
+      "step": 8740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.826369483028027e-05,
+      "loss": 1.325,
+      "step": 8750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.825974706164369e-05,
+      "loss": 1.6156,
+      "step": 8760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.825579488964198e-05,
+      "loss": 1.0642,
+      "step": 8770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.825183831463576e-05,
+      "loss": 1.1888,
+      "step": 8780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.824787733698604e-05,
+      "loss": 1.4277,
+      "step": 8790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.824391195705423e-05,
+      "loss": 1.218,
+      "step": 8800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.823994217520212e-05,
+      "loss": 1.2397,
+      "step": 8810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.823596799179196e-05,
+      "loss": 1.2788,
+      "step": 8820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.823198940718634e-05,
+      "loss": 1.4237,
+      "step": 8830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.82280064217483e-05,
+      "loss": 1.4129,
+      "step": 8840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.822401903584125e-05,
+      "loss": 1.3467,
+      "step": 8850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.822002724982901e-05,
+      "loss": 1.4142,
+      "step": 8860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.82160310640758e-05,
+      "loss": 1.1949,
+      "step": 8870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.821203047894624e-05,
+      "loss": 1.2524,
+      "step": 8880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.820802549480538e-05,
+      "loss": 1.2568,
+      "step": 8890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.820401611201862e-05,
+      "loss": 1.2205,
+      "step": 8900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.820000233095179e-05,
+      "loss": 1.2848,
+      "step": 8910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.819598415197113e-05,
+      "loss": 1.2488,
+      "step": 8920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.819196157544328e-05,
+      "loss": 1.1677,
+      "step": 8930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.818793460173525e-05,
+      "loss": 1.3149,
+      "step": 8940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.818390323121449e-05,
+      "loss": 1.4495,
+      "step": 8950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.817986746424881e-05,
+      "loss": 1.5506,
+      "step": 8960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.817582730120648e-05,
+      "loss": 1.3322,
+      "step": 8970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.81717827424561e-05,
+      "loss": 1.402,
+      "step": 8980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.816773378836675e-05,
+      "loss": 1.3724,
+      "step": 8990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.816368043930785e-05,
+      "loss": 1.3759,
+      "step": 9000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.815962269564923e-05,
+      "loss": 1.2096,
+      "step": 9010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.815556055776112e-05,
+      "loss": 1.2231,
+      "step": 9020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.81514940260142e-05,
+      "loss": 1.3297,
+      "step": 9030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.814742310077946e-05,
+      "loss": 1.7098,
+      "step": 9040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.814334778242842e-05,
+      "loss": 1.3305,
+      "step": 9050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.813926807133285e-05,
+      "loss": 1.2667,
+      "step": 9060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.813518396786502e-05,
+      "loss": 1.3842,
+      "step": 9070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.81310954723976e-05,
+      "loss": 1.5595,
+      "step": 9080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.812700258530361e-05,
+      "loss": 1.1515,
+      "step": 9090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.81229053069565e-05,
+      "loss": 1.2127,
+      "step": 9100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.811880363773012e-05,
+      "loss": 1.2793,
+      "step": 9110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.811469757799873e-05,
+      "loss": 1.1972,
+      "step": 9120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.811058712813697e-05,
+      "loss": 1.381,
+      "step": 9130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.810647228851988e-05,
+      "loss": 1.5509,
+      "step": 9140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.810235305952293e-05,
+      "loss": 1.317,
+      "step": 9150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.809822944152196e-05,
+      "loss": 1.4677,
+      "step": 9160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.809410143489321e-05,
+      "loss": 1.2051,
+      "step": 9170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.808996904001337e-05,
+      "loss": 1.582,
+      "step": 9180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.808583225725944e-05,
+      "loss": 1.1585,
+      "step": 9190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.80816910870089e-05,
+      "loss": 1.2673,
+      "step": 9200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.807754552963964e-05,
+      "loss": 1.4402,
+      "step": 9210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.807339558552984e-05,
+      "loss": 1.3071,
+      "step": 9220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.806924125505821e-05,
+      "loss": 1.2665,
+      "step": 9230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.806508253860376e-05,
+      "loss": 1.2619,
+      "step": 9240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.806091943654599e-05,
+      "loss": 1.3016,
+      "step": 9250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.805675194926472e-05,
+      "loss": 1.17,
+      "step": 9260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.80525800771402e-05,
+      "loss": 1.2501,
+      "step": 9270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.804840382055311e-05,
+      "loss": 1.6287,
+      "step": 9280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.804422317988451e-05,
+      "loss": 1.1335,
+      "step": 9290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.804003815551582e-05,
+      "loss": 1.2307,
+      "step": 9300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.803584874782891e-05,
+      "loss": 1.1366,
+      "step": 9310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.803165495720605e-05,
+      "loss": 1.3496,
+      "step": 9320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.802745678402986e-05,
+      "loss": 1.3963,
+      "step": 9330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.802325422868343e-05,
+      "loss": 1.2265,
+      "step": 9340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.801904729155018e-05,
+      "loss": 1.2618,
+      "step": 9350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.8014835973014e-05,
+      "loss": 1.2732,
+      "step": 9360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.801062027345912e-05,
+      "loss": 1.0594,
+      "step": 9370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.80064001932702e-05,
+      "loss": 1.2305,
+      "step": 9380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.800217573283229e-05,
+      "loss": 1.149,
+      "step": 9390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.799794689253083e-05,
+      "loss": 1.2976,
+      "step": 9400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.79937136727517e-05,
+      "loss": 1.1004,
+      "step": 9410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.798947607388114e-05,
+      "loss": 1.242,
+      "step": 9420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.79852340963058e-05,
+      "loss": 1.4456,
+      "step": 9430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.798098774041272e-05,
+      "loss": 1.3882,
+      "step": 9440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.797673700658936e-05,
+      "loss": 1.2927,
+      "step": 9450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.797248189522357e-05,
+      "loss": 1.3206,
+      "step": 9460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.79682224067036e-05,
+      "loss": 1.3717,
+      "step": 9470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.79639585414181e-05,
+      "loss": 1.0823,
+      "step": 9480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.79596902997561e-05,
+      "loss": 1.3667,
+      "step": 9490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.79554176821071e-05,
+      "loss": 1.1936,
+      "step": 9500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.795114068886088e-05,
+      "loss": 1.2976,
+      "step": 9510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.794685932040772e-05,
+      "loss": 1.2484,
+      "step": 9520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.794257357713826e-05,
+      "loss": 1.0996,
+      "step": 9530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.793828345944356e-05,
+      "loss": 1.2768,
+      "step": 9540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.793398896771501e-05,
+      "loss": 1.298,
+      "step": 9550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.792969010234452e-05,
+      "loss": 1.4902,
+      "step": 9560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.79253868637243e-05,
+      "loss": 1.2601,
+      "step": 9570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.792107925224698e-05,
+      "loss": 1.4121,
+      "step": 9580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.791676726830563e-05,
+      "loss": 1.2839,
+      "step": 9590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.791245091229366e-05,
+      "loss": 1.0715,
+      "step": 9600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.790813018460492e-05,
+      "loss": 1.4211,
+      "step": 9610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.790380508563363e-05,
+      "loss": 1.4622,
+      "step": 9620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.789947561577445e-05,
+      "loss": 1.2848,
+      "step": 9630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.78951417754224e-05,
+      "loss": 1.2868,
+      "step": 9640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.789080356497292e-05,
+      "loss": 1.1533,
+      "step": 9650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.788646098482183e-05,
+      "loss": 1.4299,
+      "step": 9660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.788211403536536e-05,
+      "loss": 1.2973,
+      "step": 9670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.787776271700015e-05,
+      "loss": 1.1985,
+      "step": 9680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.787340703012323e-05,
+      "loss": 1.5474,
+      "step": 9690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.786904697513202e-05,
+      "loss": 1.2382,
+      "step": 9700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.786468255242434e-05,
+      "loss": 1.2455,
+      "step": 9710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.786031376239842e-05,
+      "loss": 1.2619,
+      "step": 9720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.785594060545286e-05,
+      "loss": 1.2185,
+      "step": 9730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.785156308198673e-05,
+      "loss": 1.3427,
+      "step": 9740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.78471811923994e-05,
+      "loss": 1.2401,
+      "step": 9750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.78427949370907e-05,
+      "loss": 1.4658,
+      "step": 9760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.783840431646085e-05,
+      "loss": 1.2106,
+      "step": 9770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.783400933091047e-05,
+      "loss": 1.3724,
+      "step": 9780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.782960998084055e-05,
+      "loss": 1.2552,
+      "step": 9790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.782520626665253e-05,
+      "loss": 1.3759,
+      "step": 9800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.782079818874819e-05,
+      "loss": 1.2046,
+      "step": 9810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.781638574752976e-05,
+      "loss": 1.3758,
+      "step": 9820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.781196894339981e-05,
+      "loss": 1.4048,
+      "step": 9830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.780754777676139e-05,
+      "loss": 1.1321,
+      "step": 9840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.780312224801786e-05,
+      "loss": 1.1469,
+      "step": 9850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.779869235757305e-05,
+      "loss": 1.4011,
+      "step": 9860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.779425810583113e-05,
+      "loss": 1.2885,
+      "step": 9870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.778981949319673e-05,
+      "loss": 1.5408,
+      "step": 9880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.778537652007478e-05,
+      "loss": 1.1568,
+      "step": 9890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.778092918687075e-05,
+      "loss": 1.2728,
+      "step": 9900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.777647749399035e-05,
+      "loss": 1.2054,
+      "step": 9910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.777202144183981e-05,
+      "loss": 1.3918,
+      "step": 9920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.776756103082572e-05,
+      "loss": 1.2791,
+      "step": 9930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.776309626135504e-05,
+      "loss": 1.4875,
+      "step": 9940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.775862713383516e-05,
+      "loss": 1.1692,
+      "step": 9950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.775415364867383e-05,
+      "loss": 1.3352,
+      "step": 9960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.77496758062793e-05,
+      "loss": 1.5526,
+      "step": 9970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.774519360706005e-05,
+      "loss": 1.2732,
+      "step": 9980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.77407070514251e-05,
+      "loss": 1.285,
+      "step": 9990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.773621613978379e-05,
+      "loss": 1.178,
+      "step": 10000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.773172087254592e-05,
+      "loss": 1.3514,
+      "step": 10010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.772722125012163e-05,
+      "loss": 1.3739,
+      "step": 10020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.772271727292149e-05,
+      "loss": 1.2852,
+      "step": 10030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.771820894135642e-05,
+      "loss": 1.2566,
+      "step": 10040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.771369625583782e-05,
+      "loss": 1.3434,
+      "step": 10050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.770917921677741e-05,
+      "loss": 1.2456,
+      "step": 10060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.770465782458736e-05,
+      "loss": 1.2594,
+      "step": 10070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.77001320796802e-05,
+      "loss": 1.2029,
+      "step": 10080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.76956019824689e-05,
+      "loss": 1.2304,
+      "step": 10090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.769106753336677e-05,
+      "loss": 1.4384,
+      "step": 10100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.768652873278755e-05,
+      "loss": 1.1292,
+      "step": 10110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.768198558114536e-05,
+      "loss": 1.6068,
+      "step": 10120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.767743807885477e-05,
+      "loss": 1.2568,
+      "step": 10130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.767288622633069e-05,
+      "loss": 1.2739,
+      "step": 10140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.766833002398846e-05,
+      "loss": 1.2381,
+      "step": 10150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.766376947224375e-05,
+      "loss": 1.3275,
+      "step": 10160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.765920457151272e-05,
+      "loss": 1.1325,
+      "step": 10170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.765463532221188e-05,
+      "loss": 1.2064,
+      "step": 10180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.765006172475816e-05,
+      "loss": 1.2892,
+      "step": 10190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.764548377956883e-05,
+      "loss": 1.2074,
+      "step": 10200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.764090148706162e-05,
+      "loss": 1.3337,
+      "step": 10210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.763631484765463e-05,
+      "loss": 1.2638,
+      "step": 10220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.763172386176636e-05,
+      "loss": 1.2423,
+      "step": 10230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.76271285298157e-05,
+      "loss": 1.258,
+      "step": 10240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.762252885222194e-05,
+      "loss": 1.1784,
+      "step": 10250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.761792482940477e-05,
+      "loss": 1.3826,
+      "step": 10260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.761331646178427e-05,
+      "loss": 1.2646,
+      "step": 10270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.760870374978093e-05,
+      "loss": 1.4281,
+      "step": 10280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.760408669381564e-05,
+      "loss": 1.1588,
+      "step": 10290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.759946529430966e-05,
+      "loss": 1.2727,
+      "step": 10300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.759483955168464e-05,
+      "loss": 1.3762,
+      "step": 10310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.759020946636268e-05,
+      "loss": 1.3344,
+      "step": 10320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.758557503876625e-05,
+      "loss": 1.4003,
+      "step": 10330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.758093626931816e-05,
+      "loss": 1.3842,
+      "step": 10340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.75762931584417e-05,
+      "loss": 1.3667,
+      "step": 10350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.757164570656052e-05,
+      "loss": 1.1048,
+      "step": 10360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.756699391409867e-05,
+      "loss": 1.2294,
+      "step": 10370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.756233778148059e-05,
+      "loss": 1.3891,
+      "step": 10380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.75576773091311e-05,
+      "loss": 1.1871,
+      "step": 10390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.755301249747546e-05,
+      "loss": 1.51,
+      "step": 10400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.75483433469393e-05,
+      "loss": 1.3794,
+      "step": 10410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.754366985794862e-05,
+      "loss": 1.1809,
+      "step": 10420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.753899203092988e-05,
+      "loss": 1.223,
+      "step": 10430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.753430986630989e-05,
+      "loss": 1.1459,
+      "step": 10440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.752962336451587e-05,
+      "loss": 1.3384,
+      "step": 10450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.75249325259754e-05,
+      "loss": 1.3032,
+      "step": 10460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.75202373511165e-05,
+      "loss": 1.3291,
+      "step": 10470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.75155378403676e-05,
+      "loss": 1.4394,
+      "step": 10480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.751083399415746e-05,
+      "loss": 1.3307,
+      "step": 10490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.750612581291529e-05,
+      "loss": 1.5223,
+      "step": 10500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.750141329707068e-05,
+      "loss": 1.2802,
+      "step": 10510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.74966964470536e-05,
+      "loss": 1.2532,
+      "step": 10520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.749197526329446e-05,
+      "loss": 1.2902,
+      "step": 10530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.748724974622398e-05,
+      "loss": 1.361,
+      "step": 10540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.74825198962734e-05,
+      "loss": 1.2787,
+      "step": 10550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.747778571387425e-05,
+      "loss": 1.3774,
+      "step": 10560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.747304719945847e-05,
+      "loss": 1.2597,
+      "step": 10570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.746830435345846e-05,
+      "loss": 1.4426,
+      "step": 10580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.746355717630695e-05,
+      "loss": 1.411,
+      "step": 10590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.745880566843708e-05,
+      "loss": 1.2959,
+      "step": 10600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.745404983028242e-05,
+      "loss": 1.1221,
+      "step": 10610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.744928966227687e-05,
+      "loss": 1.4815,
+      "step": 10620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.744452516485477e-05,
+      "loss": 1.1181,
+      "step": 10630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.743975633845087e-05,
+      "loss": 1.4959,
+      "step": 10640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.743498318350026e-05,
+      "loss": 1.2657,
+      "step": 10650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.74302057004385e-05,
+      "loss": 1.1393,
+      "step": 10660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.742542388970147e-05,
+      "loss": 1.4397,
+      "step": 10670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.742063775172548e-05,
+      "loss": 1.3163,
+      "step": 10680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.741584728694724e-05,
+      "loss": 1.2981,
+      "step": 10690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.741105249580383e-05,
+      "loss": 1.4278,
+      "step": 10700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.740625337873277e-05,
+      "loss": 1.2684,
+      "step": 10710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.740144993617191e-05,
+      "loss": 1.5667,
+      "step": 10720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.739664216855957e-05,
+      "loss": 1.2765,
+      "step": 10730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.739183007633439e-05,
+      "loss": 1.0858,
+      "step": 10740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.738701365993547e-05,
+      "loss": 1.2874,
+      "step": 10750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.738219291980224e-05,
+      "loss": 1.2453,
+      "step": 10760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.737736785637459e-05,
+      "loss": 1.2517,
+      "step": 10770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.737253847009278e-05,
+      "loss": 1.4115,
+      "step": 10780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.73677047613974e-05,
+      "loss": 1.1685,
+      "step": 10790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.736286673072957e-05,
+      "loss": 1.1497,
+      "step": 10800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.735802437853066e-05,
+      "loss": 1.3992,
+      "step": 10810
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.735317770524254e-05,
+      "loss": 1.5233,
+      "step": 10820
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.734832671130744e-05,
+      "loss": 1.3203,
+      "step": 10830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.734347139716794e-05,
+      "loss": 1.1107,
+      "step": 10840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.733861176326709e-05,
+      "loss": 1.213,
+      "step": 10850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.733374781004829e-05,
+      "loss": 1.2568,
+      "step": 10860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.732887953795531e-05,
+      "loss": 1.3048,
+      "step": 10870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.73240069474324e-05,
+      "loss": 1.2171,
+      "step": 10880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.73191300389241e-05,
+      "loss": 1.5973,
+      "step": 10890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.731424881287542e-05,
+      "loss": 1.3277,
+      "step": 10900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.730936326973176e-05,
+      "loss": 1.2673,
+      "step": 10910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.730447340993884e-05,
+      "loss": 1.1229,
+      "step": 10920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.729957923394287e-05,
+      "loss": 1.1801,
+      "step": 10930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.729468074219038e-05,
+      "loss": 1.0772,
+      "step": 10940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.728977793512833e-05,
+      "loss": 1.3307,
+      "step": 10950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.728487081320408e-05,
+      "loss": 1.3974,
+      "step": 10960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.727995937686536e-05,
+      "loss": 1.1048,
+      "step": 10970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.72750436265603e-05,
+      "loss": 1.3663,
+      "step": 10980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.727012356273743e-05,
+      "loss": 1.5067,
+      "step": 10990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.726519918584569e-05,
+      "loss": 1.3096,
+      "step": 11000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.726027049633437e-05,
+      "loss": 1.3174,
+      "step": 11010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.725533749465319e-05,
+      "loss": 1.2514,
+      "step": 11020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.725040018125225e-05,
+      "loss": 1.1859,
+      "step": 11030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.724545855658205e-05,
+      "loss": 1.4342,
+      "step": 11040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.724051262109347e-05,
+      "loss": 1.283,
+      "step": 11050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.723556237523778e-05,
+      "loss": 1.2706,
+      "step": 11060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.72306078194667e-05,
+      "loss": 1.3571,
+      "step": 11070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.722564895423226e-05,
+      "loss": 1.3535,
+      "step": 11080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.722068577998694e-05,
+      "loss": 1.2717,
+      "step": 11090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.721571829718358e-05,
+      "loss": 1.4024,
+      "step": 11100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.721074650627544e-05,
+      "loss": 1.1179,
+      "step": 11110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.720577040771615e-05,
+      "loss": 1.2735,
+      "step": 11120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.720079000195977e-05,
+      "loss": 1.1875,
+      "step": 11130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.719580528946068e-05,
+      "loss": 1.3639,
+      "step": 11140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.719081627067374e-05,
+      "loss": 1.3208,
+      "step": 11150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.718582294605417e-05,
+      "loss": 1.58,
+      "step": 11160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.718082531605755e-05,
+      "loss": 1.2939,
+      "step": 11170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.717582338113987e-05,
+      "loss": 1.3281,
+      "step": 11180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.717081714175756e-05,
+      "loss": 1.359,
+      "step": 11190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.716580659836737e-05,
+      "loss": 1.2993,
+      "step": 11200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.716079175142649e-05,
+      "loss": 1.3212,
+      "step": 11210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.71557726013925e-05,
+      "loss": 1.2218,
+      "step": 11220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.715074914872335e-05,
+      "loss": 1.3414,
+      "step": 11230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.714572139387739e-05,
+      "loss": 1.229,
+      "step": 11240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.714068933731337e-05,
+      "loss": 1.2429,
+      "step": 11250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.713565297949045e-05,
+      "loss": 1.3841,
+      "step": 11260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.713061232086814e-05,
+      "loss": 1.0936,
+      "step": 11270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.712556736190636e-05,
+      "loss": 1.4631,
+      "step": 11280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.712051810306546e-05,
+      "loss": 1.3352,
+      "step": 11290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.711546454480614e-05,
+      "loss": 1.2983,
+      "step": 11300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.711040668758946e-05,
+      "loss": 1.1473,
+      "step": 11310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.710534453187696e-05,
+      "loss": 1.2945,
+      "step": 11320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.710027807813051e-05,
+      "loss": 1.4571,
+      "step": 11330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.70952073268124e-05,
+      "loss": 1.419,
+      "step": 11340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.709013227838527e-05,
+      "loss": 1.3024,
+      "step": 11350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.708505293331222e-05,
+      "loss": 1.2102,
+      "step": 11360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.707996929205668e-05,
+      "loss": 1.3852,
+      "step": 11370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.707488135508252e-05,
+      "loss": 1.2365,
+      "step": 11380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.706978912285394e-05,
+      "loss": 1.2113,
+      "step": 11390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.706469259583561e-05,
+      "loss": 1.2776,
+      "step": 11400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.705959177449253e-05,
+      "loss": 1.3543,
+      "step": 11410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.705448665929013e-05,
+      "loss": 1.2315,
+      "step": 11420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.704937725069421e-05,
+      "loss": 1.3477,
+      "step": 11430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.704426354917096e-05,
+      "loss": 1.332,
+      "step": 11440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.703914555518697e-05,
+      "loss": 1.0505,
+      "step": 11450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.703402326920922e-05,
+      "loss": 1.2145,
+      "step": 11460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.702889669170512e-05,
+      "loss": 1.3894,
+      "step": 11470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.702376582314238e-05,
+      "loss": 1.3382,
+      "step": 11480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.701863066398917e-05,
+      "loss": 1.2951,
+      "step": 11490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.701349121471407e-05,
+      "loss": 1.0731,
+      "step": 11500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.700834747578598e-05,
+      "loss": 1.3168,
+      "step": 11510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.700319944767427e-05,
+      "loss": 1.1765,
+      "step": 11520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.699804713084862e-05,
+      "loss": 1.2865,
+      "step": 11530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.699289052577916e-05,
+      "loss": 1.261,
+      "step": 11540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.698772963293638e-05,
+      "loss": 1.8619,
+      "step": 11550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.698256445279122e-05,
+      "loss": 1.2947,
+      "step": 11560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.697739498581493e-05,
+      "loss": 1.5425,
+      "step": 11570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.697222123247917e-05,
+      "loss": 1.3848,
+      "step": 11580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.696704319325607e-05,
+      "loss": 1.3001,
+      "step": 11590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.696186086861803e-05,
+      "loss": 1.4272,
+      "step": 11600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.695667425903794e-05,
+      "loss": 1.361,
+      "step": 11610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.6951483364989e-05,
+      "loss": 1.2557,
+      "step": 11620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.69462881869449e-05,
+      "loss": 1.1357,
+      "step": 11630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.694108872537962e-05,
+      "loss": 1.3093,
+      "step": 11640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.693588498076759e-05,
+      "loss": 1.5686,
+      "step": 11650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.693067695358361e-05,
+      "loss": 1.3165,
+      "step": 11660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.692546464430288e-05,
+      "loss": 1.5016,
+      "step": 11670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.692024805340098e-05,
+      "loss": 1.445,
+      "step": 11680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.691502718135392e-05,
+      "loss": 1.1653,
+      "step": 11690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.690980202863802e-05,
+      "loss": 1.0476,
+      "step": 11700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.690457259573006e-05,
+      "loss": 1.1743,
+      "step": 11710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.689933888310719e-05,
+      "loss": 1.432,
+      "step": 11720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.689410089124695e-05,
+      "loss": 1.1104,
+      "step": 11730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.688885862062727e-05,
+      "loss": 1.0394,
+      "step": 11740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.688361207172647e-05,
+      "loss": 1.2931,
+      "step": 11750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.687836124502328e-05,
+      "loss": 1.3693,
+      "step": 11760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.687310614099675e-05,
+      "loss": 1.1884,
+      "step": 11770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.686784676012643e-05,
+      "loss": 1.093,
+      "step": 11780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.686258310289215e-05,
+      "loss": 1.0592,
+      "step": 11790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.685731516977423e-05,
+      "loss": 1.335,
+      "step": 11800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.685204296125332e-05,
+      "loss": 1.6729,
+      "step": 11810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.684676647781043e-05,
+      "loss": 1.1505,
+      "step": 11820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.684148571992706e-05,
+      "loss": 1.4953,
+      "step": 11830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.683620068808501e-05,
+      "loss": 1.0413,
+      "step": 11840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.68309113827665e-05,
+      "loss": 1.4012,
+      "step": 11850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.682561780445416e-05,
+      "loss": 1.315,
+      "step": 11860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.682031995363098e-05,
+      "loss": 1.1989,
+      "step": 11870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.681501783078038e-05,
+      "loss": 1.3461,
+      "step": 11880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.680971143638608e-05,
+      "loss": 1.3148,
+      "step": 11890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.680440077093231e-05,
+      "loss": 1.3559,
+      "step": 11900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.67990858349036e-05,
+      "loss": 1.2588,
+      "step": 11910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.679376662878492e-05,
+      "loss": 1.4767,
+      "step": 11920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.678844315306159e-05,
+      "loss": 1.4732,
+      "step": 11930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.678311540821936e-05,
+      "loss": 1.5456,
+      "step": 11940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.677778339474432e-05,
+      "loss": 1.3482,
+      "step": 11950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.677244711312303e-05,
+      "loss": 1.093,
+      "step": 11960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.676710656384234e-05,
+      "loss": 1.2155,
+      "step": 11970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.676176174738956e-05,
+      "loss": 1.2428,
+      "step": 11980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.675641266425237e-05,
+      "loss": 1.2648,
+      "step": 11990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.675105931491882e-05,
+      "loss": 1.0336,
+      "step": 12000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.674570169987737e-05,
+      "loss": 1.3075,
+      "step": 12010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.674033981961688e-05,
+      "loss": 1.1876,
+      "step": 12020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.673497367462657e-05,
+      "loss": 1.4412,
+      "step": 12030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.672960326539607e-05,
+      "loss": 1.404,
+      "step": 12040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.672422859241539e-05,
+      "loss": 1.3015,
+      "step": 12050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.671884965617493e-05,
+      "loss": 1.4785,
+      "step": 12060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.671346645716549e-05,
+      "loss": 1.3855,
+      "step": 12070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.670807899587824e-05,
+      "loss": 1.328,
+      "step": 12080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.670268727280473e-05,
+      "loss": 1.3459,
+      "step": 12090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.669729128843696e-05,
+      "loss": 1.1278,
+      "step": 12100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.669189104326723e-05,
+      "loss": 1.2781,
+      "step": 12110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.668648653778832e-05,
+      "loss": 1.3047,
+      "step": 12120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.668107777249332e-05,
+      "loss": 1.5821,
+      "step": 12130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.667566474787576e-05,
+      "loss": 1.3916,
+      "step": 12140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.667024746442952e-05,
+      "loss": 1.1922,
+      "step": 12150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.666482592264891e-05,
+      "loss": 1.5275,
+      "step": 12160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.66594001230286e-05,
+      "loss": 1.4539,
+      "step": 12170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.665397006606368e-05,
+      "loss": 1.2838,
+      "step": 12180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.664853575224954e-05,
+      "loss": 1.405,
+      "step": 12190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.664309718208209e-05,
+      "loss": 1.4633,
+      "step": 12200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.663765435605753e-05,
+      "loss": 1.1995,
+      "step": 12210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.663220727467247e-05,
+      "loss": 1.4539,
+      "step": 12220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.662675593842396e-05,
+      "loss": 1.123,
+      "step": 12230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.662130034780936e-05,
+      "loss": 1.1991,
+      "step": 12240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.661584050332647e-05,
+      "loss": 1.1451,
+      "step": 12250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.661037640547345e-05,
+      "loss": 1.2448,
+      "step": 12260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.660490805474889e-05,
+      "loss": 1.2303,
+      "step": 12270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.65994354516517e-05,
+      "loss": 1.3741,
+      "step": 12280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.659395859668123e-05,
+      "loss": 1.2424,
+      "step": 12290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.658847749033722e-05,
+      "loss": 1.4204,
+      "step": 12300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.658299213311976e-05,
+      "loss": 1.2381,
+      "step": 12310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.657750252552939e-05,
+      "loss": 1.3357,
+      "step": 12320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.657200866806695e-05,
+      "loss": 1.4023,
+      "step": 12330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.656651056123374e-05,
+      "loss": 1.2667,
+      "step": 12340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.656100820553143e-05,
+      "loss": 1.259,
+      "step": 12350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.655550160146204e-05,
+      "loss": 1.1893,
+      "step": 12360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.654999074952805e-05,
+      "loss": 1.2523,
+      "step": 12370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.654447565023226e-05,
+      "loss": 1.4266,
+      "step": 12380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.65389563040779e-05,
+      "loss": 1.4251,
+      "step": 12390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.653343271156857e-05,
+      "loss": 1.0538,
+      "step": 12400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.652790487320824e-05,
+      "loss": 1.1824,
+      "step": 12410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.65223727895013e-05,
+      "loss": 1.4353,
+      "step": 12420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.651683646095253e-05,
+      "loss": 1.171,
+      "step": 12430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.651129588806706e-05,
+      "loss": 1.2732,
+      "step": 12440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.650575107135043e-05,
+      "loss": 1.2488,
+      "step": 12450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.650020201130857e-05,
+      "loss": 1.3298,
+      "step": 12460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.64946487084478e-05,
+      "loss": 1.3317,
+      "step": 12470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.648909116327481e-05,
+      "loss": 1.4012,
+      "step": 12480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.64835293762967e-05,
+      "loss": 1.4585,
+      "step": 12490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.647796334802092e-05,
+      "loss": 1.3453,
+      "step": 12500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.647239307895535e-05,
+      "loss": 1.2789,
+      "step": 12510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.646681856960824e-05,
+      "loss": 1.1763,
+      "step": 12520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.646123982048823e-05,
+      "loss": 1.3391,
+      "step": 12530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.645565683210432e-05,
+      "loss": 1.3108,
+      "step": 12540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.645006960496592e-05,
+      "loss": 1.2904,
+      "step": 12550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.644447813958285e-05,
+      "loss": 1.197,
+      "step": 12560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.643888243646526e-05,
+      "loss": 1.1268,
+      "step": 12570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.643328249612376e-05,
+      "loss": 1.4279,
+      "step": 12580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.642767831906927e-05,
+      "loss": 1.427,
+      "step": 12590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.642206990581315e-05,
+      "loss": 1.3183,
+      "step": 12600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.641645725686712e-05,
+      "loss": 1.3977,
+      "step": 12610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.641084037274328e-05,
+      "loss": 1.1507,
+      "step": 12620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.640521925395418e-05,
+      "loss": 1.2651,
+      "step": 12630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.639959390101266e-05,
+      "loss": 1.1262,
+      "step": 12640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.639396431443201e-05,
+      "loss": 1.2044,
+      "step": 12650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.63883304947259e-05,
+      "loss": 1.3134,
+      "step": 12660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.638269244240837e-05,
+      "loss": 1.3075,
+      "step": 12670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.637705015799385e-05,
+      "loss": 1.1927,
+      "step": 12680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.637140364199716e-05,
+      "loss": 1.1793,
+      "step": 12690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.636575289493351e-05,
+      "loss": 1.3144,
+      "step": 12700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.636009791731848e-05,
+      "loss": 1.3483,
+      "step": 12710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.635443870966805e-05,
+      "loss": 1.5586,
+      "step": 12720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.634877527249859e-05,
+      "loss": 1.4635,
+      "step": 12730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.634310760632685e-05,
+      "loss": 1.1159,
+      "step": 12740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.633743571166997e-05,
+      "loss": 1.1738,
+      "step": 12750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.633175958904545e-05,
+      "loss": 1.2917,
+      "step": 12760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.632607923897122e-05,
+      "loss": 1.2186,
+      "step": 12770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.632039466196557e-05,
+      "loss": 1.2737,
+      "step": 12780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.631470585854714e-05,
+      "loss": 1.4299,
+      "step": 12790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.630901282923503e-05,
+      "loss": 1.3434,
+      "step": 12800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.63033155745487e-05,
+      "loss": 1.3396,
+      "step": 12810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.629761409500795e-05,
+      "loss": 1.3017,
+      "step": 12820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.629190839113303e-05,
+      "loss": 1.4041,
+      "step": 12830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.628619846344454e-05,
+      "loss": 1.5323,
+      "step": 12840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.628048431246343e-05,
+      "loss": 1.2393,
+      "step": 12850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.627476593871114e-05,
+      "loss": 1.3642,
+      "step": 12860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.626904334270938e-05,
+      "loss": 1.2699,
+      "step": 12870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.626331652498034e-05,
+      "loss": 1.1997,
+      "step": 12880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.625758548604652e-05,
+      "loss": 1.5095,
+      "step": 12890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.625185022643086e-05,
+      "loss": 1.4039,
+      "step": 12900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.624611074665664e-05,
+      "loss": 1.2304,
+      "step": 12910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.624036704724755e-05,
+      "loss": 1.3279,
+      "step": 12920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.623461912872768e-05,
+      "loss": 1.1611,
+      "step": 12930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.622886699162148e-05,
+      "loss": 1.2507,
+      "step": 12940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.622311063645379e-05,
+      "loss": 1.2207,
+      "step": 12950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.621735006374983e-05,
+      "loss": 1.3434,
+      "step": 12960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.621158527403524e-05,
+      "loss": 1.1867,
+      "step": 12970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.620581626783598e-05,
+      "loss": 1.2722,
+      "step": 12980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.620004304567844e-05,
+      "loss": 1.1997,
+      "step": 12990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.619426560808942e-05,
+      "loss": 1.2303,
+      "step": 13000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.618848395559604e-05,
+      "loss": 1.2767,
+      "step": 13010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.618269808872582e-05,
+      "loss": 1.2423,
+      "step": 13020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.617690800800671e-05,
+      "loss": 1.2134,
+      "step": 13030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.6171113713967e-05,
+      "loss": 1.0973,
+      "step": 13040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.616531520713542e-05,
+      "loss": 1.2826,
+      "step": 13050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.615951248804096e-05,
+      "loss": 1.1461,
+      "step": 13060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.615370555721314e-05,
+      "loss": 1.0808,
+      "step": 13070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.61478944151818e-05,
+      "loss": 1.2635,
+      "step": 13080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.614207906247715e-05,
+      "loss": 1.0739,
+      "step": 13090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.61362594996298e-05,
+      "loss": 1.3752,
+      "step": 13100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.613043572717074e-05,
+      "loss": 1.2528,
+      "step": 13110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.612460774563136e-05,
+      "loss": 1.4285,
+      "step": 13120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.611877555554341e-05,
+      "loss": 1.2699,
+      "step": 13130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.611293915743906e-05,
+      "loss": 1.3874,
+      "step": 13140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.610709855185082e-05,
+      "loss": 1.1458,
+      "step": 13150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.610125373931162e-05,
+      "loss": 1.2177,
+      "step": 13160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.609540472035474e-05,
+      "loss": 1.2575,
+      "step": 13170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.608955149551388e-05,
+      "loss": 1.3255,
+      "step": 13180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.60836940653231e-05,
+      "loss": 1.3443,
+      "step": 13190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.607783243031682e-05,
+      "loss": 1.4566,
+      "step": 13200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.607196659102992e-05,
+      "loss": 1.3381,
+      "step": 13210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.606609654799761e-05,
+      "loss": 1.2155,
+      "step": 13220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.606022230175547e-05,
+      "loss": 1.1797,
+      "step": 13230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.605434385283947e-05,
+      "loss": 1.0969,
+      "step": 13240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.604846120178602e-05,
+      "loss": 1.2215,
+      "step": 13250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.604257434913184e-05,
+      "loss": 1.1516,
+      "step": 13260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.603668329541408e-05,
+      "loss": 1.4677,
+      "step": 13270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.603078804117026e-05,
+      "loss": 1.2844,
+      "step": 13280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.602488858693826e-05,
+      "loss": 1.1449,
+      "step": 13290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.601898493325636e-05,
+      "loss": 1.2592,
+      "step": 13300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.601307708066327e-05,
+      "loss": 1.3818,
+      "step": 13310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.6007165029698e-05,
+      "loss": 1.3662,
+      "step": 13320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.60012487809e-05,
+      "loss": 1.2059,
+      "step": 13330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.59953283348091e-05,
+      "loss": 1.4641,
+      "step": 13340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.598940369196547e-05,
+      "loss": 1.2971,
+      "step": 13350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.598347485290971e-05,
+      "loss": 1.2398,
+      "step": 13360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.597754181818277e-05,
+      "loss": 1.4895,
+      "step": 13370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.597160458832601e-05,
+      "loss": 1.2297,
+      "step": 13380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.596566316388116e-05,
+      "loss": 1.0977,
+      "step": 13390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.595971754539035e-05,
+      "loss": 1.2662,
+      "step": 13400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.595376773339604e-05,
+      "loss": 1.2493,
+      "step": 13410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.594781372844112e-05,
+      "loss": 1.3357,
+      "step": 13420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.594185553106889e-05,
+      "loss": 1.3568,
+      "step": 13430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.593589314182295e-05,
+      "loss": 1.2074,
+      "step": 13440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.592992656124733e-05,
+      "loss": 1.3435,
+      "step": 13450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.592395578988646e-05,
+      "loss": 1.2205,
+      "step": 13460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.591798082828511e-05,
+      "loss": 1.3639,
+      "step": 13470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.591200167698846e-05,
+      "loss": 1.2047,
+      "step": 13480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.590601833654208e-05,
+      "loss": 1.1246,
+      "step": 13490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.59000308074919e-05,
+      "loss": 1.2006,
+      "step": 13500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.589403909038424e-05,
+      "loss": 1.2141,
+      "step": 13510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.588804318576578e-05,
+      "loss": 1.2047,
+      "step": 13520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.588204309418363e-05,
+      "loss": 1.3082,
+      "step": 13530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.587603881618527e-05,
+      "loss": 1.0781,
+      "step": 13540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.587003035231852e-05,
+      "loss": 1.2985,
+      "step": 13550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.58640177031316e-05,
+      "loss": 1.3457,
+      "step": 13560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.585800086917316e-05,
+      "loss": 1.0767,
+      "step": 13570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.585197985099217e-05,
+      "loss": 1.4532,
+      "step": 13580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.584595464913801e-05,
+      "loss": 1.367,
+      "step": 13590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.583992526416046e-05,
+      "loss": 1.2669,
+      "step": 13600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.58338916966096e-05,
+      "loss": 1.3832,
+      "step": 13610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.582785394703604e-05,
+      "loss": 1.378,
+      "step": 13620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.58218120159906e-05,
+      "loss": 1.3135,
+      "step": 13630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.58157659040246e-05,
+      "loss": 1.2424,
+      "step": 13640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.58097156116897e-05,
+      "loss": 1.2324,
+      "step": 13650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.580366113953796e-05,
+      "loss": 1.3732,
+      "step": 13660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.579760248812179e-05,
+      "loss": 1.0931,
+      "step": 13670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.5791539657994e-05,
+      "loss": 1.4051,
+      "step": 13680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.578547264970779e-05,
+      "loss": 1.461,
+      "step": 13690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.577940146381675e-05,
+      "loss": 1.3012,
+      "step": 13700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.577332610087478e-05,
+      "loss": 1.2116,
+      "step": 13710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.576724656143627e-05,
+      "loss": 1.3947,
+      "step": 13720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.57611628460559e-05,
+      "loss": 1.1196,
+      "step": 13730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.57550749552888e-05,
+      "loss": 1.4405,
+      "step": 13740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.57489828896904e-05,
+      "loss": 1.3217,
+      "step": 13750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.574288664981662e-05,
+      "loss": 1.2536,
+      "step": 13760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.573678623622363e-05,
+      "loss": 1.2568,
+      "step": 13770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.57306816494681e-05,
+      "loss": 1.0089,
+      "step": 13780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.572457289010702e-05,
+      "loss": 1.4428,
+      "step": 13790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.571845995869776e-05,
+      "loss": 1.3519,
+      "step": 13800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.571234285579811e-05,
+      "loss": 1.2439,
+      "step": 13810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.570622158196618e-05,
+      "loss": 1.3892,
+      "step": 13820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.57000961377605e-05,
+      "loss": 1.1095,
+      "step": 13830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.569396652373998e-05,
+      "loss": 1.1698,
+      "step": 13840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.568783274046391e-05,
+      "loss": 1.5212,
+      "step": 13850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.568169478849197e-05,
+      "loss": 1.2814,
+      "step": 13860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.567555266838416e-05,
+      "loss": 1.2799,
+      "step": 13870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.566940638070094e-05,
+      "loss": 1.4536,
+      "step": 13880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.56632559260031e-05,
+      "loss": 1.3424,
+      "step": 13890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.565710130485183e-05,
+      "loss": 1.1792,
+      "step": 13900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.565094251780871e-05,
+      "loss": 1.402,
+      "step": 13910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.564477956543568e-05,
+      "loss": 1.5138,
+      "step": 13920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.563861244829504e-05,
+      "loss": 1.2251,
+      "step": 13930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.563244116694954e-05,
+      "loss": 1.2607,
+      "step": 13940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.562626572196224e-05,
+      "loss": 1.2343,
+      "step": 13950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.56200861138966e-05,
+      "loss": 1.2829,
+      "step": 13960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.561390234331649e-05,
+      "loss": 1.2447,
+      "step": 13970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.560771441078613e-05,
+      "loss": 1.458,
+      "step": 13980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.56015223168701e-05,
+      "loss": 1.2906,
+      "step": 13990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.559532606213341e-05,
+      "loss": 1.1875,
+      "step": 14000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.558912564714142e-05,
+      "loss": 1.2395,
+      "step": 14010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.558292107245986e-05,
+      "loss": 1.3027,
+      "step": 14020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.557671233865488e-05,
+      "loss": 1.0481,
+      "step": 14030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.557049944629297e-05,
+      "loss": 1.2331,
+      "step": 14040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.556428239594103e-05,
+      "loss": 1.3488,
+      "step": 14050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.555806118816629e-05,
+      "loss": 1.296,
+      "step": 14060
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.555183582353642e-05,
+      "loss": 1.0644,
+      "step": 14070
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.554560630261942e-05,
+      "loss": 1.3796,
+      "step": 14080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.553937262598369e-05,
+      "loss": 1.4246,
+      "step": 14090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.553313479419803e-05,
+      "loss": 1.2945,
+      "step": 14100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.55268928078316e-05,
+      "loss": 1.3542,
+      "step": 14110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.55206466674539e-05,
+      "loss": 1.4731,
+      "step": 14120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.55143963736349e-05,
+      "loss": 1.4023,
+      "step": 14130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.550814192694485e-05,
+      "loss": 1.3068,
+      "step": 14140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.550188332795446e-05,
+      "loss": 1.2309,
+      "step": 14150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.549562057723476e-05,
+      "loss": 1.1666,
+      "step": 14160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.548935367535717e-05,
+      "loss": 1.1862,
+      "step": 14170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.548308262289353e-05,
+      "loss": 1.2942,
+      "step": 14180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.547680742041603e-05,
+      "loss": 1.3143,
+      "step": 14190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.547052806849722e-05,
+      "loss": 1.1719,
+      "step": 14200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.546424456771006e-05,
+      "loss": 1.178,
+      "step": 14210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.545795691862786e-05,
+      "loss": 1.7178,
+      "step": 14220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.545166512182435e-05,
+      "loss": 1.3496,
+      "step": 14230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.544536917787359e-05,
+      "loss": 1.2592,
+      "step": 14240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.543906908735004e-05,
+      "loss": 1.5064,
+      "step": 14250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.543276485082855e-05,
+      "loss": 1.4023,
+      "step": 14260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.542645646888434e-05,
+      "loss": 1.2379,
+      "step": 14270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.5420143942093e-05,
+      "loss": 1.2176,
+      "step": 14280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.54138272710305e-05,
+      "loss": 1.2369,
+      "step": 14290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.54075064562732e-05,
+      "loss": 1.3587,
+      "step": 14300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.540118149839783e-05,
+      "loss": 1.2812,
+      "step": 14310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.539485239798149e-05,
+      "loss": 1.1354,
+      "step": 14320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.538851915560169e-05,
+      "loss": 1.1987,
+      "step": 14330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.538218177183626e-05,
+      "loss": 1.3057,
+      "step": 14340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.537584024726345e-05,
+      "loss": 1.3382,
+      "step": 14350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.53694945824619e-05,
+      "loss": 1.4262,
+      "step": 14360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.53631447780106e-05,
+      "loss": 1.184,
+      "step": 14370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.535679083448892e-05,
+      "loss": 1.2905,
+      "step": 14380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.53504327524766e-05,
+      "loss": 1.3353,
+      "step": 14390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.534407053255378e-05,
+      "loss": 1.412,
+      "step": 14400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.5337704175301e-05,
+      "loss": 1.3826,
+      "step": 14410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.533133368129909e-05,
+      "loss": 1.2468,
+      "step": 14420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.532495905112938e-05,
+      "loss": 1.1994,
+      "step": 14430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.531858028537342e-05,
+      "loss": 1.3416,
+      "step": 14440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.53121973846133e-05,
+      "loss": 1.2393,
+      "step": 14450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.53058103494314e-05,
+      "loss": 1.1162,
+      "step": 14460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.52994191804105e-05,
+      "loss": 1.1246,
+      "step": 14470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.529302387813372e-05,
+      "loss": 1.259,
+      "step": 14480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.528662444318461e-05,
+      "loss": 1.4229,
+      "step": 14490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.528022087614708e-05,
+      "loss": 1.3106,
+      "step": 14500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.527381317760538e-05,
+      "loss": 1.3853,
+      "step": 14510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.526740134814418e-05,
+      "loss": 1.3745,
+      "step": 14520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.526098538834853e-05,
+      "loss": 1.3033,
+      "step": 14530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.525456529880383e-05,
+      "loss": 1.2273,
+      "step": 14540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.524814108009588e-05,
+      "loss": 1.3499,
+      "step": 14550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.524171273281083e-05,
+      "loss": 1.2546,
+      "step": 14560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.523528025753523e-05,
+      "loss": 1.2898,
+      "step": 14570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.522884365485598e-05,
+      "loss": 1.3143,
+      "step": 14580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.522240292536041e-05,
+      "loss": 1.304,
+      "step": 14590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.521595806963617e-05,
+      "loss": 1.3761,
+      "step": 14600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.520950908827131e-05,
+      "loss": 1.3788,
+      "step": 14610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.520305598185423e-05,
+      "loss": 1.3502,
+      "step": 14620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.519659875097378e-05,
+      "loss": 1.1854,
+      "step": 14630
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.519013739621912e-05,
+      "loss": 1.3097,
+      "step": 14640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.518367191817979e-05,
+      "loss": 1.2704,
+      "step": 14650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.517720231744573e-05,
+      "loss": 1.3759,
+      "step": 14660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.517072859460723e-05,
+      "loss": 1.2764,
+      "step": 14670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.516425075025499e-05,
+      "loss": 1.2031,
+      "step": 14680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.515776878498006e-05,
+      "loss": 1.2731,
+      "step": 14690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.515128269937389e-05,
+      "loss": 1.0977,
+      "step": 14700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.514479249402827e-05,
+      "loss": 1.5174,
+      "step": 14710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.513829816953539e-05,
+      "loss": 1.184,
+      "step": 14720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.513179972648781e-05,
+      "loss": 1.2613,
+      "step": 14730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.512529716547848e-05,
+      "loss": 1.2177,
+      "step": 14740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.511879048710071e-05,
+      "loss": 1.4718,
+      "step": 14750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.511227969194818e-05,
+      "loss": 1.268,
+      "step": 14760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.510576478061495e-05,
+      "loss": 1.3361,
+      "step": 14770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.509924575369548e-05,
+      "loss": 1.1751,
+      "step": 14780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.509272261178457e-05,
+      "loss": 1.5077,
+      "step": 14790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.508619535547743e-05,
+      "loss": 1.3486,
+      "step": 14800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.507966398536959e-05,
+      "loss": 1.4725,
+      "step": 14810
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.507312850205703e-05,
+      "loss": 1.3602,
+      "step": 14820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.506658890613604e-05,
+      "loss": 1.4216,
+      "step": 14830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.506004519820334e-05,
+      "loss": 1.54,
+      "step": 14840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.505349737885597e-05,
+      "loss": 1.0727,
+      "step": 14850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.50469454486914e-05,
+      "loss": 0.9894,
+      "step": 14860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.504038940830742e-05,
+      "loss": 1.3033,
+      "step": 14870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.503382925830223e-05,
+      "loss": 1.4206,
+      "step": 14880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.50272649992744e-05,
+      "loss": 1.2722,
+      "step": 14890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.502069663182289e-05,
+      "loss": 1.617,
+      "step": 14900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.501412415654702e-05,
+      "loss": 1.2825,
+      "step": 14910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.500754757404644e-05,
+      "loss": 0.9748,
+      "step": 14920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.500096688492127e-05,
+      "loss": 1.2113,
+      "step": 14930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.49943820897719e-05,
+      "loss": 1.3671,
+      "step": 14940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.498779318919917e-05,
+      "loss": 1.2903,
+      "step": 14950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.49812001838043e-05,
+      "loss": 1.0919,
+      "step": 14960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.49746030741888e-05,
+      "loss": 1.3182,
+      "step": 14970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.496800186095466e-05,
+      "loss": 1.2445,
+      "step": 14980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.496139654470418e-05,
+      "loss": 1.3082,
+      "step": 14990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.495478712604002e-05,
+      "loss": 1.4041,
+      "step": 15000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.494817360556528e-05,
+      "loss": 1.2026,
+      "step": 15010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.494155598388339e-05,
+      "loss": 1.2226,
+      "step": 15020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.493493426159813e-05,
+      "loss": 1.1804,
+      "step": 15030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.492830843931375e-05,
+      "loss": 1.2866,
+      "step": 15040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.492167851763475e-05,
+      "loss": 1.1852,
+      "step": 15050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.491504449716608e-05,
+      "loss": 1.2098,
+      "step": 15060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.490840637851307e-05,
+      "loss": 1.4086,
+      "step": 15070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.490176416228138e-05,
+      "loss": 1.2262,
+      "step": 15080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.489511784907707e-05,
+      "loss": 1.2831,
+      "step": 15090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.488846743950657e-05,
+      "loss": 1.2364,
+      "step": 15100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.48818129341767e-05,
+      "loss": 1.2609,
+      "step": 15110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.48751543336946e-05,
+      "loss": 1.3107,
+      "step": 15120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.486849163866786e-05,
+      "loss": 1.4669,
+      "step": 15130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.486182484970438e-05,
+      "loss": 1.2959,
+      "step": 15140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.485515396741247e-05,
+      "loss": 1.3684,
+      "step": 15150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.48484789924008e-05,
+      "loss": 1.3405,
+      "step": 15160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.48417999252784e-05,
+      "loss": 1.2239,
+      "step": 15170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.48351167666547e-05,
+      "loss": 1.1562,
+      "step": 15180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.48284295171395e-05,
+      "loss": 1.1647,
+      "step": 15190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.482173817734296e-05,
+      "loss": 1.5485,
+      "step": 15200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.48150427478756e-05,
+      "loss": 1.2677,
+      "step": 15210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.480834322934838e-05,
+      "loss": 1.0979,
+      "step": 15220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.480163962237251e-05,
+      "loss": 1.2109,
+      "step": 15230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.47949319275597e-05,
+      "loss": 1.3806,
+      "step": 15240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.478822014552197e-05,
+      "loss": 1.1895,
+      "step": 15250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.478150427687174e-05,
+      "loss": 1.1608,
+      "step": 15260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.477478432222174e-05,
+      "loss": 1.3987,
+      "step": 15270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.476806028218515e-05,
+      "loss": 1.3672,
+      "step": 15280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.476133215737551e-05,
+      "loss": 1.2423,
+      "step": 15290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.475459994840668e-05,
+      "loss": 1.3597,
+      "step": 15300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.474786365589292e-05,
+      "loss": 1.3544,
+      "step": 15310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.474112328044891e-05,
+      "loss": 1.1315,
+      "step": 15320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.473437882268964e-05,
+      "loss": 1.5525,
+      "step": 15330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.472763028323049e-05,
+      "loss": 1.3624,
+      "step": 15340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.472087766268722e-05,
+      "loss": 1.0602,
+      "step": 15350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.471412096167598e-05,
+      "loss": 1.2712,
+      "step": 15360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.470736018081325e-05,
+      "loss": 1.3102,
+      "step": 15370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.470059532071589e-05,
+      "loss": 1.3245,
+      "step": 15380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.469382638200118e-05,
+      "loss": 1.4467,
+      "step": 15390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.468705336528673e-05,
+      "loss": 1.2914,
+      "step": 15400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.468027627119052e-05,
+      "loss": 1.3871,
+      "step": 15410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.46734951003309e-05,
+      "loss": 1.4056,
+      "step": 15420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.466670985332664e-05,
+      "loss": 1.3288,
+      "step": 15430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.465992053079681e-05,
+      "loss": 1.384,
+      "step": 15440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.465312713336092e-05,
+      "loss": 1.343,
+      "step": 15450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.46463296616388e-05,
+      "loss": 1.2668,
+      "step": 15460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.463952811625068e-05,
+      "loss": 1.4113,
+      "step": 15470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.463272249781715e-05,
+      "loss": 1.255,
+      "step": 15480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.462591280695917e-05,
+      "loss": 1.4448,
+      "step": 15490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.461909904429808e-05,
+      "loss": 1.1872,
+      "step": 15500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.46122812104556e-05,
+      "loss": 1.1688,
+      "step": 15510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.460545930605382e-05,
+      "loss": 1.3372,
+      "step": 15520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.459863333171515e-05,
+      "loss": 1.1087,
+      "step": 15530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.459180328806244e-05,
+      "loss": 1.392,
+      "step": 15540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.458496917571889e-05,
+      "loss": 1.2342,
+      "step": 15550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.457813099530804e-05,
+      "loss": 1.3217,
+      "step": 15560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.457128874745382e-05,
+      "loss": 1.1034,
+      "step": 15570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.456444243278058e-05,
+      "loss": 1.2555,
+      "step": 15580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.455759205191298e-05,
+      "loss": 1.2213,
+      "step": 15590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.455073760547605e-05,
+      "loss": 1.3855,
+      "step": 15600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.454387909409524e-05,
+      "loss": 1.0471,
+      "step": 15610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.453701651839633e-05,
+      "loss": 1.1452,
+      "step": 15620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.453014987900547e-05,
+      "loss": 1.4643,
+      "step": 15630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.452327917654919e-05,
+      "loss": 1.3044,
+      "step": 15640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.451640441165443e-05,
+      "loss": 1.4161,
+      "step": 15650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.450952558494843e-05,
+      "loss": 1.397,
+      "step": 15660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.450264269705885e-05,
+      "loss": 1.2303,
+      "step": 15670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.44957557486137e-05,
+      "loss": 1.3004,
+      "step": 15680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.448886474024135e-05,
+      "loss": 1.2444,
+      "step": 15690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.44819696725706e-05,
+      "loss": 1.2657,
+      "step": 15700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.447507054623051e-05,
+      "loss": 1.1836,
+      "step": 15710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.446816736185065e-05,
+      "loss": 1.3029,
+      "step": 15720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.446126012006084e-05,
+      "loss": 1.3438,
+      "step": 15730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.445434882149133e-05,
+      "loss": 1.1912,
+      "step": 15740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.444743346677274e-05,
+      "loss": 1.1669,
+      "step": 15750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.444051405653604e-05,
+      "loss": 1.2661,
+      "step": 15760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.443359059141256e-05,
+      "loss": 1.3515,
+      "step": 15770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.442666307203405e-05,
+      "loss": 1.2333,
+      "step": 15780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.441973149903256e-05,
+      "loss": 1.3061,
+      "step": 15790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.441279587304058e-05,
+      "loss": 1.2485,
+      "step": 15800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.440585619469094e-05,
+      "loss": 1.2555,
+      "step": 15810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.43989124646168e-05,
+      "loss": 1.2486,
+      "step": 15820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.439196468345176e-05,
+      "loss": 1.4019,
+      "step": 15830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.438501285182975e-05,
+      "loss": 1.4258,
+      "step": 15840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.437805697038507e-05,
+      "loss": 1.213,
+      "step": 15850
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.43710970397524e-05,
+      "loss": 1.3611,
+      "step": 15860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.436413306056678e-05,
+      "loss": 1.2525,
+      "step": 15870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.435716503346362e-05,
+      "loss": 1.2564,
+      "step": 15880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.435019295907872e-05,
+      "loss": 1.252,
+      "step": 15890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.434321683804824e-05,
+      "loss": 1.5885,
+      "step": 15900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.433623667100866e-05,
+      "loss": 1.1777,
+      "step": 15910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.432925245859691e-05,
+      "loss": 1.0543,
+      "step": 15920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.432226420145025e-05,
+      "loss": 1.3367,
+      "step": 15930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.431527190020628e-05,
+      "loss": 1.2429,
+      "step": 15940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.430827555550304e-05,
+      "loss": 1.2294,
+      "step": 15950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.430127516797886e-05,
+      "loss": 1.3431,
+      "step": 15960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.42942707382725e-05,
+      "loss": 1.3269,
+      "step": 15970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.428726226702304e-05,
+      "loss": 1.2602,
+      "step": 15980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.428024975487e-05,
+      "loss": 1.3442,
+      "step": 15990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.427323320245318e-05,
+      "loss": 1.2503,
+      "step": 16000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.426621261041279e-05,
+      "loss": 1.3511,
+      "step": 16010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.425918797938947e-05,
+      "loss": 1.1665,
+      "step": 16020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.42521593100241e-05,
+      "loss": 1.122,
+      "step": 16030
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.4245126602958e-05,
+      "loss": 1.2693,
+      "step": 16040
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.423808985883289e-05,
+      "loss": 1.3612,
+      "step": 16050
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.423104907829082e-05,
+      "loss": 1.3019,
+      "step": 16060
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.42240042619742e-05,
+      "loss": 1.2934,
+      "step": 16070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.421695541052582e-05,
+      "loss": 1.2136,
+      "step": 16080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.420990252458883e-05,
+      "loss": 1.2657,
+      "step": 16090
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.420284560480679e-05,
+      "loss": 1.26,
+      "step": 16100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.419578465182357e-05,
+      "loss": 1.4128,
+      "step": 16110
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.418871966628342e-05,
+      "loss": 1.2646,
+      "step": 16120
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.4181650648831e-05,
+      "loss": 1.2914,
+      "step": 16130
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.417457760011128e-05,
+      "loss": 1.0937,
+      "step": 16140
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.416750052076966e-05,
+      "loss": 1.2951,
+      "step": 16150
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.416041941145184e-05,
+      "loss": 1.2279,
+      "step": 16160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.415333427280394e-05,
+      "loss": 1.3275,
+      "step": 16170
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.414624510547243e-05,
+      "loss": 1.1433,
+      "step": 16180
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.413915191010414e-05,
+      "loss": 1.4721,
+      "step": 16190
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.413205468734627e-05,
+      "loss": 1.1668,
+      "step": 16200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.412495343784642e-05,
+      "loss": 1.3639,
+      "step": 16210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.41178481622525e-05,
+      "loss": 1.2019,
+      "step": 16220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.411073886121282e-05,
+      "loss": 1.2207,
+      "step": 16230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.410362553537606e-05,
+      "loss": 1.1989,
+      "step": 16240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.409650818539127e-05,
+      "loss": 1.2435,
+      "step": 16250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.408938681190785e-05,
+      "loss": 1.2853,
+      "step": 16260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.408226141557557e-05,
+      "loss": 1.3813,
+      "step": 16270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.407513199704458e-05,
+      "loss": 1.4325,
+      "step": 16280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.406799855696539e-05,
+      "loss": 1.2085,
+      "step": 16290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.406086109598888e-05,
+      "loss": 1.2851,
+      "step": 16300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.405371961476628e-05,
+      "loss": 1.3606,
+      "step": 16310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.404657411394921e-05,
+      "loss": 1.2742,
+      "step": 16320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.403942459418967e-05,
+      "loss": 1.2141,
+      "step": 16330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.403227105613995e-05,
+      "loss": 1.3764,
+      "step": 16340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.402511350045279e-05,
+      "loss": 1.1283,
+      "step": 16350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.401795192778126e-05,
+      "loss": 1.2786,
+      "step": 16360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.401078633877884e-05,
+      "loss": 1.1966,
+      "step": 16370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.40036167340993e-05,
+      "loss": 1.4799,
+      "step": 16380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.399644311439682e-05,
+      "loss": 1.3863,
+      "step": 16390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.398926548032594e-05,
+      "loss": 1.2172,
+      "step": 16400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.398208383254159e-05,
+      "loss": 1.4485,
+      "step": 16410
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.397489817169905e-05,
+      "loss": 1.4933,
+      "step": 16420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.396770849845392e-05,
+      "loss": 1.3489,
+      "step": 16430
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.396051481346225e-05,
+      "loss": 1.2826,
+      "step": 16440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.395331711738038e-05,
+      "loss": 1.233,
+      "step": 16450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.394611541086508e-05,
+      "loss": 1.1535,
+      "step": 16460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.393890969457343e-05,
+      "loss": 1.0996,
+      "step": 16470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.39316999691629e-05,
+      "loss": 1.3907,
+      "step": 16480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.392448623529137e-05,
+      "loss": 1.3211,
+      "step": 16490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.3917268493617e-05,
+      "loss": 1.3094,
+      "step": 16500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.391004674479838e-05,
+      "loss": 1.3606,
+      "step": 16510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.390282098949443e-05,
+      "loss": 1.3413,
+      "step": 16520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.389559122836444e-05,
+      "loss": 1.3874,
+      "step": 16530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.388835746206812e-05,
+      "loss": 1.2007,
+      "step": 16540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.388111969126545e-05,
+      "loss": 1.4112,
+      "step": 16550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.387387791661685e-05,
+      "loss": 1.4398,
+      "step": 16560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.386663213878308e-05,
+      "loss": 1.2158,
+      "step": 16570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.385938235842526e-05,
+      "loss": 1.6013,
+      "step": 16580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.38521285762049e-05,
+      "loss": 1.2519,
+      "step": 16590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.384487079278383e-05,
+      "loss": 1.1838,
+      "step": 16600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.383760900882429e-05,
+      "loss": 1.2601,
+      "step": 16610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.383034322498886e-05,
+      "loss": 1.297,
+      "step": 16620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.38230734419405e-05,
+      "loss": 1.2262,
+      "step": 16630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.381579966034252e-05,
+      "loss": 1.543,
+      "step": 16640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.380852188085862e-05,
+      "loss": 1.3487,
+      "step": 16650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.38012401041528e-05,
+      "loss": 1.2489,
+      "step": 16660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.379395433088953e-05,
+      "loss": 1.2502,
+      "step": 16670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.378666456173355e-05,
+      "loss": 1.3287,
+      "step": 16680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.377937079735e-05,
+      "loss": 1.333,
+      "step": 16690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.377207303840441e-05,
+      "loss": 1.2786,
+      "step": 16700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.376477128556263e-05,
+      "loss": 1.1994,
+      "step": 16710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.375746553949087e-05,
+      "loss": 1.1303,
+      "step": 16720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.37501558008558e-05,
+      "loss": 1.132,
+      "step": 16730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.37428420703243e-05,
+      "loss": 1.2864,
+      "step": 16740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.373552434856374e-05,
+      "loss": 1.0516,
+      "step": 16750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.372820263624182e-05,
+      "loss": 1.2866,
+      "step": 16760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.372087693402658e-05,
+      "loss": 1.1245,
+      "step": 16770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.371354724258643e-05,
+      "loss": 1.4265,
+      "step": 16780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.370621356259018e-05,
+      "loss": 1.1376,
+      "step": 16790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.369887589470694e-05,
+      "loss": 1.4306,
+      "step": 16800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.369153423960626e-05,
+      "loss": 1.1678,
+      "step": 16810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.368418859795799e-05,
+      "loss": 1.2773,
+      "step": 16820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.367683897043238e-05,
+      "loss": 1.2845,
+      "step": 16830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.366948535770002e-05,
+      "loss": 1.3584,
+      "step": 16840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.366212776043188e-05,
+      "loss": 1.3543,
+      "step": 16850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.36547661792993e-05,
+      "loss": 1.146,
+      "step": 16860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.364740061497397e-05,
+      "loss": 1.3677,
+      "step": 16870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.364003106812793e-05,
+      "loss": 1.2294,
+      "step": 16880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.363265753943361e-05,
+      "loss": 1.2105,
+      "step": 16890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.36252800295638e-05,
+      "loss": 1.1968,
+      "step": 16900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.361789853919164e-05,
+      "loss": 1.2331,
+      "step": 16910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.361051306899065e-05,
+      "loss": 1.1401,
+      "step": 16920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.360312361963468e-05,
+      "loss": 1.2848,
+      "step": 16930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.359573019179798e-05,
+      "loss": 1.283,
+      "step": 16940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.358833278615516e-05,
+      "loss": 1.4469,
+      "step": 16950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.358093140338116e-05,
+      "loss": 1.2231,
+      "step": 16960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.35735260441513e-05,
+      "loss": 1.6433,
+      "step": 16970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.35661167091413e-05,
+      "loss": 1.2125,
+      "step": 16980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.355870339902718e-05,
+      "loss": 1.2966,
+      "step": 16990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.355128611448536e-05,
+      "loss": 1.1292,
+      "step": 17000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.354386485619264e-05,
+      "loss": 1.3132,
+      "step": 17010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.353643962482612e-05,
+      "loss": 1.6297,
+      "step": 17020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.352901042106332e-05,
+      "loss": 1.2072,
+      "step": 17030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.35215772455821e-05,
+      "loss": 1.322,
+      "step": 17040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.35141400990607e-05,
+      "loss": 1.4751,
+      "step": 17050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.350669898217768e-05,
+      "loss": 1.6172,
+      "step": 17060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.349925389561201e-05,
+      "loss": 1.2036,
+      "step": 17070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.349180484004299e-05,
+      "loss": 1.3299,
+      "step": 17080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.34843518161503e-05,
+      "loss": 1.2185,
+      "step": 17090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.347689482461398e-05,
+      "loss": 1.1816,
+      "step": 17100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.346943386611442e-05,
+      "loss": 1.4984,
+      "step": 17110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.346196894133239e-05,
+      "loss": 1.2053,
+      "step": 17120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.345450005094902e-05,
+      "loss": 1.4044,
+      "step": 17130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.344702719564576e-05,
+      "loss": 1.368,
+      "step": 17140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.34395503761045e-05,
+      "loss": 1.6843,
+      "step": 17150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.34320695930074e-05,
+      "loss": 1.2494,
+      "step": 17160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.342458484703708e-05,
+      "loss": 1.4661,
+      "step": 17170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.341709613887643e-05,
+      "loss": 1.2289,
+      "step": 17180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.340960346920877e-05,
+      "loss": 1.3002,
+      "step": 17190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.340210683871773e-05,
+      "loss": 1.3765,
+      "step": 17200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.339460624808733e-05,
+      "loss": 1.2281,
+      "step": 17210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.338710169800198e-05,
+      "loss": 1.1963,
+      "step": 17220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.337959318914638e-05,
+      "loss": 1.1941,
+      "step": 17230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.337208072220563e-05,
+      "loss": 1.1274,
+      "step": 17240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.33645642978652e-05,
+      "loss": 1.2854,
+      "step": 17250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.335704391681092e-05,
+      "loss": 1.1231,
+      "step": 17260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.334951957972898e-05,
+      "loss": 1.3419,
+      "step": 17270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.334199128730587e-05,
+      "loss": 1.2927,
+      "step": 17280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.333445904022856e-05,
+      "loss": 1.1611,
+      "step": 17290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.332692283918427e-05,
+      "loss": 1.237,
+      "step": 17300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.331938268486065e-05,
+      "loss": 1.2885,
+      "step": 17310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.331183857794569e-05,
+      "loss": 1.1829,
+      "step": 17320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.33042905191277e-05,
+      "loss": 1.4356,
+      "step": 17330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.329673850909544e-05,
+      "loss": 1.328,
+      "step": 17340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.328918254853794e-05,
+      "loss": 1.6546,
+      "step": 17350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.328162263814463e-05,
+      "loss": 1.2788,
+      "step": 17360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.327405877860533e-05,
+      "loss": 1.4134,
+      "step": 17370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.326649097061015e-05,
+      "loss": 1.0333,
+      "step": 17380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.325891921484964e-05,
+      "loss": 1.4625,
+      "step": 17390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.325134351201463e-05,
+      "loss": 0.9974,
+      "step": 17400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.324376386279637e-05,
+      "loss": 1.1835,
+      "step": 17410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.323618026788645e-05,
+      "loss": 1.1063,
+      "step": 17420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.322859272797684e-05,
+      "loss": 1.3024,
+      "step": 17430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.322100124375981e-05,
+      "loss": 1.2947,
+      "step": 17440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.321340581592804e-05,
+      "loss": 1.2849,
+      "step": 17450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.320580644517458e-05,
+      "loss": 1.4357,
+      "step": 17460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.319820313219281e-05,
+      "loss": 1.294,
+      "step": 17470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.319059587767647e-05,
+      "loss": 1.3898,
+      "step": 17480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.318298468231967e-05,
+      "loss": 1.2725,
+      "step": 17490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.31753695468169e-05,
+      "loss": 1.2791,
+      "step": 17500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.316775047186296e-05,
+      "loss": 1.0679,
+      "step": 17510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.316012745815306e-05,
+      "loss": 1.2829,
+      "step": 17520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.315250050638274e-05,
+      "loss": 1.2124,
+      "step": 17530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.314486961724788e-05,
+      "loss": 1.4354,
+      "step": 17540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.313723479144478e-05,
+      "loss": 1.0578,
+      "step": 17550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.312959602967006e-05,
+      "loss": 1.3993,
+      "step": 17560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.31219533326207e-05,
+      "loss": 1.3431,
+      "step": 17570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.311430670099402e-05,
+      "loss": 1.5156,
+      "step": 17580
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.310665613548775e-05,
+      "loss": 1.2122,
+      "step": 17590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.309900163679996e-05,
+      "loss": 1.2211,
+      "step": 17600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.309134320562902e-05,
+      "loss": 1.3607,
+      "step": 17610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.308368084267376e-05,
+      "loss": 1.4265,
+      "step": 17620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.307601454863331e-05,
+      "loss": 1.3635,
+      "step": 17630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.306834432420712e-05,
+      "loss": 1.2899,
+      "step": 17640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.30606701700951e-05,
+      "loss": 1.1361,
+      "step": 17650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.305299208699742e-05,
+      "loss": 1.2038,
+      "step": 17660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.30453100756147e-05,
+      "loss": 1.2701,
+      "step": 17670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.303762413664781e-05,
+      "loss": 1.351,
+      "step": 17680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.302993427079809e-05,
+      "loss": 1.0281,
+      "step": 17690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.302224047876717e-05,
+      "loss": 1.3172,
+      "step": 17700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.301454276125704e-05,
+      "loss": 1.431,
+      "step": 17710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.300684111897008e-05,
+      "loss": 1.5343,
+      "step": 17720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.2999135552609e-05,
+      "loss": 1.2087,
+      "step": 17730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.299142606287688e-05,
+      "loss": 1.2349,
+      "step": 17740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.298371265047718e-05,
+      "loss": 1.4836,
+      "step": 17750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.297599531611366e-05,
+      "loss": 1.2087,
+      "step": 17760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.29682740604905e-05,
+      "loss": 1.0398,
+      "step": 17770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.296054888431219e-05,
+      "loss": 1.3909,
+      "step": 17780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.295281978828362e-05,
+      "loss": 1.2426,
+      "step": 17790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.294508677311e-05,
+      "loss": 1.1036,
+      "step": 17800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.293734983949691e-05,
+      "loss": 1.3398,
+      "step": 17810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.292960898815032e-05,
+      "loss": 1.3364,
+      "step": 17820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.292186421977648e-05,
+      "loss": 1.1406,
+      "step": 17830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.29141155350821e-05,
+      "loss": 0.9932,
+      "step": 17840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.290636293477417e-05,
+      "loss": 1.2885,
+      "step": 17850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.289860641956004e-05,
+      "loss": 1.3683,
+      "step": 17860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.289084599014746e-05,
+      "loss": 1.27,
+      "step": 17870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.288308164724452e-05,
+      "loss": 1.2984,
+      "step": 17880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.287531339155963e-05,
+      "loss": 1.4012,
+      "step": 17890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.286754122380165e-05,
+      "loss": 1.1286,
+      "step": 17900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.285976514467968e-05,
+      "loss": 1.168,
+      "step": 17910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.285198515490325e-05,
+      "loss": 1.1317,
+      "step": 17920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.284420125518223e-05,
+      "loss": 1.3636,
+      "step": 17930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.283641344622684e-05,
+      "loss": 0.9423,
+      "step": 17940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.28286217287477e-05,
+      "loss": 1.3026,
+      "step": 17950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.28208261034557e-05,
+      "loss": 1.2703,
+      "step": 17960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.281302657106217e-05,
+      "loss": 1.3029,
+      "step": 17970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.280522313227873e-05,
+      "loss": 1.4293,
+      "step": 17980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.279741578781742e-05,
+      "loss": 1.1543,
+      "step": 17990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.27896045383906e-05,
+      "loss": 1.1441,
+      "step": 18000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.278178938471098e-05,
+      "loss": 1.2155,
+      "step": 18010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.277397032749164e-05,
+      "loss": 1.3241,
+      "step": 18020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.276614736744604e-05,
+      "loss": 1.1803,
+      "step": 18030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.275832050528794e-05,
+      "loss": 1.0748,
+      "step": 18040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.27504897417315e-05,
+      "loss": 1.4759,
+      "step": 18050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.274265507749121e-05,
+      "loss": 1.514,
+      "step": 18060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.273481651328194e-05,
+      "loss": 1.1424,
+      "step": 18070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.272697404981892e-05,
+      "loss": 1.3387,
+      "step": 18080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.271912768781768e-05,
+      "loss": 1.3487,
+      "step": 18090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.271127742799417e-05,
+      "loss": 1.3531,
+      "step": 18100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.270342327106467e-05,
+      "loss": 1.5283,
+      "step": 18110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.269556521774582e-05,
+      "loss": 1.429,
+      "step": 18120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.268770326875462e-05,
+      "loss": 1.4716,
+      "step": 18130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.26798374248084e-05,
+      "loss": 1.1919,
+      "step": 18140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.267196768662488e-05,
+      "loss": 1.2246,
+      "step": 18150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.266409405492209e-05,
+      "loss": 1.0055,
+      "step": 18160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.265621653041848e-05,
+      "loss": 1.2463,
+      "step": 18170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.264833511383279e-05,
+      "loss": 1.2515,
+      "step": 18180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.264044980588416e-05,
+      "loss": 1.299,
+      "step": 18190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.263256060729206e-05,
+      "loss": 1.3153,
+      "step": 18200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.262466751877634e-05,
+      "loss": 1.3898,
+      "step": 18210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.261677054105717e-05,
+      "loss": 1.2996,
+      "step": 18220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.260886967485511e-05,
+      "loss": 1.3104,
+      "step": 18230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.260096492089105e-05,
+      "loss": 1.1722,
+      "step": 18240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.259305627988624e-05,
+      "loss": 1.2576,
+      "step": 18250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.258514375256231e-05,
+      "loss": 1.0259,
+      "step": 18260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.25772273396412e-05,
+      "loss": 1.3075,
+      "step": 18270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.256930704184524e-05,
+      "loss": 1.1774,
+      "step": 18280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.256138285989709e-05,
+      "loss": 1.2613,
+      "step": 18290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.255345479451977e-05,
+      "loss": 1.315,
+      "step": 18300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.25455228464367e-05,
+      "loss": 1.2115,
+      "step": 18310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.253758701637156e-05,
+      "loss": 1.1275,
+      "step": 18320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.252964730504848e-05,
+      "loss": 1.2247,
+      "step": 18330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.252170371319189e-05,
+      "loss": 1.3678,
+      "step": 18340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.25137562415266e-05,
+      "loss": 1.3899,
+      "step": 18350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.250580489077774e-05,
+      "loss": 1.1713,
+      "step": 18360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.24978496616708e-05,
+      "loss": 1.2295,
+      "step": 18370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.248989055493171e-05,
+      "loss": 1.4002,
+      "step": 18380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.248192757128661e-05,
+      "loss": 1.5625,
+      "step": 18390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.24739607114621e-05,
+      "loss": 1.2804,
+      "step": 18400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.24659899761851e-05,
+      "loss": 1.3862,
+      "step": 18410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.245801536618286e-05,
+      "loss": 1.2346,
+      "step": 18420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.245003688218303e-05,
+      "loss": 1.4635,
+      "step": 18430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.244205452491359e-05,
+      "loss": 1.3998,
+      "step": 18440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.243406829510287e-05,
+      "loss": 1.1615,
+      "step": 18450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.242607819347958e-05,
+      "loss": 1.219,
+      "step": 18460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.24180842207727e-05,
+      "loss": 1.2745,
+      "step": 18470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.24100863777117e-05,
+      "loss": 1.1178,
+      "step": 18480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.240208466502625e-05,
+      "loss": 1.0944,
+      "step": 18490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.239407908344652e-05,
+      "loss": 1.3658,
+      "step": 18500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.238606963370293e-05,
+      "loss": 1.2966,
+      "step": 18510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.23780563165263e-05,
+      "loss": 1.2591,
+      "step": 18520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.237003913264776e-05,
+      "loss": 1.0441,
+      "step": 18530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.236201808279886e-05,
+      "loss": 1.2045,
+      "step": 18540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.235399316771143e-05,
+      "loss": 1.3507,
+      "step": 18550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.234596438811771e-05,
+      "loss": 1.1446,
+      "step": 18560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.233793174475028e-05,
+      "loss": 1.1164,
+      "step": 18570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.232989523834204e-05,
+      "loss": 1.2103,
+      "step": 18580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.232185486962627e-05,
+      "loss": 1.3485,
+      "step": 18590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.231381063933659e-05,
+      "loss": 1.1548,
+      "step": 18600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.230576254820702e-05,
+      "loss": 1.2474,
+      "step": 18610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.229771059697184e-05,
+      "loss": 1.2301,
+      "step": 18620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.228965478636575e-05,
+      "loss": 1.3635,
+      "step": 18630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.228159511712381e-05,
+      "loss": 1.4085,
+      "step": 18640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.227353158998139e-05,
+      "loss": 1.2616,
+      "step": 18650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.226546420567422e-05,
+      "loss": 1.1465,
+      "step": 18660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.22573929649384e-05,
+      "loss": 1.249,
+      "step": 18670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.224931786851041e-05,
+      "loss": 1.1245,
+      "step": 18680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.2241238917127e-05,
+      "loss": 1.3263,
+      "step": 18690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.223315611152531e-05,
+      "loss": 1.2855,
+      "step": 18700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.22250694524429e-05,
+      "loss": 1.2554,
+      "step": 18710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.221697894061756e-05,
+      "loss": 1.4284,
+      "step": 18720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.220888457678752e-05,
+      "loss": 1.4104,
+      "step": 18730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.220078636169134e-05,
+      "loss": 1.3955,
+      "step": 18740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.21926842960679e-05,
+      "loss": 1.1677,
+      "step": 18750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.21845783806565e-05,
+      "loss": 1.0718,
+      "step": 18760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.21764686161967e-05,
+      "loss": 1.2162,
+      "step": 18770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.21683550034285e-05,
+      "loss": 1.2101,
+      "step": 18780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.216023754309218e-05,
+      "loss": 1.1701,
+      "step": 18790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.215211623592842e-05,
+      "loss": 1.2087,
+      "step": 18800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.214399108267822e-05,
+      "loss": 1.2124,
+      "step": 18810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.213586208408294e-05,
+      "loss": 1.1733,
+      "step": 18820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.21277292408843e-05,
+      "loss": 1.1636,
+      "step": 18830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.21195925538244e-05,
+      "loss": 1.3481,
+      "step": 18840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.21114520236456e-05,
+      "loss": 1.4059,
+      "step": 18850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.21033076510907e-05,
+      "loss": 1.2956,
+      "step": 18860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.209515943690281e-05,
+      "loss": 1.3075,
+      "step": 18870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.208700738182539e-05,
+      "loss": 1.2157,
+      "step": 18880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.207885148660226e-05,
+      "loss": 1.2487,
+      "step": 18890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.207069175197762e-05,
+      "loss": 1.4118,
+      "step": 18900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.206252817869594e-05,
+      "loss": 1.3536,
+      "step": 18910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.205436076750212e-05,
+      "loss": 1.3249,
+      "step": 18920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.204618951914137e-05,
+      "loss": 1.4284,
+      "step": 18930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.203801443435924e-05,
+      "loss": 1.4646,
+      "step": 18940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.20298355139017e-05,
+      "loss": 1.2792,
+      "step": 18950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.202165275851497e-05,
+      "loss": 1.1213,
+      "step": 18960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.201346616894568e-05,
+      "loss": 1.1528,
+      "step": 18970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.200527574594082e-05,
+      "loss": 1.2696,
+      "step": 18980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.199708149024769e-05,
+      "loss": 1.2345,
+      "step": 18990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.198888340261397e-05,
+      "loss": 1.5037,
+      "step": 19000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.198068148378767e-05,
+      "loss": 1.1251,
+      "step": 19010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.197247573451718e-05,
+      "loss": 1.1563,
+      "step": 19020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.196426615555116e-05,
+      "loss": 1.1353,
+      "step": 19030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.195605274763875e-05,
+      "loss": 1.2654,
+      "step": 19040
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.19478355115293e-05,
+      "loss": 1.3081,
+      "step": 19050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.193961444797263e-05,
+      "loss": 1.3418,
+      "step": 19060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.193138955771883e-05,
+      "loss": 1.3319,
+      "step": 19070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.192316084151835e-05,
+      "loss": 1.2861,
+      "step": 19080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.191492830012202e-05,
+      "loss": 1.4887,
+      "step": 19090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.190669193428101e-05,
+      "loss": 1.1558,
+      "step": 19100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.189845174474681e-05,
+      "loss": 1.2116,
+      "step": 19110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.18902077322713e-05,
+      "loss": 1.0554,
+      "step": 19120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.188195989760668e-05,
+      "loss": 1.2884,
+      "step": 19130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.18737082415055e-05,
+      "loss": 1.2485,
+      "step": 19140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.186545276472067e-05,
+      "loss": 1.0407,
+      "step": 19150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.185719346800546e-05,
+      "loss": 1.2874,
+      "step": 19160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.184893035211346e-05,
+      "loss": 1.2883,
+      "step": 19170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.184066341779862e-05,
+      "loss": 1.1917,
+      "step": 19180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.183239266581523e-05,
+      "loss": 1.1768,
+      "step": 19190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.182411809691795e-05,
+      "loss": 1.2362,
+      "step": 19200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.181583971186179e-05,
+      "loss": 1.2328,
+      "step": 19210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.180755751140207e-05,
+      "loss": 1.217,
+      "step": 19220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.17992714962945e-05,
+      "loss": 1.2306,
+      "step": 19230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.179098166729513e-05,
+      "loss": 1.0566,
+      "step": 19240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.178268802516032e-05,
+      "loss": 1.3224,
+      "step": 19250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.177439057064683e-05,
+      "loss": 1.0436,
+      "step": 19260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.176608930451173e-05,
+      "loss": 1.0848,
+      "step": 19270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.175778422751246e-05,
+      "loss": 1.2881,
+      "step": 19280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.174947534040681e-05,
+      "loss": 1.337,
+      "step": 19290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.174116264395289e-05,
+      "loss": 1.0918,
+      "step": 19300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.173284613890918e-05,
+      "loss": 1.1272,
+      "step": 19310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.17245258260345e-05,
+      "loss": 1.2602,
+      "step": 19320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.171620170608803e-05,
+      "loss": 1.4133,
+      "step": 19330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.17078737798293e-05,
+      "loss": 1.2353,
+      "step": 19340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.169954204801816e-05,
+      "loss": 1.4011,
+      "step": 19350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.169120651141481e-05,
+      "loss": 1.2604,
+      "step": 19360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.168286717077981e-05,
+      "loss": 1.0883,
+      "step": 19370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.16745240268741e-05,
+      "loss": 1.3261,
+      "step": 19380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.16661770804589e-05,
+      "loss": 1.2684,
+      "step": 19390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.165782633229582e-05,
+      "loss": 1.2597,
+      "step": 19400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.164947178314683e-05,
+      "loss": 1.1418,
+      "step": 19410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.16411134337742e-05,
+      "loss": 1.2888,
+      "step": 19420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.163275128494056e-05,
+      "loss": 0.9184,
+      "step": 19430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.162438533740892e-05,
+      "loss": 1.2341,
+      "step": 19440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.161601559194261e-05,
+      "loss": 1.4176,
+      "step": 19450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.16076420493053e-05,
+      "loss": 1.1862,
+      "step": 19460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.159926471026104e-05,
+      "loss": 1.4162,
+      "step": 19470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.159088357557417e-05,
+      "loss": 1.5603,
+      "step": 19480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.158249864600946e-05,
+      "loss": 1.4133,
+      "step": 19490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.157410992233191e-05,
+      "loss": 1.3486,
+      "step": 19500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.156571740530699e-05,
+      "loss": 1.2212,
+      "step": 19510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.155732109570043e-05,
+      "loss": 1.2289,
+      "step": 19520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.154892099427834e-05,
+      "loss": 1.0904,
+      "step": 19530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.154051710180718e-05,
+      "loss": 1.1836,
+      "step": 19540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.153210941905374e-05,
+      "loss": 1.2065,
+      "step": 19550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.152369794678515e-05,
+      "loss": 1.3365,
+      "step": 19560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.151528268576892e-05,
+      "loss": 1.0631,
+      "step": 19570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.150686363677287e-05,
+      "loss": 1.3591,
+      "step": 19580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.149844080056518e-05,
+      "loss": 1.4261,
+      "step": 19590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.149001417791438e-05,
+      "loss": 1.3315,
+      "step": 19600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.148158376958932e-05,
+      "loss": 1.2509,
+      "step": 19610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.147314957635926e-05,
+      "loss": 1.366,
+      "step": 19620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.146471159899372e-05,
+      "loss": 1.2429,
+      "step": 19630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.145626983826264e-05,
+      "loss": 1.1623,
+      "step": 19640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.144782429493623e-05,
+      "loss": 1.2967,
+      "step": 19650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.143937496978512e-05,
+      "loss": 1.0439,
+      "step": 19660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.143092186358026e-05,
+      "loss": 1.0677,
+      "step": 19670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.14224649770929e-05,
+      "loss": 1.3888,
+      "step": 19680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.14140043110947e-05,
+      "loss": 1.2772,
+      "step": 19690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.140553986635763e-05,
+      "loss": 1.3543,
+      "step": 19700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.139707164365399e-05,
+      "loss": 1.1133,
+      "step": 19710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.138859964375648e-05,
+      "loss": 1.37,
+      "step": 19720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.13801238674381e-05,
+      "loss": 1.1923,
+      "step": 19730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.137164431547221e-05,
+      "loss": 1.3854,
+      "step": 19740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.136316098863247e-05,
+      "loss": 1.4615,
+      "step": 19750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.135467388769298e-05,
+      "loss": 1.2193,
+      "step": 19760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.13461830134281e-05,
+      "loss": 1.2675,
+      "step": 19770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.133768836661257e-05,
+      "loss": 1.246,
+      "step": 19780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.132918994802147e-05,
+      "loss": 1.2972,
+      "step": 19790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.132068775843022e-05,
+      "loss": 1.2293,
+      "step": 19800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.131218179861456e-05,
+      "loss": 1.2536,
+      "step": 19810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.130367206935063e-05,
+      "loss": 1.2778,
+      "step": 19820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.129515857141488e-05,
+      "loss": 1.2956,
+      "step": 19830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.128664130558411e-05,
+      "loss": 1.3419,
+      "step": 19840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.127812027263545e-05,
+      "loss": 1.352,
+      "step": 19850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.126959547334637e-05,
+      "loss": 1.1594,
+      "step": 19860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.126106690849474e-05,
+      "loss": 1.3724,
+      "step": 19870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.125253457885869e-05,
+      "loss": 1.2078,
+      "step": 19880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.124399848521677e-05,
+      "loss": 1.4347,
+      "step": 19890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.123545862834781e-05,
+      "loss": 1.2864,
+      "step": 19900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.122691500903104e-05,
+      "loss": 1.2206,
+      "step": 19910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.121836762804597e-05,
+      "loss": 1.3752,
+      "step": 19920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.120981648617252e-05,
+      "loss": 1.108,
+      "step": 19930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.120126158419092e-05,
+      "loss": 1.266,
+      "step": 19940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.119270292288173e-05,
+      "loss": 1.3314,
+      "step": 19950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.118414050302588e-05,
+      "loss": 1.1617,
+      "step": 19960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.117557432540462e-05,
+      "loss": 1.3017,
+      "step": 19970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.116700439079958e-05,
+      "loss": 1.2081,
+      "step": 19980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.115843069999266e-05,
+      "loss": 1.2585,
+      "step": 19990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.114985325376619e-05,
+      "loss": 1.4547,
+      "step": 20000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.114127205290279e-05,
+      "loss": 1.3955,
+      "step": 20010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.113268709818543e-05,
+      "loss": 1.318,
+      "step": 20020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.112409839039742e-05,
+      "loss": 1.4196,
+      "step": 20030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.111550593032244e-05,
+      "loss": 1.1473,
+      "step": 20040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.110690971874449e-05,
+      "loss": 1.3523,
+      "step": 20050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.109830975644788e-05,
+      "loss": 1.324,
+      "step": 20060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.108970604421736e-05,
+      "loss": 1.0995,
+      "step": 20070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.10810985828379e-05,
+      "loss": 1.2547,
+      "step": 20080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.107248737309488e-05,
+      "loss": 1.3314,
+      "step": 20090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.106387241577405e-05,
+      "loss": 1.5519,
+      "step": 20100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.105525371166144e-05,
+      "loss": 1.204,
+      "step": 20110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.104663126154343e-05,
+      "loss": 1.47,
+      "step": 20120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.103800506620678e-05,
+      "loss": 1.2855,
+      "step": 20130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.102937512643856e-05,
+      "loss": 1.0478,
+      "step": 20140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.102074144302622e-05,
+      "loss": 1.3733,
+      "step": 20150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.101210401675748e-05,
+      "loss": 1.4146,
+      "step": 20160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.100346284842048e-05,
+      "loss": 1.2943,
+      "step": 20170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.099481793880362e-05,
+      "loss": 1.2103,
+      "step": 20180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.098616928869575e-05,
+      "loss": 1.2752,
+      "step": 20190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.097751689888595e-05,
+      "loss": 1.1991,
+      "step": 20200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.096886077016371e-05,
+      "loss": 1.3585,
+      "step": 20210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.096020090331885e-05,
+      "loss": 1.401,
+      "step": 20220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.09515372991415e-05,
+      "loss": 1.2516,
+      "step": 20230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.094286995842218e-05,
+      "loss": 1.2289,
+      "step": 20240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.09341988819517e-05,
+      "loss": 1.3863,
+      "step": 20250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.092552407052123e-05,
+      "loss": 1.2974,
+      "step": 20260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.091684552492233e-05,
+      "loss": 1.3099,
+      "step": 20270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.090816324594682e-05,
+      "loss": 1.3949,
+      "step": 20280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.08994772343869e-05,
+      "loss": 1.283,
+      "step": 20290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.089078749103511e-05,
+      "loss": 1.2604,
+      "step": 20300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.088209401668433e-05,
+      "loss": 1.3067,
+      "step": 20310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.087339681212779e-05,
+      "loss": 1.299,
+      "step": 20320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.086469587815904e-05,
+      "loss": 1.4764,
+      "step": 20330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.085599121557197e-05,
+      "loss": 1.4707,
+      "step": 20340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.084728282516084e-05,
+      "loss": 1.0965,
+      "step": 20350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.083857070772019e-05,
+      "loss": 1.3065,
+      "step": 20360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.082985486404501e-05,
+      "loss": 1.2415,
+      "step": 20370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.082113529493048e-05,
+      "loss": 1.1175,
+      "step": 20380
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.081241200117227e-05,
+      "loss": 1.244,
+      "step": 20390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.080368498356628e-05,
+      "loss": 1.2925,
+      "step": 20400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.079495424290878e-05,
+      "loss": 1.3885,
+      "step": 20410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.078621977999642e-05,
+      "loss": 1.1691,
+      "step": 20420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.077748159562613e-05,
+      "loss": 1.4151,
+      "step": 20430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.076873969059526e-05,
+      "loss": 1.4083,
+      "step": 20440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.075999406570137e-05,
+      "loss": 1.2528,
+      "step": 20450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.07512447217425e-05,
+      "loss": 1.2137,
+      "step": 20460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.074249165951694e-05,
+      "loss": 1.3431,
+      "step": 20470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.073373487982335e-05,
+      "loss": 1.4769,
+      "step": 20480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.072497438346072e-05,
+      "loss": 1.2392,
+      "step": 20490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.07162101712284e-05,
+      "loss": 1.2963,
+      "step": 20500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.070744224392606e-05,
+      "loss": 1.1272,
+      "step": 20510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.06986706023537e-05,
+      "loss": 1.3848,
+      "step": 20520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.068989524731168e-05,
+      "loss": 1.4085,
+      "step": 20530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.068111617960067e-05,
+      "loss": 1.2047,
+      "step": 20540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.067233340002173e-05,
+      "loss": 1.3008,
+      "step": 20550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.066354690937622e-05,
+      "loss": 1.3661,
+      "step": 20560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.065475670846582e-05,
+      "loss": 1.2619,
+      "step": 20570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.06459627980926e-05,
+      "loss": 1.1163,
+      "step": 20580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.063716517905894e-05,
+      "loss": 0.9725,
+      "step": 20590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.062836385216755e-05,
+      "loss": 1.2133,
+      "step": 20600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.061955881822151e-05,
+      "loss": 0.9211,
+      "step": 20610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.061075007802419e-05,
+      "loss": 1.4721,
+      "step": 20620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.060193763237934e-05,
+      "loss": 1.3133,
+      "step": 20630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.059312148209105e-05,
+      "loss": 1.5812,
+      "step": 20640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.058430162796371e-05,
+      "loss": 1.28,
+      "step": 20650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.057547807080208e-05,
+      "loss": 1.4741,
+      "step": 20660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.056665081141125e-05,
+      "loss": 1.2443,
+      "step": 20670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.055781985059663e-05,
+      "loss": 1.0663,
+      "step": 20680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.054898518916399e-05,
+      "loss": 1.298,
+      "step": 20690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.054014682791945e-05,
+      "loss": 1.2759,
+      "step": 20700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.053130476766943e-05,
+      "loss": 1.3189,
+      "step": 20710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.05224590092207e-05,
+      "loss": 1.3413,
+      "step": 20720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.05136095533804e-05,
+      "loss": 1.2042,
+      "step": 20730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.050475640095598e-05,
+      "loss": 1.4885,
+      "step": 20740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.049589955275518e-05,
+      "loss": 1.6309,
+      "step": 20750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.048703900958617e-05,
+      "loss": 1.4646,
+      "step": 20760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.047817477225741e-05,
+      "loss": 1.3543,
+      "step": 20770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.04693068415777e-05,
+      "loss": 1.3609,
+      "step": 20780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.046043521835617e-05,
+      "loss": 1.3234,
+      "step": 20790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.045155990340227e-05,
+      "loss": 1.4644,
+      "step": 20800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.044268089752586e-05,
+      "loss": 1.2933,
+      "step": 20810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.043379820153703e-05,
+      "loss": 1.5008,
+      "step": 20820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.042491181624633e-05,
+      "loss": 1.3722,
+      "step": 20830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.04160217424645e-05,
+      "loss": 1.1812,
+      "step": 20840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.040712798100279e-05,
+      "loss": 1.3886,
+      "step": 20850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.039823053267262e-05,
+      "loss": 1.3184,
+      "step": 20860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.038932939828586e-05,
+      "loss": 1.2823,
+      "step": 20870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.038042457865465e-05,
+      "loss": 1.1431,
+      "step": 20880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.037151607459151e-05,
+      "loss": 1.3602,
+      "step": 20890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.036260388690927e-05,
+      "loss": 1.2437,
+      "step": 20900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.035368801642112e-05,
+      "loss": 1.359,
+      "step": 20910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.034476846394055e-05,
+      "loss": 1.2629,
+      "step": 20920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.033584523028142e-05,
+      "loss": 1.394,
+      "step": 20930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.032691831625792e-05,
+      "loss": 1.2364,
+      "step": 20940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.031798772268455e-05,
+      "loss": 1.2808,
+      "step": 20950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.030905345037617e-05,
+      "loss": 1.1947,
+      "step": 20960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.030011550014798e-05,
+      "loss": 1.4801,
+      "step": 20970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.02911738728155e-05,
+      "loss": 1.2582,
+      "step": 20980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.028222856919458e-05,
+      "loss": 1.2728,
+      "step": 20990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.027327959010142e-05,
+      "loss": 1.3302,
+      "step": 21000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.026432693635257e-05,
+      "loss": 1.2725,
+      "step": 21010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.025537060876488e-05,
+      "loss": 1.1886,
+      "step": 21020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.024641060815557e-05,
+      "loss": 1.4478,
+      "step": 21030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.023744693534215e-05,
+      "loss": 1.2868,
+      "step": 21040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.02284795911425e-05,
+      "loss": 1.2076,
+      "step": 21050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.021950857637486e-05,
+      "loss": 1.3483,
+      "step": 21060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.021053389185773e-05,
+      "loss": 1.4153,
+      "step": 21070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.020155553841e-05,
+      "loss": 1.1559,
+      "step": 21080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.01925735168509e-05,
+      "loss": 1.4696,
+      "step": 21090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.018358782799995e-05,
+      "loss": 1.3476,
+      "step": 21100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.017459847267708e-05,
+      "loss": 1.3061,
+      "step": 21110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.016560545170243e-05,
+      "loss": 1.2134,
+      "step": 21120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.015660876589662e-05,
+      "loss": 1.3707,
+      "step": 21130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.01476084160805e-05,
+      "loss": 1.4401,
+      "step": 21140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.013860440307529e-05,
+      "loss": 1.3131,
+      "step": 21150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.012959672770255e-05,
+      "loss": 1.4556,
+      "step": 21160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.012058539078416e-05,
+      "loss": 1.3812,
+      "step": 21170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.011157039314235e-05,
+      "loss": 1.5042,
+      "step": 21180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.010255173559967e-05,
+      "loss": 1.4034,
+      "step": 21190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.009352941897903e-05,
+      "loss": 1.077,
+      "step": 21200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.008450344410361e-05,
+      "loss": 1.3458,
+      "step": 21210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.007547381179703e-05,
+      "loss": 1.3218,
+      "step": 21220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.006644052288312e-05,
+      "loss": 1.5008,
+      "step": 21230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.005740357818612e-05,
+      "loss": 1.3053,
+      "step": 21240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.004836297853063e-05,
+      "loss": 1.3624,
+      "step": 21250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.003931872474149e-05,
+      "loss": 1.4755,
+      "step": 21260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.003027081764395e-05,
+      "loss": 1.3678,
+      "step": 21270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.002121925806355e-05,
+      "loss": 1.3631,
+      "step": 21280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.00121640468262e-05,
+      "loss": 1.1419,
+      "step": 21290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.000310518475812e-05,
+      "loss": 1.1803,
+      "step": 21300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.999404267268588e-05,
+      "loss": 1.3218,
+      "step": 21310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.998497651143634e-05,
+      "loss": 1.3041,
+      "step": 21320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.997590670183673e-05,
+      "loss": 1.3191,
+      "step": 21330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.996683324471464e-05,
+      "loss": 1.1343,
+      "step": 21340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.995775614089795e-05,
+      "loss": 1.3106,
+      "step": 21350
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.994867539121484e-05,
+      "loss": 1.2002,
+      "step": 21360
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.993959099649391e-05,
+      "loss": 1.4418,
+      "step": 21370
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.993050295756404e-05,
+      "loss": 1.3346,
+      "step": 21380
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.992141127525443e-05,
+      "loss": 1.3023,
+      "step": 21390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.991231595039465e-05,
+      "loss": 1.2754,
+      "step": 21400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.990321698381459e-05,
+      "loss": 1.4973,
+      "step": 21410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.989411437634445e-05,
+      "loss": 1.3964,
+      "step": 21420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.988500812881478e-05,
+      "loss": 1.2517,
+      "step": 21430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.987589824205649e-05,
+      "loss": 1.1416,
+      "step": 21440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.986678471690075e-05,
+      "loss": 1.3243,
+      "step": 21450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.985766755417915e-05,
+      "loss": 1.2148,
+      "step": 21460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.984854675472351e-05,
+      "loss": 1.3652,
+      "step": 21470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.98394223193661e-05,
+      "loss": 1.3323,
+      "step": 21480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.983029424893945e-05,
+      "loss": 1.2251,
+      "step": 21490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.98211625442764e-05,
+      "loss": 1.1773,
+      "step": 21500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.981202720621019e-05,
+      "loss": 1.1245,
+      "step": 21510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.980288823557432e-05,
+      "loss": 1.2081,
+      "step": 21520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.979374563320267e-05,
+      "loss": 1.1515,
+      "step": 21530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.978459939992946e-05,
+      "loss": 1.2431,
+      "step": 21540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.97754495365892e-05,
+      "loss": 1.1934,
+      "step": 21550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.976629604401677e-05,
+      "loss": 1.1427,
+      "step": 21560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.975713892304733e-05,
+      "loss": 1.3095,
+      "step": 21570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.974797817451642e-05,
+      "loss": 1.2245,
+      "step": 21580
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.97388137992599e-05,
+      "loss": 1.0905,
+      "step": 21590
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.972964579811395e-05,
+      "loss": 1.473,
+      "step": 21600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.972047417191511e-05,
+      "loss": 1.2342,
+      "step": 21610
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.971129892150019e-05,
+      "loss": 1.2303,
+      "step": 21620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.970212004770637e-05,
+      "loss": 1.4936,
+      "step": 21630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.969293755137118e-05,
+      "loss": 1.146,
+      "step": 21640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.968375143333246e-05,
+      "loss": 1.4262,
+      "step": 21650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.967456169442835e-05,
+      "loss": 1.3256,
+      "step": 21660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.966536833549737e-05,
+      "loss": 1.5495,
+      "step": 21670
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.965617135737834e-05,
+      "loss": 1.3995,
+      "step": 21680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.964697076091045e-05,
+      "loss": 1.2397,
+      "step": 21690
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.963776654693314e-05,
+      "loss": 1.2213,
+      "step": 21700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.962855871628628e-05,
+      "loss": 1.3756,
+      "step": 21710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.961934726980999e-05,
+      "loss": 1.042,
+      "step": 21720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.961013220834477e-05,
+      "loss": 1.3547,
+      "step": 21730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.960091353273142e-05,
+      "loss": 1.0708,
+      "step": 21740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.959169124381105e-05,
+      "loss": 1.0841,
+      "step": 21750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.958246534242518e-05,
+      "loss": 1.2466,
+      "step": 21760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.957323582941557e-05,
+      "loss": 1.3134,
+      "step": 21770
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.956400270562438e-05,
+      "loss": 1.3125,
+      "step": 21780
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.955476597189406e-05,
+      "loss": 1.2383,
+      "step": 21790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.954552562906737e-05,
+      "loss": 1.2097,
+      "step": 21800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.953628167798747e-05,
+      "loss": 1.2633,
+      "step": 21810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.952703411949777e-05,
+      "loss": 1.3761,
+      "step": 21820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.951778295444206e-05,
+      "loss": 1.3495,
+      "step": 21830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.950852818366445e-05,
+      "loss": 1.1768,
+      "step": 21840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.949926980800935e-05,
+      "loss": 1.2767,
+      "step": 21850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.949000782832155e-05,
+      "loss": 1.0537,
+      "step": 21860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.948074224544614e-05,
+      "loss": 1.2656,
+      "step": 21870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.947147306022852e-05,
+      "loss": 1.2311,
+      "step": 21880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.946220027351445e-05,
+      "loss": 1.3701,
+      "step": 21890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.945292388615001e-05,
+      "loss": 1.3113,
+      "step": 21900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.944364389898158e-05,
+      "loss": 1.548,
+      "step": 21910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.943436031285594e-05,
+      "loss": 1.1314,
+      "step": 21920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.942507312862012e-05,
+      "loss": 1.1341,
+      "step": 21930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.941578234712152e-05,
+      "loss": 1.3171,
+      "step": 21940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.940648796920784e-05,
+      "loss": 1.1372,
+      "step": 21950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.939718999572714e-05,
+      "loss": 1.2167,
+      "step": 21960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.938788842752784e-05,
+      "loss": 1.3607,
+      "step": 21970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.937858326545857e-05,
+      "loss": 1.2811,
+      "step": 21980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.936927451036841e-05,
+      "loss": 1.2619,
+      "step": 21990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.93599621631067e-05,
+      "loss": 1.3984,
+      "step": 22000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.93506462245231e-05,
+      "loss": 1.1479,
+      "step": 22010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.934132669546769e-05,
+      "loss": 1.1322,
+      "step": 22020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.933200357679075e-05,
+      "loss": 1.0771,
+      "step": 22030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.932267686934299e-05,
+      "loss": 1.2705,
+      "step": 22040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.931334657397539e-05,
+      "loss": 1.0712,
+      "step": 22050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.930401269153928e-05,
+      "loss": 1.3846,
+      "step": 22060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.92946752228863e-05,
+      "loss": 1.0875,
+      "step": 22070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.928533416886844e-05,
+      "loss": 1.2723,
+      "step": 22080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.9275989530338e-05,
+      "loss": 1.334,
+      "step": 22090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.926664130814761e-05,
+      "loss": 1.1721,
+      "step": 22100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.925728950315026e-05,
+      "loss": 1.2566,
+      "step": 22110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.92479341161992e-05,
+      "loss": 1.3477,
+      "step": 22120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.923857514814804e-05,
+      "loss": 1.332,
+      "step": 22130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.922921259985075e-05,
+      "loss": 1.2383,
+      "step": 22140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.92198464721616e-05,
+      "loss": 1.1081,
+      "step": 22150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.921047676593514e-05,
+      "loss": 1.1657,
+      "step": 22160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.920110348202635e-05,
+      "loss": 1.1711,
+      "step": 22170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.919172662129042e-05,
+      "loss": 1.2467,
+      "step": 22180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.918234618458294e-05,
+      "loss": 1.1162,
+      "step": 22190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.917296217275985e-05,
+      "loss": 1.2301,
+      "step": 22200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.916357458667732e-05,
+      "loss": 1.3358,
+      "step": 22210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.915418342719193e-05,
+      "loss": 1.1441,
+      "step": 22220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.914478869516056e-05,
+      "loss": 1.1867,
+      "step": 22230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.91353903914404e-05,
+      "loss": 1.2155,
+      "step": 22240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.912598851688898e-05,
+      "loss": 1.3128,
+      "step": 22250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.911658307236417e-05,
+      "loss": 1.2874,
+      "step": 22260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.910717405872415e-05,
+      "loss": 1.336,
+      "step": 22270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.90977614768274e-05,
+      "loss": 1.1446,
+      "step": 22280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.908834532753279e-05,
+      "loss": 1.0065,
+      "step": 22290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.907892561169946e-05,
+      "loss": 1.0934,
+      "step": 22300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.906950233018687e-05,
+      "loss": 1.2727,
+      "step": 22310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.906007548385488e-05,
+      "loss": 1.0686,
+      "step": 22320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.905064507356357e-05,
+      "loss": 1.3194,
+      "step": 22330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.904121110017345e-05,
+      "loss": 1.2063,
+      "step": 22340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.903177356454526e-05,
+      "loss": 1.2497,
+      "step": 22350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.902233246754015e-05,
+      "loss": 1.1947,
+      "step": 22360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.90128878100195e-05,
+      "loss": 1.3794,
+      "step": 22370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.900343959284515e-05,
+      "loss": 1.3253,
+      "step": 22380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.899398781687908e-05,
+      "loss": 1.4376,
+      "step": 22390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.89845324829838e-05,
+      "loss": 1.2154,
+      "step": 22400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.897507359202197e-05,
+      "loss": 1.2786,
+      "step": 22410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.896561114485669e-05,
+      "loss": 1.3624,
+      "step": 22420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.895614514235131e-05,
+      "loss": 1.3831,
+      "step": 22430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.894667558536956e-05,
+      "loss": 1.253,
+      "step": 22440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.893720247477548e-05,
+      "loss": 1.2499,
+      "step": 22450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.89277258114334e-05,
+      "loss": 1.3783,
+      "step": 22460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.891824559620801e-05,
+      "loss": 1.3984,
+      "step": 22470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.89087618299643e-05,
+      "loss": 1.3286,
+      "step": 22480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.889927451356763e-05,
+      "loss": 1.3391,
+      "step": 22490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.888978364788363e-05,
+      "loss": 1.1588,
+      "step": 22500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.888028923377827e-05,
+      "loss": 1.4793,
+      "step": 22510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.887079127211786e-05,
+      "loss": 1.1875,
+      "step": 22520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.886128976376901e-05,
+      "loss": 1.3139,
+      "step": 22530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.88517847095987e-05,
+      "loss": 1.1123,
+      "step": 22540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.884227611047415e-05,
+      "loss": 1.2371,
+      "step": 22550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.883276396726299e-05,
+      "loss": 1.2583,
+      "step": 22560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.882324828083312e-05,
+      "loss": 1.4052,
+      "step": 22570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.88137290520528e-05,
+      "loss": 1.224,
+      "step": 22580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.880420628179056e-05,
+      "loss": 1.216,
+      "step": 22590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.879467997091531e-05,
+      "loss": 1.0818,
+      "step": 22600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.878515012029626e-05,
+      "loss": 1.4376,
+      "step": 22610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.877561673080293e-05,
+      "loss": 1.259,
+      "step": 22620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.87660798033052e-05,
+      "loss": 1.328,
+      "step": 22630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.875653933867321e-05,
+      "loss": 1.2694,
+      "step": 22640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.874699533777748e-05,
+      "loss": 1.2843,
+      "step": 22650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.873744780148884e-05,
+      "loss": 1.3297,
+      "step": 22660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.872789673067843e-05,
+      "loss": 1.2643,
+      "step": 22670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.871834212621772e-05,
+      "loss": 1.3479,
+      "step": 22680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.87087839889785e-05,
+      "loss": 1.1739,
+      "step": 22690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.869922231983287e-05,
+      "loss": 1.247,
+      "step": 22700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.868965711965332e-05,
+      "loss": 1.1796,
+      "step": 22710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.868008838931252e-05,
+      "loss": 1.6035,
+      "step": 22720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.867051612968362e-05,
+      "loss": 1.4405,
+      "step": 22730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.866094034163998e-05,
+      "loss": 1.3115,
+      "step": 22740
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.865136102605535e-05,
+      "loss": 1.1976,
+      "step": 22750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.864177818380378e-05,
+      "loss": 1.3366,
+      "step": 22760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.863219181575961e-05,
+      "loss": 1.2898,
+      "step": 22770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.862260192279757e-05,
+      "loss": 1.4552,
+      "step": 22780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.861300850579261e-05,
+      "loss": 1.2881,
+      "step": 22790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.860341156562012e-05,
+      "loss": 1.4083,
+      "step": 22800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.859381110315572e-05,
+      "loss": 1.2931,
+      "step": 22810
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.85842071192754e-05,
+      "loss": 1.2471,
+      "step": 22820
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.857459961485546e-05,
+      "loss": 1.3725,
+      "step": 22830
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.856498859077252e-05,
+      "loss": 1.2849,
+      "step": 22840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.855537404790349e-05,
+      "loss": 1.3101,
+      "step": 22850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.854575598712564e-05,
+      "loss": 1.2674,
+      "step": 22860
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.853613440931658e-05,
+      "loss": 1.3122,
+      "step": 22870
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.85265093153542e-05,
+      "loss": 1.3803,
+      "step": 22880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.85168807061167e-05,
+      "loss": 1.2958,
+      "step": 22890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.850724858248264e-05,
+      "loss": 1.3293,
+      "step": 22900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.849761294533087e-05,
+      "loss": 1.3776,
+      "step": 22910
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.848797379554062e-05,
+      "loss": 1.1595,
+      "step": 22920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.847833113399134e-05,
+      "loss": 1.1645,
+      "step": 22930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.846868496156287e-05,
+      "loss": 1.3574,
+      "step": 22940
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.845903527913537e-05,
+      "loss": 1.3097,
+      "step": 22950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.84493820875893e-05,
+      "loss": 1.1649,
+      "step": 22960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.843972538780546e-05,
+      "loss": 1.2799,
+      "step": 22970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.843006518066493e-05,
+      "loss": 1.2892,
+      "step": 22980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.842040146704915e-05,
+      "loss": 1.6626,
+      "step": 22990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.841073424783987e-05,
+      "loss": 1.1009,
+      "step": 23000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.840106352391915e-05,
+      "loss": 1.2724,
+      "step": 23010
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.839138929616939e-05,
+      "loss": 1.5149,
+      "step": 23020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.838171156547327e-05,
+      "loss": 1.3218,
+      "step": 23030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.837203033271384e-05,
+      "loss": 1.235,
+      "step": 23040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.836234559877442e-05,
+      "loss": 1.3536,
+      "step": 23050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.83526573645387e-05,
+      "loss": 1.4914,
+      "step": 23060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.834296563089064e-05,
+      "loss": 1.1581,
+      "step": 23070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.833327039871455e-05,
+      "loss": 1.2058,
+      "step": 23080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.832357166889507e-05,
+      "loss": 1.4767,
+      "step": 23090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.831386944231714e-05,
+      "loss": 1.4766,
+      "step": 23100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.830416371986599e-05,
+      "loss": 1.1972,
+      "step": 23110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.829445450242722e-05,
+      "loss": 1.4058,
+      "step": 23120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.828474179088676e-05,
+      "loss": 1.2726,
+      "step": 23130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.827502558613074e-05,
+      "loss": 1.3691,
+      "step": 23140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.826530588904579e-05,
+      "loss": 1.321,
+      "step": 23150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.825558270051871e-05,
+      "loss": 1.2878,
+      "step": 23160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.824585602143668e-05,
+      "loss": 1.4032,
+      "step": 23170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.823612585268723e-05,
+      "loss": 1.2364,
+      "step": 23180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.822639219515811e-05,
+      "loss": 1.3157,
+      "step": 23190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.82166550497375e-05,
+      "loss": 1.2563,
+      "step": 23200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.820691441731382e-05,
+      "loss": 1.5744,
+      "step": 23210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.819717029877581e-05,
+      "loss": 1.3628,
+      "step": 23220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.81874226950126e-05,
+      "loss": 1.2474,
+      "step": 23230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.817767160691358e-05,
+      "loss": 1.4045,
+      "step": 23240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.816791703536847e-05,
+      "loss": 1.3834,
+      "step": 23250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.815815898126728e-05,
+      "loss": 1.2779,
+      "step": 23260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.814839744550039e-05,
+      "loss": 1.2073,
+      "step": 23270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.813863242895846e-05,
+      "loss": 0.9973,
+      "step": 23280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.812886393253248e-05,
+      "loss": 1.2626,
+      "step": 23290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.811909195711376e-05,
+      "loss": 1.4005,
+      "step": 23300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.810931650359394e-05,
+      "loss": 1.2073,
+      "step": 23310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.809953757286494e-05,
+      "loss": 1.2985,
+      "step": 23320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.808975516581903e-05,
+      "loss": 1.3684,
+      "step": 23330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.807996928334878e-05,
+      "loss": 1.2892,
+      "step": 23340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.807017992634708e-05,
+      "loss": 1.2638,
+      "step": 23350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.806038709570717e-05,
+      "loss": 1.2908,
+      "step": 23360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.805059079232253e-05,
+      "loss": 1.2028,
+      "step": 23370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.804079101708704e-05,
+      "loss": 1.2889,
+      "step": 23380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.803098777089485e-05,
+      "loss": 1.2676,
+      "step": 23390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.802118105464044e-05,
+      "loss": 1.3761,
+      "step": 23400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.801137086921858e-05,
+      "loss": 1.1783,
+      "step": 23410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.800155721552443e-05,
+      "loss": 1.4158,
+      "step": 23420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.799174009445338e-05,
+      "loss": 1.2277,
+      "step": 23430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.798191950690117e-05,
+      "loss": 1.2096,
+      "step": 23440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.79720954537639e-05,
+      "loss": 1.2766,
+      "step": 23450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.796226793593792e-05,
+      "loss": 1.3304,
+      "step": 23460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.795243695431991e-05,
+      "loss": 1.2767,
+      "step": 23470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.794260250980688e-05,
+      "loss": 1.1089,
+      "step": 23480
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.793276460329619e-05,
+      "loss": 1.2314,
+      "step": 23490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.792292323568545e-05,
+      "loss": 1.2272,
+      "step": 23500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.791307840787261e-05,
+      "loss": 1.2688,
+      "step": 23510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.790323012075596e-05,
+      "loss": 1.2017,
+      "step": 23520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.78933783752341e-05,
+      "loss": 1.2572,
+      "step": 23530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.78835231722059e-05,
+      "loss": 1.5967,
+      "step": 23540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.78736645125706e-05,
+      "loss": 1.2391,
+      "step": 23550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.786380239722772e-05,
+      "loss": 1.3014,
+      "step": 23560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.785393682707711e-05,
+      "loss": 1.2119,
+      "step": 23570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.784406780301894e-05,
+      "loss": 1.3249,
+      "step": 23580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.783419532595372e-05,
+      "loss": 1.2792,
+      "step": 23590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.782431939678218e-05,
+      "loss": 1.2616,
+      "step": 23600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.781444001640548e-05,
+      "loss": 1.2989,
+      "step": 23610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.780455718572502e-05,
+      "loss": 1.1899,
+      "step": 23620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.779467090564256e-05,
+      "loss": 1.3393,
+      "step": 23630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.778478117706014e-05,
+      "loss": 1.2664,
+      "step": 23640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.777488800088013e-05,
+      "loss": 1.1082,
+      "step": 23650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.776499137800522e-05,
+      "loss": 1.456,
+      "step": 23660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.775509130933839e-05,
+      "loss": 1.1701,
+      "step": 23670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.774518779578297e-05,
+      "loss": 1.2321,
+      "step": 23680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.773528083824258e-05,
+      "loss": 1.1775,
+      "step": 23690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.772537043762117e-05,
+      "loss": 1.2802,
+      "step": 23700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.771545659482298e-05,
+      "loss": 1.2145,
+      "step": 23710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.770553931075259e-05,
+      "loss": 1.3986,
+      "step": 23720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.769561858631488e-05,
+      "loss": 1.108,
+      "step": 23730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.768569442241504e-05,
+      "loss": 1.3167,
+      "step": 23740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.767576681995858e-05,
+      "loss": 1.3237,
+      "step": 23750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.766583577985134e-05,
+      "loss": 1.1841,
+      "step": 23760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.765590130299944e-05,
+      "loss": 1.2273,
+      "step": 23770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.764596339030935e-05,
+      "loss": 1.3044,
+      "step": 23780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.763602204268781e-05,
+      "loss": 1.4951,
+      "step": 23790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.762607726104193e-05,
+      "loss": 1.3126,
+      "step": 23800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.76161290462791e-05,
+      "loss": 1.2374,
+      "step": 23810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.760617739930698e-05,
+      "loss": 1.0595,
+      "step": 23820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.759622232103363e-05,
+      "loss": 1.517,
+      "step": 23830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.758626381236738e-05,
+      "loss": 1.3972,
+      "step": 23840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.757630187421686e-05,
+      "loss": 1.4081,
+      "step": 23850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.756633650749102e-05,
+      "loss": 1.3255,
+      "step": 23860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.755636771309916e-05,
+      "loss": 1.2636,
+      "step": 23870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.754639549195083e-05,
+      "loss": 1.2255,
+      "step": 23880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.753641984495596e-05,
+      "loss": 1.2831,
+      "step": 23890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.752644077302472e-05,
+      "loss": 1.1728,
+      "step": 23900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.751645827706767e-05,
+      "loss": 1.1858,
+      "step": 23910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.750647235799561e-05,
+      "loss": 1.4628,
+      "step": 23920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.749648301671971e-05,
+      "loss": 1.3749,
+      "step": 23930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.748649025415141e-05,
+      "loss": 1.3737,
+      "step": 23940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.747649407120248e-05,
+      "loss": 1.1647,
+      "step": 23950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.746649446878501e-05,
+      "loss": 1.234,
+      "step": 23960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.74564914478114e-05,
+      "loss": 1.2896,
+      "step": 23970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.744648500919435e-05,
+      "loss": 1.1208,
+      "step": 23980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.743647515384686e-05,
+      "loss": 1.3439,
+      "step": 23990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.742646188268229e-05,
+      "loss": 1.2527,
+      "step": 24000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.741644519661425e-05,
+      "loss": 1.0824,
+      "step": 24010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.740642509655672e-05,
+      "loss": 1.2831,
+      "step": 24020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.739640158342395e-05,
+      "loss": 1.339,
+      "step": 24030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.738637465813054e-05,
+      "loss": 1.1319,
+      "step": 24040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.737634432159132e-05,
+      "loss": 1.1888,
+      "step": 24050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.736631057472153e-05,
+      "loss": 1.2445,
+      "step": 24060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.735627341843669e-05,
+      "loss": 1.3269,
+      "step": 24070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.734623285365259e-05,
+      "loss": 1.3556,
+      "step": 24080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.73361888812854e-05,
+      "loss": 1.2679,
+      "step": 24090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.732614150225151e-05,
+      "loss": 1.333,
+      "step": 24100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.731609071746772e-05,
+      "loss": 1.1531,
+      "step": 24110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.730603652785105e-05,
+      "loss": 1.3322,
+      "step": 24120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.729597893431894e-05,
+      "loss": 1.4441,
+      "step": 24130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.7285917937789e-05,
+      "loss": 1.4038,
+      "step": 24140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.727585353917927e-05,
+      "loss": 1.2727,
+      "step": 24150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.726578573940804e-05,
+      "loss": 1.2898,
+      "step": 24160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.725571453939394e-05,
+      "loss": 1.484,
+      "step": 24170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.724563994005588e-05,
+      "loss": 1.4288,
+      "step": 24180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.723556194231313e-05,
+      "loss": 1.1676,
+      "step": 24190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.72254805470852e-05,
+      "loss": 1.3025,
+      "step": 24200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.721539575529197e-05,
+      "loss": 1.2361,
+      "step": 24210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.720530756785358e-05,
+      "loss": 1.225,
+      "step": 24220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.719521598569052e-05,
+      "loss": 1.2299,
+      "step": 24230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.71851210097236e-05,
+      "loss": 1.3079,
+      "step": 24240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.71750226408739e-05,
+      "loss": 1.1053,
+      "step": 24250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.716492088006279e-05,
+      "loss": 0.978,
+      "step": 24260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.715481572821204e-05,
+      "loss": 1.6357,
+      "step": 24270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.714470718624365e-05,
+      "loss": 1.3841,
+      "step": 24280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.713459525507996e-05,
+      "loss": 1.1877,
+      "step": 24290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.712447993564361e-05,
+      "loss": 1.2383,
+      "step": 24300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.711436122885754e-05,
+      "loss": 1.1688,
+      "step": 24310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.710423913564503e-05,
+      "loss": 1.3272,
+      "step": 24320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.709411365692964e-05,
+      "loss": 1.2633,
+      "step": 24330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.708398479363526e-05,
+      "loss": 1.2994,
+      "step": 24340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.707385254668606e-05,
+      "loss": 1.1254,
+      "step": 24350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.706371691700657e-05,
+      "loss": 1.2881,
+      "step": 24360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.705357790552154e-05,
+      "loss": 1.2455,
+      "step": 24370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.704343551315614e-05,
+      "loss": 1.3218,
+      "step": 24380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.703328974083575e-05,
+      "loss": 1.27,
+      "step": 24390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.702314058948614e-05,
+      "loss": 1.2219,
+      "step": 24400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.701298806003331e-05,
+      "loss": 1.2305,
+      "step": 24410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.700283215340366e-05,
+      "loss": 1.2769,
+      "step": 24420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.699267287052377e-05,
+      "loss": 1.363,
+      "step": 24430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.698251021232066e-05,
+      "loss": 1.1918,
+      "step": 24440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.697234417972161e-05,
+      "loss": 1.1066,
+      "step": 24450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.696217477365417e-05,
+      "loss": 1.2102,
+      "step": 24460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.695200199504622e-05,
+      "loss": 1.2908,
+      "step": 24470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.694182584482598e-05,
+      "loss": 1.3019,
+      "step": 24480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.693164632392194e-05,
+      "loss": 1.3206,
+      "step": 24490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.69214634332629e-05,
+      "loss": 1.2205,
+      "step": 24500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.691127717377801e-05,
+      "loss": 1.1099,
+      "step": 24510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.690108754639666e-05,
+      "loss": 1.2325,
+      "step": 24520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.689089455204862e-05,
+      "loss": 1.4773,
+      "step": 24530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.688069819166389e-05,
+      "loss": 1.3784,
+      "step": 24540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.687049846617284e-05,
+      "loss": 1.2263,
+      "step": 24550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.686029537650611e-05,
+      "loss": 1.2377,
+      "step": 24560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.685008892359468e-05,
+      "loss": 1.246,
+      "step": 24570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.683987910836979e-05,
+      "loss": 1.4302,
+      "step": 24580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.682966593176305e-05,
+      "loss": 1.4223,
+      "step": 24590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.681944939470631e-05,
+      "loss": 1.3111,
+      "step": 24600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.680922949813178e-05,
+      "loss": 1.3293,
+      "step": 24610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.679900624297193e-05,
+      "loss": 1.2571,
+      "step": 24620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.678877963015957e-05,
+      "loss": 1.3261,
+      "step": 24630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.677854966062783e-05,
+      "loss": 1.2469,
+      "step": 24640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.67683163353101e-05,
+      "loss": 1.0586,
+      "step": 24650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.67580796551401e-05,
+      "loss": 1.1861,
+      "step": 24660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.674783962105186e-05,
+      "loss": 1.4956,
+      "step": 24670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.67375962339797e-05,
+      "loss": 1.302,
+      "step": 24680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.672734949485829e-05,
+      "loss": 1.4479,
+      "step": 24690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.671709940462253e-05,
+      "loss": 1.1814,
+      "step": 24700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.670684596420772e-05,
+      "loss": 1.1177,
+      "step": 24710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.669658917454937e-05,
+      "loss": 1.3226,
+      "step": 24720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.668632903658336e-05,
+      "loss": 1.2454,
+      "step": 24730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.667606555124585e-05,
+      "loss": 1.2153,
+      "step": 24740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.666579871947333e-05,
+      "loss": 1.0666,
+      "step": 24750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.665552854220253e-05,
+      "loss": 1.4203,
+      "step": 24760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.664525502037059e-05,
+      "loss": 1.2008,
+      "step": 24770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.663497815491486e-05,
+      "loss": 1.303,
+      "step": 24780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.662469794677306e-05,
+      "loss": 1.2779,
+      "step": 24790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.661441439688315e-05,
+      "loss": 1.0151,
+      "step": 24800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.660412750618345e-05,
+      "loss": 1.3339,
+      "step": 24810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.659383727561259e-05,
+      "loss": 1.2459,
+      "step": 24820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.658354370610944e-05,
+      "loss": 1.0281,
+      "step": 24830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.657324679861326e-05,
+      "loss": 1.3584,
+      "step": 24840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.656294655406351e-05,
+      "loss": 1.3592,
+      "step": 24850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.655264297340009e-05,
+      "loss": 1.3096,
+      "step": 24860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.654233605756308e-05,
+      "loss": 1.1819,
+      "step": 24870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.653202580749292e-05,
+      "loss": 1.2104,
+      "step": 24880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.652171222413037e-05,
+      "loss": 1.2947,
+      "step": 24890
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.651139530841645e-05,
+      "loss": 1.1817,
+      "step": 24900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.650107506129253e-05,
+      "loss": 1.2528,
+      "step": 24910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.649075148370021e-05,
+      "loss": 1.1333,
+      "step": 24920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.648042457658152e-05,
+      "loss": 1.2834,
+      "step": 24930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.647009434087865e-05,
+      "loss": 1.3249,
+      "step": 24940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.64597607775342e-05,
+      "loss": 1.4387,
+      "step": 24950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.644942388749102e-05,
+      "loss": 1.2659,
+      "step": 24960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.64390836716923e-05,
+      "loss": 1.1693,
+      "step": 24970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.642874013108148e-05,
+      "loss": 1.2862,
+      "step": 24980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.641839326660237e-05,
+      "loss": 1.2829,
+      "step": 24990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.640804307919902e-05,
+      "loss": 1.2455,
+      "step": 25000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.639768956981583e-05,
+      "loss": 1.1734,
+      "step": 25010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.638733273939748e-05,
+      "loss": 1.3274,
+      "step": 25020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.637697258888896e-05,
+      "loss": 1.2912,
+      "step": 25030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.636660911923555e-05,
+      "loss": 1.2205,
+      "step": 25040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.635624233138286e-05,
+      "loss": 1.3405,
+      "step": 25050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.634587222627679e-05,
+      "loss": 1.263,
+      "step": 25060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.633549880486351e-05,
+      "loss": 1.3308,
+      "step": 25070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.632512206808958e-05,
+      "loss": 1.3356,
+      "step": 25080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.631474201690173e-05,
+      "loss": 1.2657,
+      "step": 25090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.630435865224713e-05,
+      "loss": 1.2769,
+      "step": 25100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.629397197507315e-05,
+      "loss": 1.2632,
+      "step": 25110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.628358198632753e-05,
+      "loss": 1.2584,
+      "step": 25120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.627318868695826e-05,
+      "loss": 1.2567,
+      "step": 25130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.626279207791367e-05,
+      "loss": 1.2489,
+      "step": 25140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.625239216014239e-05,
+      "loss": 1.2256,
+      "step": 25150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.62419889345933e-05,
+      "loss": 1.6258,
+      "step": 25160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.623158240221566e-05,
+      "loss": 1.2184,
+      "step": 25170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.622117256395896e-05,
+      "loss": 1.2458,
+      "step": 25180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.621075942077307e-05,
+      "loss": 1.1242,
+      "step": 25190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.620034297360809e-05,
+      "loss": 1.3397,
+      "step": 25200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.618992322341444e-05,
+      "loss": 1.0229,
+      "step": 25210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.617950017114284e-05,
+      "loss": 1.3809,
+      "step": 25220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.616907381774438e-05,
+      "loss": 1.336,
+      "step": 25230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.615864416417033e-05,
+      "loss": 1.2924,
+      "step": 25240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.614821121137236e-05,
+      "loss": 1.3836,
+      "step": 25250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.613777496030237e-05,
+      "loss": 1.3497,
+      "step": 25260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.612733541191262e-05,
+      "loss": 1.3889,
+      "step": 25270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.611689256715564e-05,
+      "loss": 1.2354,
+      "step": 25280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.610644642698427e-05,
+      "loss": 1.0807,
+      "step": 25290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.609599699235164e-05,
+      "loss": 1.2515,
+      "step": 25300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.60855442642112e-05,
+      "loss": 1.2451,
+      "step": 25310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.607508824351669e-05,
+      "loss": 1.2102,
+      "step": 25320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.606462893122212e-05,
+      "loss": 1.0768,
+      "step": 25330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.605416632828185e-05,
+      "loss": 1.2183,
+      "step": 25340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.604370043565054e-05,
+      "loss": 1.2816,
+      "step": 25350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.603323125428309e-05,
+      "loss": 1.3463,
+      "step": 25360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.602275878513475e-05,
+      "loss": 1.2908,
+      "step": 25370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.601228302916109e-05,
+      "loss": 1.2678,
+      "step": 25380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.60018039873179e-05,
+      "loss": 1.3399,
+      "step": 25390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.599132166056137e-05,
+      "loss": 1.327,
+      "step": 25400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.598083604984791e-05,
+      "loss": 1.2336,
+      "step": 25410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.597034715613426e-05,
+      "loss": 1.2365,
+      "step": 25420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.595985498037748e-05,
+      "loss": 1.6296,
+      "step": 25430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.594935952353487e-05,
+      "loss": 1.2119,
+      "step": 25440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.593886078656409e-05,
+      "loss": 1.3625,
+      "step": 25450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.592835877042309e-05,
+      "loss": 1.1275,
+      "step": 25460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.591785347607007e-05,
+      "loss": 1.368,
+      "step": 25470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.59073449044636e-05,
+      "loss": 1.1099,
+      "step": 25480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.589683305656249e-05,
+      "loss": 1.4078,
+      "step": 25490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.588631793332589e-05,
+      "loss": 1.3117,
+      "step": 25500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.58757995357132e-05,
+      "loss": 1.4631,
+      "step": 25510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.586527786468422e-05,
+      "loss": 1.3648,
+      "step": 25520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.58547529211989e-05,
+      "loss": 1.4038,
+      "step": 25530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.58442247062176e-05,
+      "loss": 1.2978,
+      "step": 25540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.583369322070094e-05,
+      "loss": 1.2655,
+      "step": 25550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.582315846560986e-05,
+      "loss": 1.2609,
+      "step": 25560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.581262044190559e-05,
+      "loss": 1.2292,
+      "step": 25570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.580207915054962e-05,
+      "loss": 1.2096,
+      "step": 25580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.579153459250377e-05,
+      "loss": 1.2939,
+      "step": 25590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.578098676873019e-05,
+      "loss": 1.4657,
+      "step": 25600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.577043568019128e-05,
+      "loss": 1.1646,
+      "step": 25610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.575988132784975e-05,
+      "loss": 1.2445,
+      "step": 25620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.574932371266859e-05,
+      "loss": 1.1337,
+      "step": 25630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.573876283561115e-05,
+      "loss": 1.0884,
+      "step": 25640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.572819869764101e-05,
+      "loss": 1.3262,
+      "step": 25650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.571763129972209e-05,
+      "loss": 1.2143,
+      "step": 25660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.570706064281856e-05,
+      "loss": 1.3505,
+      "step": 25670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.569648672789497e-05,
+      "loss": 1.145,
+      "step": 25680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.568590955591606e-05,
+      "loss": 1.2317,
+      "step": 25690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.567532912784695e-05,
+      "loss": 1.3809,
+      "step": 25700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.566474544465304e-05,
+      "loss": 1.251,
+      "step": 25710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.56541585073e-05,
+      "loss": 1.501,
+      "step": 25720
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.564356831675383e-05,
+      "loss": 1.189,
+      "step": 25730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.563297487398079e-05,
+      "loss": 1.3831,
+      "step": 25740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.562237817994747e-05,
+      "loss": 1.366,
+      "step": 25750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.561177823562074e-05,
+      "loss": 1.3184,
+      "step": 25760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.560117504196779e-05,
+      "loss": 1.3487,
+      "step": 25770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.559056859995607e-05,
+      "loss": 1.2711,
+      "step": 25780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.557995891055333e-05,
+      "loss": 1.163,
+      "step": 25790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.556934597472766e-05,
+      "loss": 1.1446,
+      "step": 25800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.555872979344739e-05,
+      "loss": 1.3401,
+      "step": 25810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.55481103676812e-05,
+      "loss": 1.3432,
+      "step": 25820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.553748769839801e-05,
+      "loss": 1.39,
+      "step": 25830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.552686178656708e-05,
+      "loss": 1.3649,
+      "step": 25840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.551623263315796e-05,
+      "loss": 1.2359,
+      "step": 25850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.550560023914047e-05,
+      "loss": 1.3131,
+      "step": 25860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.549496460548474e-05,
+      "loss": 1.3377,
+      "step": 25870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.54843257331612e-05,
+      "loss": 1.2271,
+      "step": 25880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.547368362314059e-05,
+      "loss": 1.3565,
+      "step": 25890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.54630382763939e-05,
+      "loss": 1.4663,
+      "step": 25900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.545238969389247e-05,
+      "loss": 1.5225,
+      "step": 25910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.544173787660788e-05,
+      "loss": 1.1,
+      "step": 25920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.543108282551205e-05,
+      "loss": 1.2345,
+      "step": 25930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.542042454157719e-05,
+      "loss": 1.3981,
+      "step": 25940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.540976302577579e-05,
+      "loss": 1.2563,
+      "step": 25950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.539909827908063e-05,
+      "loss": 1.3372,
+      "step": 25960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.538843030246479e-05,
+      "loss": 1.2642,
+      "step": 25970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.537775909690166e-05,
+      "loss": 1.3084,
+      "step": 25980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.53670846633649e-05,
+      "loss": 1.3232,
+      "step": 25990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.53564070028285e-05,
+      "loss": 1.3055,
+      "step": 26000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.53457261162667e-05,
+      "loss": 1.1639,
+      "step": 26010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.533504200465405e-05,
+      "loss": 1.248,
+      "step": 26020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.532435466896545e-05,
+      "loss": 1.1644,
+      "step": 26030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.531366411017598e-05,
+      "loss": 1.3173,
+      "step": 26040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.53029703292611e-05,
+      "loss": 1.341,
+      "step": 26050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.529227332719657e-05,
+      "loss": 1.3811,
+      "step": 26060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.528157310495839e-05,
+      "loss": 1.2061,
+      "step": 26070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.527086966352289e-05,
+      "loss": 1.3943,
+      "step": 26080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.526016300386669e-05,
+      "loss": 1.3149,
+      "step": 26090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.524945312696668e-05,
+      "loss": 1.3332,
+      "step": 26100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.523874003380008e-05,
+      "loss": 1.2111,
+      "step": 26110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.522802372534438e-05,
+      "loss": 1.2866,
+      "step": 26120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.521730420257735e-05,
+      "loss": 1.2371,
+      "step": 26130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.52065814664771e-05,
+      "loss": 1.2803,
+      "step": 26140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.519585551802198e-05,
+      "loss": 1.2091,
+      "step": 26150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.518512635819068e-05,
+      "loss": 1.2132,
+      "step": 26160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.517439398796214e-05,
+      "loss": 1.2747,
+      "step": 26170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.516365840831565e-05,
+      "loss": 1.1185,
+      "step": 26180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.515291962023072e-05,
+      "loss": 1.3429,
+      "step": 26190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.51421776246872e-05,
+      "loss": 1.163,
+      "step": 26200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.51314324226652e-05,
+      "loss": 1.3672,
+      "step": 26210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.51206840151452e-05,
+      "loss": 1.1565,
+      "step": 26220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.510993240310788e-05,
+      "loss": 1.3848,
+      "step": 26230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.509917758753423e-05,
+      "loss": 1.383,
+      "step": 26240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.50884195694056e-05,
+      "loss": 1.3075,
+      "step": 26250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.507765834970355e-05,
+      "loss": 1.1299,
+      "step": 26260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.506689392940998e-05,
+      "loss": 1.4614,
+      "step": 26270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.505612630950706e-05,
+      "loss": 1.2264,
+      "step": 26280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.504535549097727e-05,
+      "loss": 1.2389,
+      "step": 26290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.503458147480338e-05,
+      "loss": 1.416,
+      "step": 26300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.502380426196841e-05,
+      "loss": 1.373,
+      "step": 26310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.501302385345573e-05,
+      "loss": 1.3067,
+      "step": 26320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.500224025024899e-05,
+      "loss": 1.3258,
+      "step": 26330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.499145345333208e-05,
+      "loss": 1.2853,
+      "step": 26340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.498066346368926e-05,
+      "loss": 1.2557,
+      "step": 26350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.496987028230501e-05,
+      "loss": 1.1936,
+      "step": 26360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.495907391016416e-05,
+      "loss": 1.3147,
+      "step": 26370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.494827434825177e-05,
+      "loss": 1.3184,
+      "step": 26380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.493747159755325e-05,
+      "loss": 1.2931,
+      "step": 26390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.492666565905428e-05,
+      "loss": 1.3665,
+      "step": 26400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.491585653374081e-05,
+      "loss": 1.3053,
+      "step": 26410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.49050442225991e-05,
+      "loss": 1.4516,
+      "step": 26420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.489422872661571e-05,
+      "loss": 1.0048,
+      "step": 26430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.488341004677745e-05,
+      "loss": 1.2913,
+      "step": 26440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.487258818407149e-05,
+      "loss": 1.494,
+      "step": 26450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.486176313948521e-05,
+      "loss": 1.3213,
+      "step": 26460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.485093491400635e-05,
+      "loss": 1.1013,
+      "step": 26470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.48401035086229e-05,
+      "loss": 1.3415,
+      "step": 26480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.482926892432314e-05,
+      "loss": 1.3253,
+      "step": 26490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.481843116209566e-05,
+      "loss": 1.2684,
+      "step": 26500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.480759022292934e-05,
+      "loss": 1.1928,
+      "step": 26510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.479674610781331e-05,
+      "loss": 1.3124,
+      "step": 26520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.478589881773706e-05,
+      "loss": 1.275,
+      "step": 26530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.47750483536903e-05,
+      "loss": 1.4937,
+      "step": 26540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.476419471666306e-05,
+      "loss": 1.2804,
+      "step": 26550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.475333790764566e-05,
+      "loss": 1.2967,
+      "step": 26560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.474247792762873e-05,
+      "loss": 1.1503,
+      "step": 26570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.473161477760315e-05,
+      "loss": 1.2519,
+      "step": 26580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.472074845856008e-05,
+      "loss": 1.2036,
+      "step": 26590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.470987897149106e-05,
+      "loss": 1.4442,
+      "step": 26600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.469900631738779e-05,
+      "loss": 1.0919,
+      "step": 26610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.468813049724235e-05,
+      "loss": 1.1629,
+      "step": 26620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.467725151204711e-05,
+      "loss": 1.3146,
+      "step": 26630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.466636936279467e-05,
+      "loss": 1.1448,
+      "step": 26640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.465548405047794e-05,
+      "loss": 1.3973,
+      "step": 26650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.464459557609017e-05,
+      "loss": 1.2155,
+      "step": 26660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.463370394062481e-05,
+      "loss": 1.3208,
+      "step": 26670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.462280914507568e-05,
+      "loss": 1.163,
+      "step": 26680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.461191119043684e-05,
+      "loss": 1.3457,
+      "step": 26690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.460101007770265e-05,
+      "loss": 1.2833,
+      "step": 26700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.459010580786777e-05,
+      "loss": 1.1441,
+      "step": 26710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.457919838192714e-05,
+      "loss": 1.3319,
+      "step": 26720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.456828780087598e-05,
+      "loss": 1.3963,
+      "step": 26730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.455737406570979e-05,
+      "loss": 1.5014,
+      "step": 26740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.45464571774244e-05,
+      "loss": 1.1003,
+      "step": 26750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.453553713701587e-05,
+      "loss": 1.2513,
+      "step": 26760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.45246139454806e-05,
+      "loss": 1.1898,
+      "step": 26770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.451368760381526e-05,
+      "loss": 1.232,
+      "step": 26780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.450275811301678e-05,
+      "loss": 1.171,
+      "step": 26790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.449182547408243e-05,
+      "loss": 1.2793,
+      "step": 26800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.448088968800968e-05,
+      "loss": 1.4364,
+      "step": 26810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.446995075579642e-05,
+      "loss": 1.5247,
+      "step": 26820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.44590086784407e-05,
+      "loss": 1.2296,
+      "step": 26830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.444806345694092e-05,
+      "loss": 1.2471,
+      "step": 26840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.443711509229577e-05,
+      "loss": 1.2505,
+      "step": 26850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.442616358550419e-05,
+      "loss": 1.4087,
+      "step": 26860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.441520893756542e-05,
+      "loss": 1.151,
+      "step": 26870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.440425114947904e-05,
+      "loss": 1.1604,
+      "step": 26880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.439329022224483e-05,
+      "loss": 1.0405,
+      "step": 26890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.43823261568629e-05,
+      "loss": 1.2368,
+      "step": 26900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.437135895433367e-05,
+      "loss": 1.2347,
+      "step": 26910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.43603886156578e-05,
+      "loss": 1.2692,
+      "step": 26920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.434941514183627e-05,
+      "loss": 1.1797,
+      "step": 26930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.433843853387034e-05,
+      "loss": 1.2293,
+      "step": 26940
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.432745879276152e-05,
+      "loss": 1.2469,
+      "step": 26950
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.431647591951166e-05,
+      "loss": 1.4054,
+      "step": 26960
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.430548991512286e-05,
+      "loss": 1.3398,
+      "step": 26970
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.429450078059751e-05,
+      "loss": 1.2314,
+      "step": 26980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.428350851693831e-05,
+      "loss": 1.4495,
+      "step": 26990
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.427251312514822e-05,
+      "loss": 1.3226,
+      "step": 27000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.426151460623046e-05,
+      "loss": 1.4091,
+      "step": 27010
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.425051296118862e-05,
+      "loss": 1.2245,
+      "step": 27020
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.42395081910265e-05,
+      "loss": 1.4106,
+      "step": 27030
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.42285002967482e-05,
+      "loss": 1.1378,
+      "step": 27040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.421748927935813e-05,
+      "loss": 1.1341,
+      "step": 27050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.420647513986097e-05,
+      "loss": 1.1649,
+      "step": 27060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.419545787926166e-05,
+      "loss": 1.2055,
+      "step": 27070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.418443749856547e-05,
+      "loss": 1.295,
+      "step": 27080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.417341399877792e-05,
+      "loss": 1.1502,
+      "step": 27090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.416238738090484e-05,
+      "loss": 1.2067,
+      "step": 27100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.415135764595232e-05,
+      "loss": 1.1214,
+      "step": 27110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.414032479492676e-05,
+      "loss": 1.1568,
+      "step": 27120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.412928882883481e-05,
+      "loss": 1.471,
+      "step": 27130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.411824974868345e-05,
+      "loss": 1.1852,
+      "step": 27140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.41072075554799e-05,
+      "loss": 1.5221,
+      "step": 27150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.409616225023169e-05,
+      "loss": 1.1691,
+      "step": 27160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.408511383394663e-05,
+      "loss": 1.2235,
+      "step": 27170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.407406230763279e-05,
+      "loss": 1.4331,
+      "step": 27180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.406300767229857e-05,
+      "loss": 1.0593,
+      "step": 27190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.405194992895264e-05,
+      "loss": 1.3323,
+      "step": 27200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.40408890786039e-05,
+      "loss": 1.192,
+      "step": 27210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.402982512226159e-05,
+      "loss": 1.3016,
+      "step": 27220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.401875806093524e-05,
+      "loss": 1.1356,
+      "step": 27230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.400768789563463e-05,
+      "loss": 1.3031,
+      "step": 27240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.399661462736983e-05,
+      "loss": 1.2102,
+      "step": 27250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.398553825715122e-05,
+      "loss": 1.5091,
+      "step": 27260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.397445878598939e-05,
+      "loss": 1.3252,
+      "step": 27270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.396337621489532e-05,
+      "loss": 1.2352,
+      "step": 27280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.395229054488019e-05,
+      "loss": 1.2655,
+      "step": 27290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.394120177695548e-05,
+      "loss": 1.285,
+      "step": 27300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.393010991213299e-05,
+      "loss": 1.5301,
+      "step": 27310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.391901495142476e-05,
+      "loss": 1.4222,
+      "step": 27320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.390791689584314e-05,
+      "loss": 1.5654,
+      "step": 27330
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.389681574640074e-05,
+      "loss": 1.4082,
+      "step": 27340
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.388571150411046e-05,
+      "loss": 1.1596,
+      "step": 27350
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.387460416998549e-05,
+      "loss": 1.4219,
+      "step": 27360
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.386349374503927e-05,
+      "loss": 1.1008,
+      "step": 27370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.385238023028561e-05,
+      "loss": 1.335,
+      "step": 27380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.384126362673848e-05,
+      "loss": 1.3131,
+      "step": 27390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.383014393541224e-05,
+      "loss": 1.4491,
+      "step": 27400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.381902115732144e-05,
+      "loss": 1.4997,
+      "step": 27410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.380789529348099e-05,
+      "loss": 1.3243,
+      "step": 27420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.379676634490601e-05,
+      "loss": 1.2531,
+      "step": 27430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.3785634312612e-05,
+      "loss": 1.274,
+      "step": 27440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.37744991976146e-05,
+      "loss": 1.2635,
+      "step": 27450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.376336100092989e-05,
+      "loss": 1.3091,
+      "step": 27460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.37522197235741e-05,
+      "loss": 1.1116,
+      "step": 27470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.374107536656382e-05,
+      "loss": 1.2216,
+      "step": 27480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.372992793091589e-05,
+      "loss": 1.4895,
+      "step": 27490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.371877741764742e-05,
+      "loss": 1.0963,
+      "step": 27500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.370762382777584e-05,
+      "loss": 1.2577,
+      "step": 27510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.369646716231882e-05,
+      "loss": 1.1947,
+      "step": 27520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.368530742229432e-05,
+      "loss": 1.1473,
+      "step": 27530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.367414460872063e-05,
+      "loss": 1.2354,
+      "step": 27540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.366297872261623e-05,
+      "loss": 1.3187,
+      "step": 27550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.365180976499996e-05,
+      "loss": 1.1235,
+      "step": 27560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.364063773689088e-05,
+      "loss": 1.2695,
+      "step": 27570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.36294626393084e-05,
+      "loss": 1.3021,
+      "step": 27580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.361828447327214e-05,
+      "loss": 1.1396,
+      "step": 27590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.360710323980203e-05,
+      "loss": 1.452,
+      "step": 27600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.359591893991829e-05,
+      "loss": 1.3063,
+      "step": 27610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.358473157464143e-05,
+      "loss": 1.2269,
+      "step": 27620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.357354114499217e-05,
+      "loss": 1.1113,
+      "step": 27630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.35623476519916e-05,
+      "loss": 1.1943,
+      "step": 27640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.355115109666103e-05,
+      "loss": 1.2983,
+      "step": 27650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.353995148002208e-05,
+      "loss": 1.1644,
+      "step": 27660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.35287488030966e-05,
+      "loss": 1.4893,
+      "step": 27670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.35175430669068e-05,
+      "loss": 1.3958,
+      "step": 27680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.35063342724751e-05,
+      "loss": 1.2689,
+      "step": 27690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.349512242082425e-05,
+      "loss": 1.1521,
+      "step": 27700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.348390751297723e-05,
+      "loss": 1.2783,
+      "step": 27710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.347268954995731e-05,
+      "loss": 1.461,
+      "step": 27720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.34614685327881e-05,
+      "loss": 1.2763,
+      "step": 27730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.345024446249336e-05,
+      "loss": 1.1519,
+      "step": 27740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.34390173400973e-05,
+      "loss": 1.1179,
+      "step": 27750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.342778716662425e-05,
+      "loss": 1.285,
+      "step": 27760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.341655394309893e-05,
+      "loss": 1.206,
+      "step": 27770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.340531767054626e-05,
+      "loss": 1.3797,
+      "step": 27780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.339407834999147e-05,
+      "loss": 1.1791,
+      "step": 27790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.33828359824601e-05,
+      "loss": 1.1542,
+      "step": 27800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.337159056897792e-05,
+      "loss": 1.2696,
+      "step": 27810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.336034211057098e-05,
+      "loss": 1.2738,
+      "step": 27820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.334909060826566e-05,
+      "loss": 1.3142,
+      "step": 27830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.333783606308853e-05,
+      "loss": 1.3002,
+      "step": 27840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.332657847606655e-05,
+      "loss": 1.1613,
+      "step": 27850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.331531784822685e-05,
+      "loss": 1.2752,
+      "step": 27860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.33040541805969e-05,
+      "loss": 1.0929,
+      "step": 27870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.329278747420442e-05,
+      "loss": 1.2864,
+      "step": 27880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.328151773007744e-05,
+      "loss": 1.3144,
+      "step": 27890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.327024494924423e-05,
+      "loss": 1.2344,
+      "step": 27900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.325896913273335e-05,
+      "loss": 1.2351,
+      "step": 27910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.324769028157365e-05,
+      "loss": 1.3708,
+      "step": 27920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.323640839679425e-05,
+      "loss": 1.1846,
+      "step": 27930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.322512347942452e-05,
+      "loss": 1.4174,
+      "step": 27940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.321383553049416e-05,
+      "loss": 1.3728,
+      "step": 27950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.320254455103308e-05,
+      "loss": 1.1595,
+      "step": 27960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.319125054207156e-05,
+      "loss": 1.2334,
+      "step": 27970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.317995350464003e-05,
+      "loss": 1.0628,
+      "step": 27980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.316865343976932e-05,
+      "loss": 1.1752,
+      "step": 27990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.315735034849046e-05,
+      "loss": 1.088,
+      "step": 28000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.314604423183477e-05,
+      "loss": 1.4587,
+      "step": 28010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.313473509083387e-05,
+      "loss": 1.3223,
+      "step": 28020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.312342292651962e-05,
+      "loss": 1.1431,
+      "step": 28030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.311210773992421e-05,
+      "loss": 1.3173,
+      "step": 28040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.310078953208006e-05,
+      "loss": 1.0717,
+      "step": 28050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.308946830401986e-05,
+      "loss": 1.5137,
+      "step": 28060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.307814405677659e-05,
+      "loss": 1.4088,
+      "step": 28070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.306681679138355e-05,
+      "loss": 1.1914,
+      "step": 28080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.305548650887421e-05,
+      "loss": 1.2922,
+      "step": 28090
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.304415321028245e-05,
+      "loss": 1.3544,
+      "step": 28100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.303281689664229e-05,
+      "loss": 1.1383,
+      "step": 28110
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.302147756898814e-05,
+      "loss": 1.1972,
+      "step": 28120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.301013522835459e-05,
+      "loss": 1.0615,
+      "step": 28130
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.299878987577658e-05,
+      "loss": 1.2076,
+      "step": 28140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.29874415122893e-05,
+      "loss": 1.6751,
+      "step": 28150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.29760901389282e-05,
+      "loss": 1.2135,
+      "step": 28160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.2964735756729e-05,
+      "loss": 1.3364,
+      "step": 28170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.295337836672772e-05,
+      "loss": 1.5556,
+      "step": 28180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.294201796996062e-05,
+      "loss": 1.2284,
+      "step": 28190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.293065456746431e-05,
+      "loss": 1.2007,
+      "step": 28200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.291928816027557e-05,
+      "loss": 1.2953,
+      "step": 28210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.290791874943152e-05,
+      "loss": 1.0888,
+      "step": 28220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.289654633596955e-05,
+      "loss": 1.244,
+      "step": 28230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.28851709209273e-05,
+      "loss": 1.1921,
+      "step": 28240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.287379250534271e-05,
+      "loss": 1.2461,
+      "step": 28250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.286241109025398e-05,
+      "loss": 1.3144,
+      "step": 28260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.285102667669957e-05,
+      "loss": 1.1797,
+      "step": 28270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.283963926571824e-05,
+      "loss": 1.3801,
+      "step": 28280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.2828248858349e-05,
+      "loss": 1.0576,
+      "step": 28290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.281685545563117e-05,
+      "loss": 1.4676,
+      "step": 28300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.28054590586043e-05,
+      "loss": 1.2034,
+      "step": 28310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.279405966830822e-05,
+      "loss": 1.1842,
+      "step": 28320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.278265728578307e-05,
+      "loss": 1.3048,
+      "step": 28330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.277125191206923e-05,
+      "loss": 1.4345,
+      "step": 28340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.275984354820736e-05,
+      "loss": 1.4033,
+      "step": 28350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.27484321952384e-05,
+      "loss": 1.2447,
+      "step": 28360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.273701785420353e-05,
+      "loss": 1.3716,
+      "step": 28370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.272560052614424e-05,
+      "loss": 1.3972,
+      "step": 28380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.271418021210232e-05,
+      "loss": 1.146,
+      "step": 28390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.270275691311974e-05,
+      "loss": 1.2948,
+      "step": 28400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.269133063023882e-05,
+      "loss": 1.3138,
+      "step": 28410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.267990136450212e-05,
+      "loss": 1.29,
+      "step": 28420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.26684691169525e-05,
+      "loss": 1.0941,
+      "step": 28430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.265703388863306e-05,
+      "loss": 1.1417,
+      "step": 28440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.264559568058718e-05,
+      "loss": 1.2778,
+      "step": 28450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.263415449385854e-05,
+      "loss": 1.191,
+      "step": 28460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.262271032949102e-05,
+      "loss": 1.1773,
+      "step": 28470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.261126318852887e-05,
+      "loss": 1.2815,
+      "step": 28480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.259981307201655e-05,
+      "loss": 1.3834,
+      "step": 28490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.258835998099879e-05,
+      "loss": 1.2724,
+      "step": 28500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.25769039165206e-05,
+      "loss": 1.3202,
+      "step": 28510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.256544487962728e-05,
+      "loss": 1.4193,
+      "step": 28520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.255398287136438e-05,
+      "loss": 0.9349,
+      "step": 28530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.254251789277773e-05,
+      "loss": 1.3385,
+      "step": 28540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.253104994491345e-05,
+      "loss": 1.2945,
+      "step": 28550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.25195790288179e-05,
+      "loss": 1.4197,
+      "step": 28560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.250810514553768e-05,
+      "loss": 1.3701,
+      "step": 28570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.249662829611975e-05,
+      "loss": 1.2065,
+      "step": 28580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.248514848161126e-05,
+      "loss": 1.3788,
+      "step": 28590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.24736657030597e-05,
+      "loss": 1.4662,
+      "step": 28600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.246217996151277e-05,
+      "loss": 1.117,
+      "step": 28610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.245069125801846e-05,
+      "loss": 1.2402,
+      "step": 28620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.243919959362505e-05,
+      "loss": 1.2305,
+      "step": 28630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.242770496938106e-05,
+      "loss": 1.6373,
+      "step": 28640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.241620738633528e-05,
+      "loss": 1.1744,
+      "step": 28650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.240470684553683e-05,
+      "loss": 1.4429,
+      "step": 28660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.239320334803502e-05,
+      "loss": 1.5479,
+      "step": 28670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.238169689487947e-05,
+      "loss": 1.3326,
+      "step": 28680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.237018748712007e-05,
+      "loss": 1.2314,
+      "step": 28690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.235867512580694e-05,
+      "loss": 1.2692,
+      "step": 28700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.234715981199058e-05,
+      "loss": 1.287,
+      "step": 28710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.23356415467216e-05,
+      "loss": 1.2947,
+      "step": 28720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.2324120331051e-05,
+      "loss": 1.5016,
+      "step": 28730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.231259616603e-05,
+      "loss": 1.2771,
+      "step": 28740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.23010690527101e-05,
+      "loss": 1.3657,
+      "step": 28750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.228953899214308e-05,
+      "loss": 1.4173,
+      "step": 28760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.227800598538097e-05,
+      "loss": 1.2384,
+      "step": 28770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.226647003347608e-05,
+      "loss": 1.1372,
+      "step": 28780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.225493113748098e-05,
+      "loss": 1.2316,
+      "step": 28790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.22433892984485e-05,
+      "loss": 1.2086,
+      "step": 28800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.22318445174318e-05,
+      "loss": 1.2879,
+      "step": 28810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.22202967954842e-05,
+      "loss": 1.1584,
+      "step": 28820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.220874613365938e-05,
+      "loss": 1.2797,
+      "step": 28830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.219719253301128e-05,
+      "loss": 1.3394,
+      "step": 28840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.218563599459404e-05,
+      "loss": 1.2505,
+      "step": 28850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.217407651946213e-05,
+      "loss": 1.1877,
+      "step": 28860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.216251410867028e-05,
+      "loss": 1.3494,
+      "step": 28870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.215094876327348e-05,
+      "loss": 1.2865,
+      "step": 28880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.213938048432697e-05,
+      "loss": 1.253,
+      "step": 28890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.212780927288629e-05,
+      "loss": 1.0997,
+      "step": 28900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.211623513000723e-05,
+      "loss": 1.2415,
+      "step": 28910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.210465805674585e-05,
+      "loss": 1.3405,
+      "step": 28920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.209307805415848e-05,
+      "loss": 1.2628,
+      "step": 28930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.20814951233017e-05,
+      "loss": 1.4601,
+      "step": 28940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.206990926523239e-05,
+      "loss": 1.3528,
+      "step": 28950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.205832048100768e-05,
+      "loss": 1.0674,
+      "step": 28960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.204672877168493e-05,
+      "loss": 1.3054,
+      "step": 28970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.203513413832184e-05,
+      "loss": 1.3358,
+      "step": 28980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.202353658197632e-05,
+      "loss": 1.8508,
+      "step": 28990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.201193610370659e-05,
+      "loss": 1.3068,
+      "step": 29000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.200033270457109e-05,
+      "loss": 1.2477,
+      "step": 29010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.198872638562856e-05,
+      "loss": 1.1866,
+      "step": 29020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.197711714793797e-05,
+      "loss": 1.3735,
+      "step": 29030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.196550499255861e-05,
+      "loss": 1.2985,
+      "step": 29040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.195388992055002e-05,
+      "loss": 1.1644,
+      "step": 29050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.194227193297197e-05,
+      "loss": 1.1818,
+      "step": 29060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.193065103088453e-05,
+      "loss": 1.2622,
+      "step": 29070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.1919027215348e-05,
+      "loss": 1.2003,
+      "step": 29080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.1907400487423e-05,
+      "loss": 1.3377,
+      "step": 29090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.18957708481704e-05,
+      "loss": 1.2792,
+      "step": 29100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.188413829865129e-05,
+      "loss": 1.4133,
+      "step": 29110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.187250283992709e-05,
+      "loss": 1.224,
+      "step": 29120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.186086447305943e-05,
+      "loss": 1.4087,
+      "step": 29130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.184922319911023e-05,
+      "loss": 1.277,
+      "step": 29140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.183757901914168e-05,
+      "loss": 1.2031,
+      "step": 29150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.182593193421625e-05,
+      "loss": 1.2813,
+      "step": 29160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.181428194539664e-05,
+      "loss": 1.1518,
+      "step": 29170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.180262905374581e-05,
+      "loss": 1.4077,
+      "step": 29180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.179097326032703e-05,
+      "loss": 1.2569,
+      "step": 29190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.177931456620382e-05,
+      "loss": 1.3568,
+      "step": 29200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.176765297243992e-05,
+      "loss": 1.308,
+      "step": 29210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.17559884800994e-05,
+      "loss": 1.2244,
+      "step": 29220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.174432109024655e-05,
+      "loss": 1.4062,
+      "step": 29230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.173265080394593e-05,
+      "loss": 1.1777,
+      "step": 29240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.172097762226238e-05,
+      "loss": 1.2748,
+      "step": 29250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.170930154626101e-05,
+      "loss": 1.2154,
+      "step": 29260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.169762257700715e-05,
+      "loss": 1.233,
+      "step": 29270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.168594071556645e-05,
+      "loss": 1.181,
+      "step": 29280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.16742559630048e-05,
+      "loss": 1.2714,
+      "step": 29290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.166256832038833e-05,
+      "loss": 1.2402,
+      "step": 29300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.165087778878349e-05,
+      "loss": 1.2843,
+      "step": 29310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.163918436925692e-05,
+      "loss": 1.1469,
+      "step": 29320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.162748806287558e-05,
+      "loss": 1.1818,
+      "step": 29330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.161578887070668e-05,
+      "loss": 1.2172,
+      "step": 29340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.160408679381768e-05,
+      "loss": 1.2522,
+      "step": 29350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.159238183327633e-05,
+      "loss": 1.2492,
+      "step": 29360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.158067399015061e-05,
+      "loss": 1.306,
+      "step": 29370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.15689632655088e-05,
+      "loss": 1.2349,
+      "step": 29380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.155724966041939e-05,
+      "loss": 1.4653,
+      "step": 29390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.154553317595118e-05,
+      "loss": 1.1166,
+      "step": 29400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.153381381317322e-05,
+      "loss": 1.3382,
+      "step": 29410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.152209157315483e-05,
+      "loss": 1.2821,
+      "step": 29420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.151036645696556e-05,
+      "loss": 1.3065,
+      "step": 29430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.149863846567527e-05,
+      "loss": 1.415,
+      "step": 29440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.148690760035403e-05,
+      "loss": 1.1433,
+      "step": 29450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14751738620722e-05,
+      "loss": 1.2578,
+      "step": 29460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.146343725190045e-05,
+      "loss": 1.2699,
+      "step": 29470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.145169777090959e-05,
+      "loss": 1.1895,
+      "step": 29480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.143995542017084e-05,
+      "loss": 1.3623,
+      "step": 29490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.142821020075554e-05,
+      "loss": 1.4797,
+      "step": 29500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.141646211373541e-05,
+      "loss": 1.1159,
+      "step": 29510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.140471116018237e-05,
+      "loss": 1.3417,
+      "step": 29520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13929573411686e-05,
+      "loss": 1.3552,
+      "step": 29530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.138120065776655e-05,
+      "loss": 1.1364,
+      "step": 29540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.136944111104896e-05,
+      "loss": 1.0759,
+      "step": 29550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.135767870208877e-05,
+      "loss": 1.1594,
+      "step": 29560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.134591343195925e-05,
+      "loss": 1.1217,
+      "step": 29570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.133414530173389e-05,
+      "loss": 1.2963,
+      "step": 29580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.132237431248643e-05,
+      "loss": 1.5932,
+      "step": 29590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.131060046529093e-05,
+      "loss": 1.1978,
+      "step": 29600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.129882376122164e-05,
+      "loss": 1.1566,
+      "step": 29610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.128704420135312e-05,
+      "loss": 1.2011,
+      "step": 29620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.127526178676015e-05,
+      "loss": 1.057,
+      "step": 29630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12634765185178e-05,
+      "loss": 1.4486,
+      "step": 29640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.125168839770142e-05,
+      "loss": 1.2399,
+      "step": 29650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.123989742538655e-05,
+      "loss": 1.1102,
+      "step": 29660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.122810360264909e-05,
+      "loss": 1.2983,
+      "step": 29670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.121630693056509e-05,
+      "loss": 1.276,
+      "step": 29680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.120450741021094e-05,
+      "loss": 1.1742,
+      "step": 29690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.119270504266325e-05,
+      "loss": 1.3373,
+      "step": 29700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.118089982899893e-05,
+      "loss": 1.4534,
+      "step": 29710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.116909177029511e-05,
+      "loss": 1.3018,
+      "step": 29720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.115728086762917e-05,
+      "loss": 1.1947,
+      "step": 29730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11454671220788e-05,
+      "loss": 1.1674,
+      "step": 29740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.113365053472193e-05,
+      "loss": 1.1703,
+      "step": 29750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11218311066367e-05,
+      "loss": 1.3395,
+      "step": 29760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.111000883890159e-05,
+      "loss": 1.1125,
+      "step": 29770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.109818373259529e-05,
+      "loss": 1.3501,
+      "step": 29780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.108635578879674e-05,
+      "loss": 1.2262,
+      "step": 29790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.107452500858517e-05,
+      "loss": 1.233,
+      "step": 29800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.106269139304007e-05,
+      "loss": 1.2201,
+      "step": 29810
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.105085494324117e-05,
+      "loss": 1.3223,
+      "step": 29820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.103901566026843e-05,
+      "loss": 1.3127,
+      "step": 29830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.102717354520213e-05,
+      "loss": 0.8995,
+      "step": 29840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.101532859912279e-05,
+      "loss": 1.3091,
+      "step": 29850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.100348082311115e-05,
+      "loss": 1.537,
+      "step": 29860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.099163021824827e-05,
+      "loss": 1.3161,
+      "step": 29870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.097977678561542e-05,
+      "loss": 1.3172,
+      "step": 29880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.096792052629413e-05,
+      "loss": 1.1917,
+      "step": 29890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.09560614413662e-05,
+      "loss": 1.2352,
+      "step": 29900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.094419953191373e-05,
+      "loss": 1.2005,
+      "step": 29910
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.0932334799019e-05,
+      "loss": 1.215,
+      "step": 29920
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.092046724376457e-05,
+      "loss": 1.3485,
+      "step": 29930
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.090859686723331e-05,
+      "loss": 1.3266,
+      "step": 29940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.089672367050829e-05,
+      "loss": 1.3012,
+      "step": 29950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.088484765467286e-05,
+      "loss": 1.1656,
+      "step": 29960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.087296882081061e-05,
+      "loss": 1.3282,
+      "step": 29970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.086108717000542e-05,
+      "loss": 1.2474,
+      "step": 29980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.084920270334141e-05,
+      "loss": 1.3896,
+      "step": 29990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.083731542190294e-05,
+      "loss": 1.2215,
+      "step": 30000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.082542532677463e-05,
+      "loss": 1.2579,
+      "step": 30010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.08135324190414e-05,
+      "loss": 1.3407,
+      "step": 30020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.080163669978838e-05,
+      "loss": 1.26,
+      "step": 30030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.078973817010096e-05,
+      "loss": 1.169,
+      "step": 30040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.077783683106481e-05,
+      "loss": 1.4966,
+      "step": 30050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.076593268376585e-05,
+      "loss": 1.3155,
+      "step": 30060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.075402572929025e-05,
+      "loss": 1.2524,
+      "step": 30070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.07421159687244e-05,
+      "loss": 1.4385,
+      "step": 30080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.073020340315503e-05,
+      "loss": 1.2795,
+      "step": 30090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.071828803366908e-05,
+      "loss": 1.3565,
+      "step": 30100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.07063698613537e-05,
+      "loss": 1.5225,
+      "step": 30110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.069444888729638e-05,
+      "loss": 1.3163,
+      "step": 30120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.068252511258479e-05,
+      "loss": 1.2209,
+      "step": 30130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.067059853830691e-05,
+      "loss": 1.2795,
+      "step": 30140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.065866916555099e-05,
+      "loss": 1.4623,
+      "step": 30150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.064673699540543e-05,
+      "loss": 1.0715,
+      "step": 30160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.063480202895903e-05,
+      "loss": 1.2935,
+      "step": 30170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.062286426730074e-05,
+      "loss": 1.2512,
+      "step": 30180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.061092371151977e-05,
+      "loss": 1.0615,
+      "step": 30190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.059898036270565e-05,
+      "loss": 1.3225,
+      "step": 30200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.058703422194813e-05,
+      "loss": 1.2801,
+      "step": 30210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.05750852903372e-05,
+      "loss": 1.22,
+      "step": 30220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.056313356896311e-05,
+      "loss": 1.2293,
+      "step": 30230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.055117905891637e-05,
+      "loss": 1.1926,
+      "step": 30240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.053922176128777e-05,
+      "loss": 1.4466,
+      "step": 30250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.052726167716829e-05,
+      "loss": 1.365,
+      "step": 30260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.051529880764923e-05,
+      "loss": 1.313,
+      "step": 30270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.050333315382213e-05,
+      "loss": 1.2731,
+      "step": 30280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.049136471677874e-05,
+      "loss": 1.1884,
+      "step": 30290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.047939349761114e-05,
+      "loss": 1.3017,
+      "step": 30300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.046741949741157e-05,
+      "loss": 1.1294,
+      "step": 30310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.045544271727259e-05,
+      "loss": 1.3334,
+      "step": 30320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.044346315828703e-05,
+      "loss": 1.3379,
+      "step": 30330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.04314808215479e-05,
+      "loss": 1.2214,
+      "step": 30340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.041949570814852e-05,
+      "loss": 1.2562,
+      "step": 30350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.040750781918247e-05,
+      "loss": 1.0802,
+      "step": 30360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.039551715574351e-05,
+      "loss": 0.9623,
+      "step": 30370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.038352371892576e-05,
+      "loss": 1.1255,
+      "step": 30380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.037152750982351e-05,
+      "loss": 1.2509,
+      "step": 30390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.035952852953132e-05,
+      "loss": 1.2069,
+      "step": 30400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.034752677914404e-05,
+      "loss": 1.1465,
+      "step": 30410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.033552225975672e-05,
+      "loss": 1.3333,
+      "step": 30420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.03235149724647e-05,
+      "loss": 1.2588,
+      "step": 30430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.031150491836357e-05,
+      "loss": 1.2525,
+      "step": 30440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.029949209854915e-05,
+      "loss": 1.1298,
+      "step": 30450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.028747651411754e-05,
+      "loss": 1.4997,
+      "step": 30460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.027545816616507e-05,
+      "loss": 1.3049,
+      "step": 30470
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.026343705578832e-05,
+      "loss": 1.1496,
+      "step": 30480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.025141318408413e-05,
+      "loss": 1.1987,
+      "step": 30490
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.023938655214963e-05,
+      "loss": 1.3914,
+      "step": 30500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.022735716108211e-05,
+      "loss": 1.3865,
+      "step": 30510
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.021532501197923e-05,
+      "loss": 1.1831,
+      "step": 30520
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.020329010593881e-05,
+      "loss": 1.4677,
+      "step": 30530
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.019125244405893e-05,
+      "loss": 1.3643,
+      "step": 30540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.017921202743798e-05,
+      "loss": 1.1504,
+      "step": 30550
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.016716885717456e-05,
+      "loss": 1.2218,
+      "step": 30560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.015512293436749e-05,
+      "loss": 1.2422,
+      "step": 30570
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.014307426011591e-05,
+      "loss": 1.2782,
+      "step": 30580
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.013102283551917e-05,
+      "loss": 1.2293,
+      "step": 30590
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.011896866167688e-05,
+      "loss": 1.3277,
+      "step": 30600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.01069117396889e-05,
+      "loss": 1.0918,
+      "step": 30610
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.009485207065531e-05,
+      "loss": 1.3868,
+      "step": 30620
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.008278965567651e-05,
+      "loss": 1.422,
+      "step": 30630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.007072449585309e-05,
+      "loss": 1.3878,
+      "step": 30640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.005865659228592e-05,
+      "loss": 1.1369,
+      "step": 30650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.004658594607612e-05,
+      "loss": 1.3,
+      "step": 30660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.003451255832503e-05,
+      "loss": 1.1346,
+      "step": 30670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.00224364301343e-05,
+      "loss": 1.1894,
+      "step": 30680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.001035756260573e-05,
+      "loss": 1.4919,
+      "step": 30690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.999827595684149e-05,
+      "loss": 1.5689,
+      "step": 30700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.998619161394391e-05,
+      "loss": 1.2959,
+      "step": 30710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.997410453501564e-05,
+      "loss": 1.0812,
+      "step": 30720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.996201472115949e-05,
+      "loss": 1.3914,
+      "step": 30730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.994992217347861e-05,
+      "loss": 1.0835,
+      "step": 30740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.993782689307634e-05,
+      "loss": 1.2736,
+      "step": 30750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.99257288810563e-05,
+      "loss": 1.3569,
+      "step": 30760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.991362813852235e-05,
+      "loss": 1.3787,
+      "step": 30770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.990152466657858e-05,
+      "loss": 1.1633,
+      "step": 30780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.988941846632939e-05,
+      "loss": 1.2479,
+      "step": 30790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.987730953887933e-05,
+      "loss": 1.1677,
+      "step": 30800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.98651978853333e-05,
+      "loss": 1.3026,
+      "step": 30810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.985308350679638e-05,
+      "loss": 1.1655,
+      "step": 30820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.984096640437393e-05,
+      "loss": 1.4358,
+      "step": 30830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.982884657917156e-05,
+      "loss": 1.3939,
+      "step": 30840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.981672403229509e-05,
+      "loss": 1.1359,
+      "step": 30850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.980459876485065e-05,
+      "loss": 1.4047,
+      "step": 30860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.979247077794457e-05,
+      "loss": 1.4101,
+      "step": 30870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.978034007268347e-05,
+      "loss": 1.2853,
+      "step": 30880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.976820665017413e-05,
+      "loss": 1.3256,
+      "step": 30890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.975607051152372e-05,
+      "loss": 1.3236,
+      "step": 30900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.974393165783951e-05,
+      "loss": 1.3947,
+      "step": 30910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.973179009022913e-05,
+      "loss": 1.3211,
+      "step": 30920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.971964580980039e-05,
+      "loss": 1.2352,
+      "step": 30930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.970749881766137e-05,
+      "loss": 1.3411,
+      "step": 30940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.969534911492042e-05,
+      "loss": 1.1491,
+      "step": 30950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.96831967026861e-05,
+      "loss": 1.4631,
+      "step": 30960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.967104158206725e-05,
+      "loss": 1.3809,
+      "step": 30970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.96588837541729e-05,
+      "loss": 1.1897,
+      "step": 30980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.964672322011241e-05,
+      "loss": 1.3747,
+      "step": 30990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.963455998099532e-05,
+      "loss": 1.414,
+      "step": 31000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.962239403793144e-05,
+      "loss": 1.3563,
+      "step": 31010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.961022539203082e-05,
+      "loss": 1.1784,
+      "step": 31020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.95980540444038e-05,
+      "loss": 1.2734,
+      "step": 31030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.958587999616092e-05,
+      "loss": 1.3385,
+      "step": 31040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.957370324841293e-05,
+      "loss": 1.3774,
+      "step": 31050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.956152380227093e-05,
+      "loss": 1.1256,
+      "step": 31060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.954934165884617e-05,
+      "loss": 1.3006,
+      "step": 31070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.953715681925023e-05,
+      "loss": 1.1921,
+      "step": 31080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.952496928459485e-05,
+      "loss": 1.3989,
+      "step": 31090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.951277905599209e-05,
+      "loss": 1.4106,
+      "step": 31100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.950058613455418e-05,
+      "loss": 1.3119,
+      "step": 31110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.948839052139367e-05,
+      "loss": 1.2639,
+      "step": 31120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.947619221762335e-05,
+      "loss": 1.2607,
+      "step": 31130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.946399122435616e-05,
+      "loss": 1.1748,
+      "step": 31140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.945178754270542e-05,
+      "loss": 1.2188,
+      "step": 31150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.943958117378458e-05,
+      "loss": 1.1876,
+      "step": 31160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.942737211870743e-05,
+      "loss": 1.2138,
+      "step": 31170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.941516037858796e-05,
+      "loss": 1.1043,
+      "step": 31180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.940294595454036e-05,
+      "loss": 1.3516,
+      "step": 31190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.939072884767915e-05,
+      "loss": 1.1847,
+      "step": 31200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.937850905911905e-05,
+      "loss": 1.6446,
+      "step": 31210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.936628658997502e-05,
+      "loss": 1.1596,
+      "step": 31220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.935406144136226e-05,
+      "loss": 1.4597,
+      "step": 31230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.934183361439629e-05,
+      "loss": 1.2567,
+      "step": 31240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.932960311019275e-05,
+      "loss": 1.1242,
+      "step": 31250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.931736992986762e-05,
+      "loss": 1.3772,
+      "step": 31260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.930513407453709e-05,
+      "loss": 1.3123,
+      "step": 31270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.929289554531758e-05,
+      "loss": 1.2481,
+      "step": 31280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.92806543433258e-05,
+      "loss": 1.2453,
+      "step": 31290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.926841046967865e-05,
+      "loss": 1.4209,
+      "step": 31300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.925616392549328e-05,
+      "loss": 1.1744,
+      "step": 31310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.924391471188714e-05,
+      "loss": 1.2668,
+      "step": 31320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.923166282997788e-05,
+      "loss": 1.3251,
+      "step": 31330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.921940828088339e-05,
+      "loss": 1.2625,
+      "step": 31340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.920715106572181e-05,
+      "loss": 1.3701,
+      "step": 31350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.919489118561151e-05,
+      "loss": 1.0377,
+      "step": 31360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.918262864167116e-05,
+      "loss": 1.0791,
+      "step": 31370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.917036343501959e-05,
+      "loss": 1.3158,
+      "step": 31380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.915809556677595e-05,
+      "loss": 0.9222,
+      "step": 31390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.914582503805956e-05,
+      "loss": 1.1447,
+      "step": 31400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.913355184999004e-05,
+      "loss": 1.3185,
+      "step": 31410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.912127600368723e-05,
+      "loss": 1.2184,
+      "step": 31420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.910899750027123e-05,
+      "loss": 1.4154,
+      "step": 31430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.909671634086235e-05,
+      "loss": 1.2962,
+      "step": 31440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.908443252658118e-05,
+      "loss": 1.2536,
+      "step": 31450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.90721460585485e-05,
+      "loss": 1.4812,
+      "step": 31460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.905985693788537e-05,
+      "loss": 1.2097,
+      "step": 31470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.904756516571312e-05,
+      "loss": 1.162,
+      "step": 31480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.903527074315326e-05,
+      "loss": 1.2438,
+      "step": 31490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.902297367132757e-05,
+      "loss": 1.2634,
+      "step": 31500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.901067395135808e-05,
+      "loss": 1.3383,
+      "step": 31510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.899837158436705e-05,
+      "loss": 1.2777,
+      "step": 31520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.8986066571477e-05,
+      "loss": 1.4775,
+      "step": 31530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.897375891381066e-05,
+      "loss": 1.3543,
+      "step": 31540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.896144861249103e-05,
+      "loss": 1.1671,
+      "step": 31550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.894913566864131e-05,
+      "loss": 1.4008,
+      "step": 31560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.893682008338501e-05,
+      "loss": 1.3362,
+      "step": 31570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.892450185784581e-05,
+      "loss": 1.3773,
+      "step": 31580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.89121809931477e-05,
+      "loss": 1.2376,
+      "step": 31590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.889985749041482e-05,
+      "loss": 1.0669,
+      "step": 31600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.888753135077164e-05,
+      "loss": 1.3547,
+      "step": 31610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.887520257534282e-05,
+      "loss": 1.3871,
+      "step": 31620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.886287116525328e-05,
+      "loss": 1.3832,
+      "step": 31630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.885053712162819e-05,
+      "loss": 1.4147,
+      "step": 31640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.883820044559291e-05,
+      "loss": 1.4529,
+      "step": 31650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.882586113827312e-05,
+      "loss": 1.3178,
+      "step": 31660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.881351920079465e-05,
+      "loss": 1.4368,
+      "step": 31670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.880117463428364e-05,
+      "loss": 1.3981,
+      "step": 31680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.878882743986647e-05,
+      "loss": 1.1767,
+      "step": 31690
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.877647761866969e-05,
+      "loss": 1.2676,
+      "step": 31700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.876412517182014e-05,
+      "loss": 1.1743,
+      "step": 31710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.875177010044493e-05,
+      "loss": 1.3237,
+      "step": 31720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.873941240567133e-05,
+      "loss": 1.4447,
+      "step": 31730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.872705208862694e-05,
+      "loss": 1.5072,
+      "step": 31740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.871468915043952e-05,
+      "loss": 1.3557,
+      "step": 31750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.87023235922371e-05,
+      "loss": 1.4993,
+      "step": 31760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.868995541514796e-05,
+      "loss": 1.3452,
+      "step": 31770
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.867758462030062e-05,
+      "loss": 1.1674,
+      "step": 31780
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.866521120882382e-05,
+      "loss": 1.2627,
+      "step": 31790
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.865283518184654e-05,
+      "loss": 1.219,
+      "step": 31800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.864045654049803e-05,
+      "loss": 1.364,
+      "step": 31810
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.862807528590772e-05,
+      "loss": 1.215,
+      "step": 31820
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.861569141920533e-05,
+      "loss": 1.3557,
+      "step": 31830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.860330494152081e-05,
+      "loss": 1.4345,
+      "step": 31840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.859091585398433e-05,
+      "loss": 1.3383,
+      "step": 31850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.85785241577263e-05,
+      "loss": 1.078,
+      "step": 31860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.856612985387741e-05,
+      "loss": 1.296,
+      "step": 31870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.85537329435685e-05,
+      "loss": 1.3709,
+      "step": 31880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.854133342793075e-05,
+      "loss": 1.1628,
+      "step": 31890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.85289313080955e-05,
+      "loss": 1.2544,
+      "step": 31900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.851652658519437e-05,
+      "loss": 1.2134,
+      "step": 31910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.850411926035922e-05,
+      "loss": 1.1733,
+      "step": 31920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.849170933472207e-05,
+      "loss": 1.3353,
+      "step": 31930
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.847929680941532e-05,
+      "loss": 1.1677,
+      "step": 31940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.846688168557148e-05,
+      "loss": 1.2535,
+      "step": 31950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.845446396432335e-05,
+      "loss": 1.2486,
+      "step": 31960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.844204364680395e-05,
+      "loss": 1.3036,
+      "step": 31970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.842962073414657e-05,
+      "loss": 1.2812,
+      "step": 31980
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.841719522748471e-05,
+      "loss": 1.2309,
+      "step": 31990
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.84047671279521e-05,
+      "loss": 1.2043,
+      "step": 32000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.839233643668272e-05,
+      "loss": 1.4122,
+      "step": 32010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.837990315481076e-05,
+      "loss": 1.4687,
+      "step": 32020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.836746728347072e-05,
+      "loss": 1.2183,
+      "step": 32030
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.835502882379729e-05,
+      "loss": 1.1445,
+      "step": 32040
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.834258777692532e-05,
+      "loss": 1.24,
+      "step": 32050
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.833014414399003e-05,
+      "loss": 1.5049,
+      "step": 32060
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.831769792612679e-05,
+      "loss": 1.232,
+      "step": 32070
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.830524912447126e-05,
+      "loss": 1.2991,
+      "step": 32080
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.829279774015927e-05,
+      "loss": 1.3614,
+      "step": 32090
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.828034377432693e-05,
+      "loss": 1.2771,
+      "step": 32100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.82678872281106e-05,
+      "loss": 1.2044,
+      "step": 32110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.825542810264683e-05,
+      "loss": 1.2124,
+      "step": 32120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.824296639907243e-05,
+      "loss": 1.2771,
+      "step": 32130
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.823050211852445e-05,
+      "loss": 1.2584,
+      "step": 32140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.821803526214019e-05,
+      "loss": 1.4263,
+      "step": 32150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.820556583105711e-05,
+      "loss": 1.2793,
+      "step": 32160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.819309382641302e-05,
+      "loss": 1.1144,
+      "step": 32170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.818061924934584e-05,
+      "loss": 1.3202,
+      "step": 32180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.816814210099384e-05,
+      "loss": 1.2168,
+      "step": 32190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.815566238249546e-05,
+      "loss": 1.478,
+      "step": 32200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.814318009498938e-05,
+      "loss": 1.448,
+      "step": 32210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.813069523961451e-05,
+      "loss": 1.2886,
+      "step": 32220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.811820781751003e-05,
+      "loss": 1.3078,
+      "step": 32230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.810571782981531e-05,
+      "loss": 1.3631,
+      "step": 32240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.809322527766999e-05,
+      "loss": 1.2765,
+      "step": 32250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.80807301622139e-05,
+      "loss": 1.0878,
+      "step": 32260
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.806823248458719e-05,
+      "loss": 1.3806,
+      "step": 32270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.805573224593012e-05,
+      "loss": 1.2058,
+      "step": 32280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.804322944738329e-05,
+      "loss": 1.293,
+      "step": 32290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.803072409008749e-05,
+      "loss": 1.1969,
+      "step": 32300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.801821617518372e-05,
+      "loss": 1.3988,
+      "step": 32310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.800570570381327e-05,
+      "loss": 1.2118,
+      "step": 32320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.799319267711763e-05,
+      "loss": 1.1554,
+      "step": 32330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.79806770962385e-05,
+      "loss": 1.1899,
+      "step": 32340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.796815896231787e-05,
+      "loss": 1.1534,
+      "step": 32350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.795563827649792e-05,
+      "loss": 1.4639,
+      "step": 32360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.794311503992107e-05,
+      "loss": 1.2374,
+      "step": 32370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.793058925372998e-05,
+      "loss": 1.2222,
+      "step": 32380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.791806091906754e-05,
+      "loss": 1.1837,
+      "step": 32390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.790553003707689e-05,
+      "loss": 1.2761,
+      "step": 32400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.789299660890137e-05,
+      "loss": 1.2179,
+      "step": 32410
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.788046063568458e-05,
+      "loss": 1.4358,
+      "step": 32420
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.786792211857031e-05,
+      "loss": 1.4125,
+      "step": 32430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.785538105870262e-05,
+      "loss": 1.3298,
+      "step": 32440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.784283745722583e-05,
+      "loss": 1.2219,
+      "step": 32450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.783029131528443e-05,
+      "loss": 1.0797,
+      "step": 32460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.781774263402317e-05,
+      "loss": 1.2882,
+      "step": 32470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.780519141458703e-05,
+      "loss": 1.1884,
+      "step": 32480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.779263765812121e-05,
+      "loss": 1.0929,
+      "step": 32490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.778008136577117e-05,
+      "loss": 1.1277,
+      "step": 32500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.776752253868257e-05,
+      "loss": 1.3553,
+      "step": 32510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.775496117800134e-05,
+      "loss": 1.3348,
+      "step": 32520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.774239728487361e-05,
+      "loss": 1.3471,
+      "step": 32530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.772983086044571e-05,
+      "loss": 1.232,
+      "step": 32540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.771726190586428e-05,
+      "loss": 1.14,
+      "step": 32550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.770469042227613e-05,
+      "loss": 1.4092,
+      "step": 32560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.769211641082832e-05,
+      "loss": 1.2098,
+      "step": 32570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.767953987266816e-05,
+      "loss": 1.3986,
+      "step": 32580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.766696080894315e-05,
+      "loss": 1.5044,
+      "step": 32590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.765437922080105e-05,
+      "loss": 1.456,
+      "step": 32600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.764179510938985e-05,
+      "loss": 1.3229,
+      "step": 32610
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.762920847585772e-05,
+      "loss": 1.2522,
+      "step": 32620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.761661932135318e-05,
+      "loss": 1.1673,
+      "step": 32630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.760402764702483e-05,
+      "loss": 1.4183,
+      "step": 32640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.759143345402163e-05,
+      "loss": 1.1912,
+      "step": 32650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.757883674349266e-05,
+      "loss": 1.4115,
+      "step": 32660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.756623751658729e-05,
+      "loss": 1.4258,
+      "step": 32670
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.755363577445516e-05,
+      "loss": 1.1299,
+      "step": 32680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.754103151824603e-05,
+      "loss": 1.1065,
+      "step": 32690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.752842474911001e-05,
+      "loss": 1.1116,
+      "step": 32700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.751581546819733e-05,
+      "loss": 1.2559,
+      "step": 32710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.750320367665851e-05,
+      "loss": 1.243,
+      "step": 32720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.74905893756443e-05,
+      "loss": 1.1619,
+      "step": 32730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.747797256630567e-05,
+      "loss": 1.3613,
+      "step": 32740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.74653532497938e-05,
+      "loss": 1.1512,
+      "step": 32750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.745273142726012e-05,
+      "loss": 1.2168,
+      "step": 32760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.744010709985629e-05,
+      "loss": 1.2785,
+      "step": 32770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.742748026873418e-05,
+      "loss": 1.3567,
+      "step": 32780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.741485093504591e-05,
+      "loss": 1.2556,
+      "step": 32790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.740221909994382e-05,
+      "loss": 1.239,
+      "step": 32800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.738958476458047e-05,
+      "loss": 1.357,
+      "step": 32810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.737694793010866e-05,
+      "loss": 1.2633,
+      "step": 32820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.73643085976814e-05,
+      "loss": 1.1337,
+      "step": 32830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.735166676845196e-05,
+      "loss": 1.3935,
+      "step": 32840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.733902244357383e-05,
+      "loss": 1.0734,
+      "step": 32850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.732637562420066e-05,
+      "loss": 1.3726,
+      "step": 32860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.731372631148644e-05,
+      "loss": 1.3189,
+      "step": 32870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.730107450658532e-05,
+      "loss": 1.3462,
+      "step": 32880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.728842021065166e-05,
+      "loss": 1.4169,
+      "step": 32890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.72757634248401e-05,
+      "loss": 1.0966,
+      "step": 32900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.72631041503055e-05,
+      "loss": 1.2818,
+      "step": 32910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.725044238820289e-05,
+      "loss": 1.2736,
+      "step": 32920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.72377781396876e-05,
+      "loss": 1.3515,
+      "step": 32930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.722511140591514e-05,
+      "loss": 1.3008,
+      "step": 32940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.721244218804126e-05,
+      "loss": 1.1625,
+      "step": 32950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.719977048722196e-05,
+      "loss": 1.2885,
+      "step": 32960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.71870963046134e-05,
+      "loss": 1.199,
+      "step": 32970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.717441964137204e-05,
+      "loss": 1.2953,
+      "step": 32980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.716174049865454e-05,
+      "loss": 1.367,
+      "step": 32990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.714905887761779e-05,
+      "loss": 1.2351,
+      "step": 33000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.713637477941886e-05,
+      "loss": 1.2883,
+      "step": 33010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.712368820521513e-05,
+      "loss": 1.254,
+      "step": 33020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.711099915616415e-05,
+      "loss": 1.1531,
+      "step": 33030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.709830763342367e-05,
+      "loss": 1.2354,
+      "step": 33040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.708561363815175e-05,
+      "loss": 1.1545,
+      "step": 33050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.707291717150663e-05,
+      "loss": 1.4253,
+      "step": 33060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.706021823464673e-05,
+      "loss": 1.281,
+      "step": 33070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.704751682873077e-05,
+      "loss": 1.1426,
+      "step": 33080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.703481295491765e-05,
+      "loss": 1.2292,
+      "step": 33090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.702210661436654e-05,
+      "loss": 1.1432,
+      "step": 33100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.70093978082368e-05,
+      "loss": 1.2121,
+      "step": 33110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.699668653768795e-05,
+      "loss": 1.0782,
+      "step": 33120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.69839728038799e-05,
+      "loss": 1.2624,
+      "step": 33130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.697125660797262e-05,
+      "loss": 1.2758,
+      "step": 33140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.695853795112643e-05,
+      "loss": 1.3233,
+      "step": 33150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.694581683450179e-05,
+      "loss": 1.2187,
+      "step": 33160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.69330932592594e-05,
+      "loss": 1.0823,
+      "step": 33170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.692036722656023e-05,
+      "loss": 1.5059,
+      "step": 33180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.690763873756539e-05,
+      "loss": 1.2895,
+      "step": 33190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.689490779343631e-05,
+      "loss": 1.3533,
+      "step": 33200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.68821743953346e-05,
+      "loss": 1.3315,
+      "step": 33210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.686943854442208e-05,
+      "loss": 1.2424,
+      "step": 33220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.685670024186082e-05,
+      "loss": 1.6076,
+      "step": 33230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.684395948881306e-05,
+      "loss": 1.2685,
+      "step": 33240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.683121628644136e-05,
+      "loss": 1.084,
+      "step": 33250
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.681847063590844e-05,
+      "loss": 1.3628,
+      "step": 33260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.680572253837721e-05,
+      "loss": 1.1774,
+      "step": 33270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.679297199501089e-05,
+      "loss": 1.4887,
+      "step": 33280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.678021900697283e-05,
+      "loss": 1.3775,
+      "step": 33290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.67674635754267e-05,
+      "loss": 1.1925,
+      "step": 33300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.675470570153633e-05,
+      "loss": 1.1798,
+      "step": 33310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.674194538646577e-05,
+      "loss": 1.2486,
+      "step": 33320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.672918263137934e-05,
+      "loss": 1.1049,
+      "step": 33330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.67164174374415e-05,
+      "loss": 1.2164,
+      "step": 33340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.670364980581704e-05,
+      "loss": 1.2386,
+      "step": 33350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.66908797376709e-05,
+      "loss": 1.1891,
+      "step": 33360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.667810723416824e-05,
+      "loss": 1.14,
+      "step": 33370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.666533229647449e-05,
+      "loss": 1.2207,
+      "step": 33380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.665255492575525e-05,
+      "loss": 1.2142,
+      "step": 33390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.663977512317639e-05,
+      "loss": 1.232,
+      "step": 33400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.662699288990394e-05,
+      "loss": 1.2781,
+      "step": 33410
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.661420822710423e-05,
+      "loss": 1.2841,
+      "step": 33420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.660142113594377e-05,
+      "loss": 1.2639,
+      "step": 33430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.658863161758925e-05,
+      "loss": 1.1885,
+      "step": 33440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.657583967320768e-05,
+      "loss": 1.2975,
+      "step": 33450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.656304530396619e-05,
+      "loss": 1.2814,
+      "step": 33460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.65502485110322e-05,
+      "loss": 0.9301,
+      "step": 33470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.653744929557331e-05,
+      "loss": 1.555,
+      "step": 33480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.65246476587574e-05,
+      "loss": 1.1058,
+      "step": 33490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.651184360175247e-05,
+      "loss": 1.3931,
+      "step": 33500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.649903712572686e-05,
+      "loss": 1.3049,
+      "step": 33510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.648622823184903e-05,
+      "loss": 1.4389,
+      "step": 33520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.647341692128773e-05,
+      "loss": 1.4891,
+      "step": 33530
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.646060319521187e-05,
+      "loss": 1.1877,
+      "step": 33540
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.644778705479065e-05,
+      "loss": 1.037,
+      "step": 33550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.643496850119342e-05,
+      "loss": 1.2066,
+      "step": 33560
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.642214753558983e-05,
+      "loss": 1.2275,
+      "step": 33570
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.640932415914964e-05,
+      "loss": 1.3623,
+      "step": 33580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.639649837304294e-05,
+      "loss": 1.3331,
+      "step": 33590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.638367017843998e-05,
+      "loss": 1.364,
+      "step": 33600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.637083957651125e-05,
+      "loss": 1.4006,
+      "step": 33610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.635800656842745e-05,
+      "loss": 1.1238,
+      "step": 33620
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.634517115535946e-05,
+      "loss": 1.2952,
+      "step": 33630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.633233333847848e-05,
+      "loss": 1.4126,
+      "step": 33640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.631949311895588e-05,
+      "loss": 1.1871,
+      "step": 33650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.630665049796319e-05,
+      "loss": 1.1875,
+      "step": 33660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.629380547667224e-05,
+      "loss": 1.3736,
+      "step": 33670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.628095805625502e-05,
+      "loss": 1.2475,
+      "step": 33680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.626810823788381e-05,
+      "loss": 1.334,
+      "step": 33690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.625525602273103e-05,
+      "loss": 1.1109,
+      "step": 33700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.624240141196938e-05,
+      "loss": 1.4416,
+      "step": 33710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.622954440677174e-05,
+      "loss": 1.159,
+      "step": 33720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.621668500831121e-05,
+      "loss": 1.2166,
+      "step": 33730
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.620382321776115e-05,
+      "loss": 1.2802,
+      "step": 33740
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.619095903629508e-05,
+      "loss": 1.1555,
+      "step": 33750
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.61780924650868e-05,
+      "loss": 1.349,
+      "step": 33760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.616522350531026e-05,
+      "loss": 1.3891,
+      "step": 33770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.61523521581397e-05,
+      "loss": 1.3187,
+      "step": 33780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.61394784247495e-05,
+      "loss": 1.3633,
+      "step": 33790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.61266023063143e-05,
+      "loss": 1.2676,
+      "step": 33800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.6113723804009e-05,
+      "loss": 1.2281,
+      "step": 33810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.610084291900863e-05,
+      "loss": 1.1839,
+      "step": 33820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.608795965248852e-05,
+      "loss": 1.4485,
+      "step": 33830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.607507400562413e-05,
+      "loss": 1.2172,
+      "step": 33840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.606218597959122e-05,
+      "loss": 1.3025,
+      "step": 33850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.604929557556572e-05,
+      "loss": 1.1577,
+      "step": 33860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.603640279472379e-05,
+      "loss": 1.4266,
+      "step": 33870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.602350763824181e-05,
+      "loss": 1.3145,
+      "step": 33880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.60106101072964e-05,
+      "loss": 1.2152,
+      "step": 33890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.59977102030643e-05,
+      "loss": 1.2002,
+      "step": 33900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.59848079267226e-05,
+      "loss": 1.5019,
+      "step": 33910
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.597190327944852e-05,
+      "loss": 1.2517,
+      "step": 33920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.595899626241952e-05,
+      "loss": 1.2598,
+      "step": 33930
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.594608687681326e-05,
+      "loss": 1.3104,
+      "step": 33940
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.593317512380766e-05,
+      "loss": 1.246,
+      "step": 33950
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.59202610045808e-05,
+      "loss": 1.3581,
+      "step": 33960
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.590734452031103e-05,
+      "loss": 1.2522,
+      "step": 33970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.589442567217687e-05,
+      "loss": 1.3645,
+      "step": 33980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.588150446135709e-05,
+      "loss": 1.1933,
+      "step": 33990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.586858088903063e-05,
+      "loss": 1.2719,
+      "step": 34000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.58556549563767e-05,
+      "loss": 1.2717,
+      "step": 34010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.58427266645747e-05,
+      "loss": 1.189,
+      "step": 34020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.582979601480427e-05,
+      "loss": 1.223,
+      "step": 34030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.581686300824518e-05,
+      "loss": 1.4804,
+      "step": 34040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.580392764607753e-05,
+      "loss": 1.2457,
+      "step": 34050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.579098992948156e-05,
+      "loss": 1.3047,
+      "step": 34060
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.577804985963775e-05,
+      "loss": 1.2986,
+      "step": 34070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.57651074377268e-05,
+      "loss": 1.3243,
+      "step": 34080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.575216266492958e-05,
+      "loss": 1.224,
+      "step": 34090
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.573921554242726e-05,
+      "loss": 1.2771,
+      "step": 34100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.572626607140114e-05,
+      "loss": 1.2465,
+      "step": 34110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.57133142530328e-05,
+      "loss": 1.4554,
+      "step": 34120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.570036008850396e-05,
+      "loss": 1.3207,
+      "step": 34130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.568740357899663e-05,
+      "loss": 1.2288,
+      "step": 34140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.567444472569298e-05,
+      "loss": 1.3859,
+      "step": 34150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.566148352977544e-05,
+      "loss": 1.2249,
+      "step": 34160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.564851999242663e-05,
+      "loss": 1.5065,
+      "step": 34170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.563555411482934e-05,
+      "loss": 1.1996,
+      "step": 34180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.562258589816665e-05,
+      "loss": 1.2967,
+      "step": 34190
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.560961534362182e-05,
+      "loss": 1.2754,
+      "step": 34200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.559664245237832e-05,
+      "loss": 1.1849,
+      "step": 34210
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.558366722561983e-05,
+      "loss": 1.4332,
+      "step": 34220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.557068966453022e-05,
+      "loss": 1.2562,
+      "step": 34230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.555770977029367e-05,
+      "loss": 1.1644,
+      "step": 34240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.554472754409444e-05,
+      "loss": 1.1976,
+      "step": 34250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.553174298711711e-05,
+      "loss": 1.1114,
+      "step": 34260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.551875610054641e-05,
+      "loss": 1.4584,
+      "step": 34270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.55057668855673e-05,
+      "loss": 1.3919,
+      "step": 34280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.549277534336497e-05,
+      "loss": 1.4206,
+      "step": 34290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.547978147512479e-05,
+      "loss": 1.2289,
+      "step": 34300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.546678528203237e-05,
+      "loss": 1.2138,
+      "step": 34310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.545378676527353e-05,
+      "loss": 1.2962,
+      "step": 34320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.544078592603429e-05,
+      "loss": 1.2363,
+      "step": 34330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.542778276550086e-05,
+      "loss": 1.249,
+      "step": 34340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.541477728485973e-05,
+      "loss": 1.3437,
+      "step": 34350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.540176948529753e-05,
+      "loss": 1.2118,
+      "step": 34360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.538875936800116e-05,
+      "loss": 1.4572,
+      "step": 34370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.537574693415765e-05,
+      "loss": 1.3596,
+      "step": 34380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.536273218495434e-05,
+      "loss": 1.2226,
+      "step": 34390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.534971512157871e-05,
+      "loss": 1.3725,
+      "step": 34400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.533669574521849e-05,
+      "loss": 1.2912,
+      "step": 34410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.532367405706162e-05,
+      "loss": 1.3452,
+      "step": 34420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.531065005829621e-05,
+      "loss": 1.2263,
+      "step": 34430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.529762375011061e-05,
+      "loss": 1.3199,
+      "step": 34440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.528459513369342e-05,
+      "loss": 1.3155,
+      "step": 34450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.527156421023333e-05,
+      "loss": 1.2495,
+      "step": 34460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.525853098091943e-05,
+      "loss": 1.2566,
+      "step": 34470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.524549544694081e-05,
+      "loss": 1.1455,
+      "step": 34480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.523245760948694e-05,
+      "loss": 1.2725,
+      "step": 34490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.521941746974741e-05,
+      "loss": 1.4364,
+      "step": 34500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.520637502891202e-05,
+      "loss": 1.1361,
+      "step": 34510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.519333028817083e-05,
+      "loss": 1.2371,
+      "step": 34520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.518028324871407e-05,
+      "loss": 1.3555,
+      "step": 34530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.516723391173218e-05,
+      "loss": 1.3577,
+      "step": 34540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.515418227841584e-05,
+      "loss": 1.2978,
+      "step": 34550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.514112834995593e-05,
+      "loss": 1.1839,
+      "step": 34560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.51280721275435e-05,
+      "loss": 1.4009,
+      "step": 34570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.511501361236985e-05,
+      "loss": 1.0374,
+      "step": 34580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.510195280562648e-05,
+      "loss": 1.2915,
+      "step": 34590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.508888970850512e-05,
+      "loss": 1.425,
+      "step": 34600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.507582432219763e-05,
+      "loss": 1.2595,
+      "step": 34610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.50627566478962e-05,
+      "loss": 1.2773,
+      "step": 34620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.504968668679311e-05,
+      "loss": 1.2683,
+      "step": 34630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.503661444008096e-05,
+      "loss": 1.4671,
+      "step": 34640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.502353990895244e-05,
+      "loss": 1.3415,
+      "step": 34650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.501046309460055e-05,
+      "loss": 1.3181,
+      "step": 34660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.499738399821845e-05,
+      "loss": 1.3761,
+      "step": 34670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.49843026209995e-05,
+      "loss": 1.412,
+      "step": 34680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.497121896413732e-05,
+      "loss": 1.3649,
+      "step": 34690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.495813302882566e-05,
+      "loss": 1.2426,
+      "step": 34700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.494504481625854e-05,
+      "loss": 1.3072,
+      "step": 34710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.493195432763018e-05,
+      "loss": 1.343,
+      "step": 34720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.491886156413497e-05,
+      "loss": 1.0702,
+      "step": 34730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.490576652696755e-05,
+      "loss": 1.1581,
+      "step": 34740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.489266921732274e-05,
+      "loss": 1.3473,
+      "step": 34750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.48795696363956e-05,
+      "loss": 1.2574,
+      "step": 34760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.486646778538135e-05,
+      "loss": 1.288,
+      "step": 34770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.485336366547547e-05,
+      "loss": 1.2667,
+      "step": 34780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.48402572778736e-05,
+      "loss": 1.1552,
+      "step": 34790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.48271486237716e-05,
+      "loss": 1.239,
+      "step": 34800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.481403770436554e-05,
+      "loss": 1.2492,
+      "step": 34810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.480092452085175e-05,
+      "loss": 1.3376,
+      "step": 34820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.478780907442665e-05,
+      "loss": 1.3622,
+      "step": 34830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.477469136628696e-05,
+      "loss": 1.379,
+      "step": 34840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.47615713976296e-05,
+      "loss": 1.2066,
+      "step": 34850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.474844916965165e-05,
+      "loss": 1.3111,
+      "step": 34860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.473532468355043e-05,
+      "loss": 1.1847,
+      "step": 34870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.472219794052344e-05,
+      "loss": 1.3582,
+      "step": 34880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.470906894176844e-05,
+      "loss": 1.0988,
+      "step": 34890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.469593768848331e-05,
+      "loss": 1.1431,
+      "step": 34900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.468280418186624e-05,
+      "loss": 1.333,
+      "step": 34910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.466966842311555e-05,
+      "loss": 1.2121,
+      "step": 34920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.465653041342977e-05,
+      "loss": 1.2633,
+      "step": 34930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.464339015400768e-05,
+      "loss": 1.4368,
+      "step": 34940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.46302476460482e-05,
+      "loss": 1.3346,
+      "step": 34950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.46171028907505e-05,
+      "loss": 1.147,
+      "step": 34960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.4603955889314e-05,
+      "loss": 1.2308,
+      "step": 34970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.459080664293821e-05,
+      "loss": 1.3684,
+      "step": 34980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.457765515282293e-05,
+      "loss": 1.4716,
+      "step": 34990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.456450142016814e-05,
+      "loss": 1.3971,
+      "step": 35000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.455134544617402e-05,
+      "loss": 1.2953,
+      "step": 35010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.453818723204098e-05,
+      "loss": 1.3975,
+      "step": 35020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.452502677896961e-05,
+      "loss": 1.1598,
+      "step": 35030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.451186408816069e-05,
+      "loss": 1.3326,
+      "step": 35040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.449869916081524e-05,
+      "loss": 1.155,
+      "step": 35050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.448553199813445e-05,
+      "loss": 1.3561,
+      "step": 35060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.447236260131975e-05,
+      "loss": 1.1837,
+      "step": 35070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.445919097157277e-05,
+      "loss": 1.2882,
+      "step": 35080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.44460171100953e-05,
+      "loss": 1.1631,
+      "step": 35090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.443284101808937e-05,
+      "loss": 1.2922,
+      "step": 35100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.44196626967572e-05,
+      "loss": 1.4103,
+      "step": 35110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.440648214730124e-05,
+      "loss": 1.4262,
+      "step": 35120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.439329937092413e-05,
+      "loss": 1.2043,
+      "step": 35130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.438011436882867e-05,
+      "loss": 1.3484,
+      "step": 35140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.436692714221792e-05,
+      "loss": 1.4364,
+      "step": 35150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.435373769229512e-05,
+      "loss": 1.2773,
+      "step": 35160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.434054602026371e-05,
+      "loss": 1.1325,
+      "step": 35170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.432735212732737e-05,
+      "loss": 1.0857,
+      "step": 35180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.43141560146899e-05,
+      "loss": 1.3902,
+      "step": 35190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.43009576835554e-05,
+      "loss": 1.1651,
+      "step": 35200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.428775713512807e-05,
+      "loss": 1.4953,
+      "step": 35210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.427455437061243e-05,
+      "loss": 1.1083,
+      "step": 35220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.42613493912131e-05,
+      "loss": 1.0344,
+      "step": 35230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.424814219813497e-05,
+      "loss": 1.2533,
+      "step": 35240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.423493279258306e-05,
+      "loss": 1.3005,
+      "step": 35250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.42217211757627e-05,
+      "loss": 1.3189,
+      "step": 35260
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.420850734887929e-05,
+      "loss": 1.4458,
+      "step": 35270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.419529131313855e-05,
+      "loss": 1.2736,
+      "step": 35280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.418207306974631e-05,
+      "loss": 1.6326,
+      "step": 35290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.416885261990869e-05,
+      "loss": 1.4359,
+      "step": 35300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.415562996483192e-05,
+      "loss": 1.5167,
+      "step": 35310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.414240510572251e-05,
+      "loss": 1.1607,
+      "step": 35320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.412917804378712e-05,
+      "loss": 1.0971,
+      "step": 35330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.411594878023262e-05,
+      "loss": 1.2977,
+      "step": 35340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.41027173162661e-05,
+      "loss": 1.2427,
+      "step": 35350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.408948365309483e-05,
+      "loss": 1.2354,
+      "step": 35360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.407624779192627e-05,
+      "loss": 1.4373,
+      "step": 35370
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.406300973396816e-05,
+      "loss": 1.4373,
+      "step": 35380
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.404976948042832e-05,
+      "loss": 1.1593,
+      "step": 35390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.403652703251487e-05,
+      "loss": 1.2892,
+      "step": 35400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.402328239143606e-05,
+      "loss": 1.0377,
+      "step": 35410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.40100355584004e-05,
+      "loss": 1.1029,
+      "step": 35420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.399678653461656e-05,
+      "loss": 1.3885,
+      "step": 35430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.398353532129341e-05,
+      "loss": 1.2575,
+      "step": 35440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.397028191964004e-05,
+      "loss": 1.071,
+      "step": 35450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.395702633086572e-05,
+      "loss": 1.3516,
+      "step": 35460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.394376855617995e-05,
+      "loss": 1.4154,
+      "step": 35470
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.393050859679239e-05,
+      "loss": 1.1707,
+      "step": 35480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.391724645391293e-05,
+      "loss": 1.238,
+      "step": 35490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.390398212875166e-05,
+      "loss": 1.3833,
+      "step": 35500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.389071562251883e-05,
+      "loss": 1.3467,
+      "step": 35510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.387744693642493e-05,
+      "loss": 1.137,
+      "step": 35520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.386417607168064e-05,
+      "loss": 1.2642,
+      "step": 35530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.385090302949682e-05,
+      "loss": 1.4112,
+      "step": 35540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.383762781108455e-05,
+      "loss": 1.3657,
+      "step": 35550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.38243504176551e-05,
+      "loss": 1.1587,
+      "step": 35560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.381107085041993e-05,
+      "loss": 1.3146,
+      "step": 35570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.379778911059074e-05,
+      "loss": 1.1741,
+      "step": 35580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.378450519937938e-05,
+      "loss": 1.1403,
+      "step": 35590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.37712191179979e-05,
+      "loss": 1.3594,
+      "step": 35600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.375793086765855e-05,
+      "loss": 1.6164,
+      "step": 35610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.374464044957383e-05,
+      "loss": 1.1541,
+      "step": 35620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.373134786495638e-05,
+      "loss": 1.1719,
+      "step": 35630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.371805311501904e-05,
+      "loss": 1.4099,
+      "step": 35640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.37047562009749e-05,
+      "loss": 1.4676,
+      "step": 35650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.369145712403716e-05,
+      "loss": 1.4472,
+      "step": 35660
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.367815588541931e-05,
+      "loss": 1.3282,
+      "step": 35670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.366485248633499e-05,
+      "loss": 1.3898,
+      "step": 35680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.365154692799804e-05,
+      "loss": 1.2438,
+      "step": 35690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.363823921162248e-05,
+      "loss": 1.3301,
+      "step": 35700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.362492933842255e-05,
+      "loss": 1.174,
+      "step": 35710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.36116173096127e-05,
+      "loss": 1.2322,
+      "step": 35720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.359830312640756e-05,
+      "loss": 1.2753,
+      "step": 35730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.358498679002195e-05,
+      "loss": 1.3126,
+      "step": 35740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.357166830167087e-05,
+      "loss": 1.1424,
+      "step": 35750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.355834766256957e-05,
+      "loss": 1.2487,
+      "step": 35760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.354502487393347e-05,
+      "loss": 1.2732,
+      "step": 35770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.353169993697815e-05,
+      "loss": 1.1733,
+      "step": 35780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.351837285291945e-05,
+      "loss": 1.2748,
+      "step": 35790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.350504362297334e-05,
+      "loss": 1.3946,
+      "step": 35800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.349171224835604e-05,
+      "loss": 1.121,
+      "step": 35810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.347837873028395e-05,
+      "loss": 1.2899,
+      "step": 35820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.346504306997366e-05,
+      "loss": 1.4403,
+      "step": 35830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.345170526864193e-05,
+      "loss": 1.3106,
+      "step": 35840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.343836532750577e-05,
+      "loss": 1.2838,
+      "step": 35850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.342502324778234e-05,
+      "loss": 1.3668,
+      "step": 35860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.341167903068902e-05,
+      "loss": 1.1647,
+      "step": 35870
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.339833267744338e-05,
+      "loss": 1.1176,
+      "step": 35880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.338498418926318e-05,
+      "loss": 1.1755,
+      "step": 35890
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.337163356736636e-05,
+      "loss": 1.1986,
+      "step": 35900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.335828081297109e-05,
+      "loss": 1.1371,
+      "step": 35910
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.334492592729571e-05,
+      "loss": 1.4632,
+      "step": 35920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.333156891155875e-05,
+      "loss": 1.426,
+      "step": 35930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.331820976697898e-05,
+      "loss": 1.2938,
+      "step": 35940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.330484849477528e-05,
+      "loss": 1.1865,
+      "step": 35950
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.32914850961668e-05,
+      "loss": 1.231,
+      "step": 35960
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.327811957237285e-05,
+      "loss": 1.2905,
+      "step": 35970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.326475192461294e-05,
+      "loss": 1.555,
+      "step": 35980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.32513821541068e-05,
+      "loss": 1.1432,
+      "step": 35990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.323801026207428e-05,
+      "loss": 1.244,
+      "step": 36000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.32246362497355e-05,
+      "loss": 1.4017,
+      "step": 36010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.321126011831075e-05,
+      "loss": 1.2672,
+      "step": 36020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.31978818690205e-05,
+      "loss": 1.2751,
+      "step": 36030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.318450150308545e-05,
+      "loss": 1.1557,
+      "step": 36040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.31711190217264e-05,
+      "loss": 1.1871,
+      "step": 36050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.315773442616446e-05,
+      "loss": 1.1532,
+      "step": 36060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.314434771762087e-05,
+      "loss": 1.3019,
+      "step": 36070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.313095889731706e-05,
+      "loss": 1.4045,
+      "step": 36080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.311756796647469e-05,
+      "loss": 1.2496,
+      "step": 36090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.310417492631559e-05,
+      "loss": 1.2231,
+      "step": 36100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.309077977806174e-05,
+      "loss": 1.0855,
+      "step": 36110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.307738252293538e-05,
+      "loss": 1.3741,
+      "step": 36120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.306398316215895e-05,
+      "loss": 1.3343,
+      "step": 36130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.3050581696955e-05,
+      "loss": 1.2559,
+      "step": 36140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.303717812854632e-05,
+      "loss": 1.2406,
+      "step": 36150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.302377245815592e-05,
+      "loss": 1.3662,
+      "step": 36160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.301036468700697e-05,
+      "loss": 1.3779,
+      "step": 36170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.299695481632283e-05,
+      "loss": 1.3917,
+      "step": 36180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.298354284732704e-05,
+      "loss": 1.188,
+      "step": 36190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.297012878124337e-05,
+      "loss": 1.3136,
+      "step": 36200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.295671261929576e-05,
+      "loss": 1.4019,
+      "step": 36210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.294329436270833e-05,
+      "loss": 1.0086,
+      "step": 36220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.29298740127054e-05,
+      "loss": 1.1557,
+      "step": 36230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.29164515705115e-05,
+      "loss": 1.1072,
+      "step": 36240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.290302703735132e-05,
+      "loss": 1.4596,
+      "step": 36250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.288960041444977e-05,
+      "loss": 1.2498,
+      "step": 36260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.28761717030319e-05,
+      "loss": 1.2593,
+      "step": 36270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.286274090432304e-05,
+      "loss": 1.3283,
+      "step": 36280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.28493080195486e-05,
+      "loss": 1.3535,
+      "step": 36290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.283587304993428e-05,
+      "loss": 1.317,
+      "step": 36300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.282243599670592e-05,
+      "loss": 1.2573,
+      "step": 36310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.280899686108952e-05,
+      "loss": 1.1539,
+      "step": 36320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.279555564431136e-05,
+      "loss": 1.2985,
+      "step": 36330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.278211234759784e-05,
+      "loss": 1.281,
+      "step": 36340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.276866697217555e-05,
+      "loss": 1.1453,
+      "step": 36350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.27552195192713e-05,
+      "loss": 1.4054,
+      "step": 36360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.274176999011206e-05,
+      "loss": 1.3218,
+      "step": 36370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.272831838592503e-05,
+      "loss": 1.2729,
+      "step": 36380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.271486470793757e-05,
+      "loss": 1.3934,
+      "step": 36390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.270140895737722e-05,
+      "loss": 1.4525,
+      "step": 36400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.268795113547174e-05,
+      "loss": 1.3551,
+      "step": 36410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.267449124344905e-05,
+      "loss": 1.1976,
+      "step": 36420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.266102928253728e-05,
+      "loss": 1.2997,
+      "step": 36430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.264756525396473e-05,
+      "loss": 1.3487,
+      "step": 36440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.263409915895992e-05,
+      "loss": 1.2133,
+      "step": 36450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.262063099875148e-05,
+      "loss": 1.1276,
+      "step": 36460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.260716077456836e-05,
+      "loss": 1.1639,
+      "step": 36470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.25936884876396e-05,
+      "loss": 1.2263,
+      "step": 36480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.25802141391944e-05,
+      "loss": 1.2567,
+      "step": 36490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.256673773046229e-05,
+      "loss": 1.2929,
+      "step": 36500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.255325926267282e-05,
+      "loss": 1.5596,
+      "step": 36510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.253977873705586e-05,
+      "loss": 1.3165,
+      "step": 36520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.252629615484138e-05,
+      "loss": 1.4751,
+      "step": 36530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.251281151725957e-05,
+      "loss": 1.1511,
+      "step": 36540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.249932482554085e-05,
+      "loss": 1.1302,
+      "step": 36550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.248583608091574e-05,
+      "loss": 1.1903,
+      "step": 36560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.247234528461502e-05,
+      "loss": 1.3213,
+      "step": 36570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.245885243786963e-05,
+      "loss": 1.0815,
+      "step": 36580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.244535754191067e-05,
+      "loss": 1.2147,
+      "step": 36590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.24318605979695e-05,
+      "loss": 1.0584,
+      "step": 36600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.241836160727758e-05,
+      "loss": 1.0879,
+      "step": 36610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.240486057106663e-05,
+      "loss": 1.3493,
+      "step": 36620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.239135749056851e-05,
+      "loss": 1.1648,
+      "step": 36630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.237785236701527e-05,
+      "loss": 1.1193,
+      "step": 36640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.23643452016392e-05,
+      "loss": 1.4995,
+      "step": 36650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.23508359956727e-05,
+      "loss": 1.1098,
+      "step": 36660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.233732475034838e-05,
+      "loss": 1.0503,
+      "step": 36670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.232381146689907e-05,
+      "loss": 1.2744,
+      "step": 36680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.231029614655776e-05,
+      "loss": 1.1492,
+      "step": 36690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.229677879055765e-05,
+      "loss": 1.3286,
+      "step": 36700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.228325940013206e-05,
+      "loss": 1.1044,
+      "step": 36710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.226973797651456e-05,
+      "loss": 1.1734,
+      "step": 36720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.225621452093888e-05,
+      "loss": 1.0754,
+      "step": 36730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.224268903463896e-05,
+      "loss": 1.2066,
+      "step": 36740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.222916151884889e-05,
+      "loss": 1.2909,
+      "step": 36750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.221563197480296e-05,
+      "loss": 1.1018,
+      "step": 36760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.220210040373566e-05,
+      "loss": 1.1874,
+      "step": 36770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.218856680688164e-05,
+      "loss": 1.1103,
+      "step": 36780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.217503118547574e-05,
+      "loss": 1.423,
+      "step": 36790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.216149354075301e-05,
+      "loss": 1.4922,
+      "step": 36800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.214795387394866e-05,
+      "loss": 1.4544,
+      "step": 36810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.213441218629808e-05,
+      "loss": 1.087,
+      "step": 36820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.212086847903685e-05,
+      "loss": 1.2778,
+      "step": 36830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.210732275340078e-05,
+      "loss": 1.2039,
+      "step": 36840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.209377501062576e-05,
+      "loss": 1.3447,
+      "step": 36850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.208022525194798e-05,
+      "loss": 1.2708,
+      "step": 36860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.206667347860372e-05,
+      "loss": 1.2575,
+      "step": 36870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.205311969182951e-05,
+      "loss": 1.4718,
+      "step": 36880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.203956389286204e-05,
+      "loss": 1.384,
+      "step": 36890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.202600608293818e-05,
+      "loss": 1.1135,
+      "step": 36900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.201244626329497e-05,
+      "loss": 1.4001,
+      "step": 36910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.199888443516967e-05,
+      "loss": 1.1688,
+      "step": 36920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.198532059979967e-05,
+      "loss": 1.1732,
+      "step": 36930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.197175475842261e-05,
+      "loss": 1.3144,
+      "step": 36940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.195818691227628e-05,
+      "loss": 1.2133,
+      "step": 36950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.194461706259862e-05,
+      "loss": 1.3347,
+      "step": 36960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.193104521062782e-05,
+      "loss": 1.2667,
+      "step": 36970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.191747135760217e-05,
+      "loss": 1.3224,
+      "step": 36980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.190389550476024e-05,
+      "loss": 1.1556,
+      "step": 36990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.18903176533407e-05,
+      "loss": 1.2832,
+      "step": 37000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.187673780458244e-05,
+      "loss": 1.233,
+      "step": 37010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.186315595972453e-05,
+      "loss": 1.0699,
+      "step": 37020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.184957212000622e-05,
+      "loss": 1.324,
+      "step": 37030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.183598628666694e-05,
+      "loss": 1.1498,
+      "step": 37040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.182239846094629e-05,
+      "loss": 1.2845,
+      "step": 37050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.18088086440841e-05,
+      "loss": 1.2213,
+      "step": 37060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.179521683732029e-05,
+      "loss": 1.3878,
+      "step": 37070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.178162304189506e-05,
+      "loss": 1.3064,
+      "step": 37080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.176802725904876e-05,
+      "loss": 1.2849,
+      "step": 37090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.175442949002187e-05,
+      "loss": 1.1505,
+      "step": 37100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.17408297360551e-05,
+      "loss": 1.325,
+      "step": 37110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.172722799838935e-05,
+      "loss": 1.2663,
+      "step": 37120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.171362427826567e-05,
+      "loss": 1.5372,
+      "step": 37130
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.170001857692533e-05,
+      "loss": 1.2545,
+      "step": 37140
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.168641089560972e-05,
+      "loss": 1.2594,
+      "step": 37150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.167280123556047e-05,
+      "loss": 1.1489,
+      "step": 37160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.165918959801935e-05,
+      "loss": 1.3339,
+      "step": 37170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.164557598422834e-05,
+      "loss": 1.12,
+      "step": 37180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.163196039542958e-05,
+      "loss": 1.1718,
+      "step": 37190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.16183428328654e-05,
+      "loss": 1.4046,
+      "step": 37200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.160472329777832e-05,
+      "loss": 1.2872,
+      "step": 37210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.159110179141098e-05,
+      "loss": 1.2491,
+      "step": 37220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.157747831500632e-05,
+      "loss": 1.0425,
+      "step": 37230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.156385286980734e-05,
+      "loss": 1.195,
+      "step": 37240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.155022545705726e-05,
+      "loss": 1.2482,
+      "step": 37250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.153659607799951e-05,
+      "loss": 1.1739,
+      "step": 37260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.152296473387767e-05,
+      "loss": 1.1204,
+      "step": 37270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.15093314259355e-05,
+      "loss": 1.3818,
+      "step": 37280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.149569615541696e-05,
+      "loss": 1.3263,
+      "step": 37290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.148205892356616e-05,
+      "loss": 1.1836,
+      "step": 37300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.146841973162739e-05,
+      "loss": 1.3659,
+      "step": 37310
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.145477858084514e-05,
+      "loss": 1.2263,
+      "step": 37320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.14411354724641e-05,
+      "loss": 1.2881,
+      "step": 37330
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.142749040772905e-05,
+      "loss": 1.3868,
+      "step": 37340
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.141384338788504e-05,
+      "loss": 1.0782,
+      "step": 37350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.14001944141773e-05,
+      "loss": 1.1946,
+      "step": 37360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.138654348785114e-05,
+      "loss": 1.1812,
+      "step": 37370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.137289061015215e-05,
+      "loss": 1.384,
+      "step": 37380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.135923578232604e-05,
+      "loss": 1.1335,
+      "step": 37390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.134557900561873e-05,
+      "loss": 1.4056,
+      "step": 37400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.133192028127631e-05,
+      "loss": 1.1561,
+      "step": 37410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.131825961054504e-05,
+      "loss": 1.4686,
+      "step": 37420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.130459699467134e-05,
+      "loss": 1.2232,
+      "step": 37430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.129093243490186e-05,
+      "loss": 1.4046,
+      "step": 37440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.127726593248337e-05,
+      "loss": 1.0989,
+      "step": 37450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.126359748866287e-05,
+      "loss": 1.3128,
+      "step": 37460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.124992710468749e-05,
+      "loss": 1.2046,
+      "step": 37470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.123625478180456e-05,
+      "loss": 1.4009,
+      "step": 37480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.122258052126156e-05,
+      "loss": 1.3694,
+      "step": 37490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.120890432430622e-05,
+      "loss": 1.1301,
+      "step": 37500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.119522619218639e-05,
+      "loss": 1.2419,
+      "step": 37510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.118154612615006e-05,
+      "loss": 1.3203,
+      "step": 37520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.116786412744548e-05,
+      "loss": 1.2458,
+      "step": 37530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.115418019732102e-05,
+      "loss": 1.107,
+      "step": 37540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.114049433702525e-05,
+      "loss": 1.2188,
+      "step": 37550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.112680654780692e-05,
+      "loss": 1.302,
+      "step": 37560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.111311683091492e-05,
+      "loss": 1.1232,
+      "step": 37570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.109942518759835e-05,
+      "loss": 1.2322,
+      "step": 37580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.108573161910649e-05,
+      "loss": 1.3502,
+      "step": 37590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.107203612668877e-05,
+      "loss": 1.1995,
+      "step": 37600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.105833871159482e-05,
+      "loss": 1.3416,
+      "step": 37610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.104463937507442e-05,
+      "loss": 1.338,
+      "step": 37620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.103093811837754e-05,
+      "loss": 1.6498,
+      "step": 37630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.101723494275431e-05,
+      "loss": 1.438,
+      "step": 37640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.10035298494551e-05,
+      "loss": 1.4754,
+      "step": 37650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.098982283973036e-05,
+      "loss": 1.31,
+      "step": 37660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.097611391483077e-05,
+      "loss": 1.4978,
+      "step": 37670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.096240307600717e-05,
+      "loss": 1.2826,
+      "step": 37680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.094869032451058e-05,
+      "loss": 1.6213,
+      "step": 37690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.09349756615922e-05,
+      "loss": 1.3583,
+      "step": 37700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.09212590885034e-05,
+      "loss": 1.3246,
+      "step": 37710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.090754060649572e-05,
+      "loss": 1.1309,
+      "step": 37720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.089382021682084e-05,
+      "loss": 1.2324,
+      "step": 37730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.088009792073069e-05,
+      "loss": 1.2048,
+      "step": 37740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.086637371947734e-05,
+      "loss": 1.2455,
+      "step": 37750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.085264761431301e-05,
+      "loss": 1.1813,
+      "step": 37760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.08389196064901e-05,
+      "loss": 1.0894,
+      "step": 37770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.08251896972612e-05,
+      "loss": 1.2165,
+      "step": 37780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.081145788787909e-05,
+      "loss": 1.2214,
+      "step": 37790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.07977241795967e-05,
+      "loss": 1.187,
+      "step": 37800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.078398857366709e-05,
+      "loss": 1.3102,
+      "step": 37810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.07702510713436e-05,
+      "loss": 1.2209,
+      "step": 37820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.075651167387962e-05,
+      "loss": 0.9734,
+      "step": 37830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.074277038252883e-05,
+      "loss": 1.2992,
+      "step": 37840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.072902719854499e-05,
+      "loss": 1.4796,
+      "step": 37850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.07152821231821e-05,
+      "loss": 1.4592,
+      "step": 37860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.070153515769428e-05,
+      "loss": 1.0596,
+      "step": 37870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.068778630333583e-05,
+      "loss": 1.4112,
+      "step": 37880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.067403556136128e-05,
+      "loss": 1.2487,
+      "step": 37890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.066028293302525e-05,
+      "loss": 1.3506,
+      "step": 37900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.06465284195826e-05,
+      "loss": 1.1637,
+      "step": 37910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.063277202228831e-05,
+      "loss": 1.5547,
+      "step": 37920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.061901374239757e-05,
+      "loss": 1.4841,
+      "step": 37930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.060525358116572e-05,
+      "loss": 1.2703,
+      "step": 37940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.059149153984828e-05,
+      "loss": 1.4132,
+      "step": 37950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.057772761970097e-05,
+      "loss": 1.3561,
+      "step": 37960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.056396182197959e-05,
+      "loss": 1.2494,
+      "step": 37970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.05501941479402e-05,
+      "loss": 1.1008,
+      "step": 37980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.053642459883903e-05,
+      "loss": 1.0237,
+      "step": 37990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.052265317593244e-05,
+      "loss": 1.3784,
+      "step": 38000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.050887988047696e-05,
+      "loss": 1.4226,
+      "step": 38010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.049510471372933e-05,
+      "loss": 1.0991,
+      "step": 38020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.04813276769464e-05,
+      "loss": 1.2785,
+      "step": 38030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.046754877138528e-05,
+      "loss": 1.0058,
+      "step": 38040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.045376799830316e-05,
+      "loss": 1.2636,
+      "step": 38050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.043998535895747e-05,
+      "loss": 1.2941,
+      "step": 38060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.042620085460574e-05,
+      "loss": 1.202,
+      "step": 38070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.041241448650575e-05,
+      "loss": 1.3678,
+      "step": 38080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.03986262559154e-05,
+      "loss": 1.383,
+      "step": 38090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.038483616409274e-05,
+      "loss": 0.9835,
+      "step": 38100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.037104421229606e-05,
+      "loss": 1.0909,
+      "step": 38110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.035725040178376e-05,
+      "loss": 1.2183,
+      "step": 38120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.034345473381443e-05,
+      "loss": 1.1683,
+      "step": 38130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.032965720964682e-05,
+      "loss": 1.2521,
+      "step": 38140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.031585783053987e-05,
+      "loss": 1.2334,
+      "step": 38150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.030205659775268e-05,
+      "loss": 1.1959,
+      "step": 38160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.028825351254451e-05,
+      "loss": 1.312,
+      "step": 38170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.02744485761748e-05,
+      "loss": 1.2183,
+      "step": 38180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.026064178990313e-05,
+      "loss": 1.3987,
+      "step": 38190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.02468331549893e-05,
+      "loss": 1.2339,
+      "step": 38200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.023302267269328e-05,
+      "loss": 1.2404,
+      "step": 38210
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.021921034427509e-05,
+      "loss": 1.2197,
+      "step": 38220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.020539617099508e-05,
+      "loss": 1.3971,
+      "step": 38230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.01915801541137e-05,
+      "loss": 1.1385,
+      "step": 38240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.017776229489152e-05,
+      "loss": 1.1747,
+      "step": 38250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.016394259458936e-05,
+      "loss": 1.3953,
+      "step": 38260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.015012105446815e-05,
+      "loss": 1.208,
+      "step": 38270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.013629767578903e-05,
+      "loss": 1.2004,
+      "step": 38280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.012247245981327e-05,
+      "loss": 1.2513,
+      "step": 38290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.010864540780232e-05,
+      "loss": 1.2834,
+      "step": 38300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.009481652101783e-05,
+      "loss": 1.3833,
+      "step": 38310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.008098580072155e-05,
+      "loss": 1.3211,
+      "step": 38320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.006715324817548e-05,
+      "loss": 1.4195,
+      "step": 38330
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.00533188646417e-05,
+      "loss": 1.2599,
+      "step": 38340
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.003948265138252e-05,
+      "loss": 1.3718,
+      "step": 38350
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.002564460966043e-05,
+      "loss": 1.1053,
+      "step": 38360
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.001180474073801e-05,
+      "loss": 1.4744,
+      "step": 38370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.999796304587806e-05,
+      "loss": 1.2511,
+      "step": 38380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.998411952634353e-05,
+      "loss": 1.3381,
+      "step": 38390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.99702741833976e-05,
+      "loss": 1.2348,
+      "step": 38400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.995642701830348e-05,
+      "loss": 1.2599,
+      "step": 38410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.994257803232468e-05,
+      "loss": 1.4099,
+      "step": 38420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.992872722672483e-05,
+      "loss": 1.3039,
+      "step": 38430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.991487460276767e-05,
+      "loss": 1.2128,
+      "step": 38440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.99010201617172e-05,
+      "loss": 1.4061,
+      "step": 38450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.988716390483754e-05,
+      "loss": 1.4134,
+      "step": 38460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.987330583339294e-05,
+      "loss": 1.3917,
+      "step": 38470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.985944594864789e-05,
+      "loss": 1.3662,
+      "step": 38480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.984558425186699e-05,
+      "loss": 1.3866,
+      "step": 38490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.983172074431501e-05,
+      "loss": 1.2795,
+      "step": 38500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.981785542725693e-05,
+      "loss": 1.2584,
+      "step": 38510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.980398830195785e-05,
+      "loss": 1.2054,
+      "step": 38520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.979011936968305e-05,
+      "loss": 1.1822,
+      "step": 38530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.977624863169794e-05,
+      "loss": 1.0685,
+      "step": 38540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.97623760892682e-05,
+      "loss": 1.2506,
+      "step": 38550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.974850174365952e-05,
+      "loss": 1.3055,
+      "step": 38560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.97346255961379e-05,
+      "loss": 1.4225,
+      "step": 38570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.972074764796942e-05,
+      "loss": 1.4897,
+      "step": 38580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.970686790042033e-05,
+      "loss": 1.1962,
+      "step": 38590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.969298635475709e-05,
+      "loss": 1.2045,
+      "step": 38600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.967910301224627e-05,
+      "loss": 1.144,
+      "step": 38610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.966521787415465e-05,
+      "loss": 1.2091,
+      "step": 38620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.965133094174913e-05,
+      "loss": 1.2127,
+      "step": 38630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.96374422162968e-05,
+      "loss": 1.2075,
+      "step": 38640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.962355169906491e-05,
+      "loss": 1.2523,
+      "step": 38650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.960965939132089e-05,
+      "loss": 1.3342,
+      "step": 38660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.95957652943323e-05,
+      "loss": 1.1508,
+      "step": 38670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.958186940936688e-05,
+      "loss": 1.4154,
+      "step": 38680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.956797173769254e-05,
+      "loss": 1.0091,
+      "step": 38690
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.955407228057733e-05,
+      "loss": 1.1065,
+      "step": 38700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.954017103928949e-05,
+      "loss": 1.1108,
+      "step": 38710
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.952626801509743e-05,
+      "loss": 1.2405,
+      "step": 38720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.951236320926966e-05,
+      "loss": 1.1415,
+      "step": 38730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.949845662307493e-05,
+      "loss": 1.0715,
+      "step": 38740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.948454825778211e-05,
+      "loss": 1.1263,
+      "step": 38750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.947063811466023e-05,
+      "loss": 1.267,
+      "step": 38760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.945672619497853e-05,
+      "loss": 1.2791,
+      "step": 38770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.944281250000634e-05,
+      "loss": 1.359,
+      "step": 38780
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.942889703101319e-05,
+      "loss": 1.1779,
+      "step": 38790
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.941497978926878e-05,
+      "loss": 1.2239,
+      "step": 38800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.940106077604295e-05,
+      "loss": 1.3325,
+      "step": 38810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.938713999260574e-05,
+      "loss": 1.1363,
+      "step": 38820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.937321744022728e-05,
+      "loss": 1.2763,
+      "step": 38830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.935929312017796e-05,
+      "loss": 1.3162,
+      "step": 38840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.934536703372822e-05,
+      "loss": 1.4165,
+      "step": 38850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.933143918214876e-05,
+      "loss": 1.4522,
+      "step": 38860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.93175095667104e-05,
+      "loss": 1.2063,
+      "step": 38870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.930357818868409e-05,
+      "loss": 1.2706,
+      "step": 38880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.928964504934099e-05,
+      "loss": 1.4134,
+      "step": 38890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.927571014995238e-05,
+      "loss": 1.4629,
+      "step": 38900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.926177349178976e-05,
+      "loss": 1.1394,
+      "step": 38910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.924783507612474e-05,
+      "loss": 1.2628,
+      "step": 38920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.923389490422908e-05,
+      "loss": 1.3117,
+      "step": 38930
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.921995297737476e-05,
+      "loss": 1.3508,
+      "step": 38940
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.920600929683385e-05,
+      "loss": 1.3851,
+      "step": 38950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.919206386387862e-05,
+      "loss": 1.2411,
+      "step": 38960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.917811667978151e-05,
+      "loss": 1.2847,
+      "step": 38970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.916416774581509e-05,
+      "loss": 1.2791,
+      "step": 38980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.915021706325211e-05,
+      "loss": 1.4112,
+      "step": 38990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.913626463336547e-05,
+      "loss": 1.437,
+      "step": 39000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.912231045742822e-05,
+      "loss": 1.0508,
+      "step": 39010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.910835453671362e-05,
+      "loss": 1.1755,
+      "step": 39020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.909439687249499e-05,
+      "loss": 1.1998,
+      "step": 39030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.908043746604593e-05,
+      "loss": 1.227,
+      "step": 39040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.906647631864007e-05,
+      "loss": 1.2545,
+      "step": 39050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.905251343155135e-05,
+      "loss": 1.3315,
+      "step": 39060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.903854880605373e-05,
+      "loss": 1.3104,
+      "step": 39070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.90245824434214e-05,
+      "loss": 1.2865,
+      "step": 39080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.901061434492868e-05,
+      "loss": 0.994,
+      "step": 39090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.899664451185007e-05,
+      "loss": 1.1979,
+      "step": 39100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.898267294546023e-05,
+      "loss": 1.2603,
+      "step": 39110
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.896869964703397e-05,
+      "loss": 1.5188,
+      "step": 39120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.895472461784624e-05,
+      "loss": 1.4215,
+      "step": 39130
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.894074785917219e-05,
+      "loss": 1.3187,
+      "step": 39140
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.892676937228705e-05,
+      "loss": 1.0928,
+      "step": 39150
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.891278915846631e-05,
+      "loss": 1.0739,
+      "step": 39160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.889880721898553e-05,
+      "loss": 1.3793,
+      "step": 39170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.88848235551205e-05,
+      "loss": 1.3913,
+      "step": 39180
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.887083816814711e-05,
+      "loss": 1.2382,
+      "step": 39190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.885685105934143e-05,
+      "loss": 1.1174,
+      "step": 39200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.88428622299797e-05,
+      "loss": 1.3559,
+      "step": 39210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.882887168133829e-05,
+      "loss": 1.3627,
+      "step": 39220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.881487941469373e-05,
+      "loss": 1.1626,
+      "step": 39230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.880088543132275e-05,
+      "loss": 1.1884,
+      "step": 39240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.878688973250215e-05,
+      "loss": 1.2352,
+      "step": 39250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.877289231950901e-05,
+      "loss": 1.2119,
+      "step": 39260
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.875889319362043e-05,
+      "loss": 1.3129,
+      "step": 39270
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.874489235611378e-05,
+      "loss": 1.4021,
+      "step": 39280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.873088980826651e-05,
+      "loss": 1.1961,
+      "step": 39290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.871688555135627e-05,
+      "loss": 1.2347,
+      "step": 39300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.870287958666084e-05,
+      "loss": 1.2311,
+      "step": 39310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.868887191545816e-05,
+      "loss": 1.1066,
+      "step": 39320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.867486253902638e-05,
+      "loss": 1.3057,
+      "step": 39330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.86608514586437e-05,
+      "loss": 1.2619,
+      "step": 39340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.864683867558855e-05,
+      "loss": 1.1515,
+      "step": 39350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.863282419113952e-05,
+      "loss": 1.3945,
+      "step": 39360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.86188080065753e-05,
+      "loss": 1.293,
+      "step": 39370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.86047901231748e-05,
+      "loss": 1.3497,
+      "step": 39380
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.859077054221703e-05,
+      "loss": 1.3709,
+      "step": 39390
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.85767492649812e-05,
+      "loss": 1.1923,
+      "step": 39400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.856272629274665e-05,
+      "loss": 1.2464,
+      "step": 39410
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.854870162679286e-05,
+      "loss": 1.212,
+      "step": 39420
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.85346752683995e-05,
+      "loss": 1.3247,
+      "step": 39430
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.852064721884636e-05,
+      "loss": 1.2687,
+      "step": 39440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.850661747941342e-05,
+      "loss": 1.1797,
+      "step": 39450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.84925860513808e-05,
+      "loss": 1.4658,
+      "step": 39460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.847855293602874e-05,
+      "loss": 1.2345,
+      "step": 39470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.846451813463771e-05,
+      "loss": 1.3541,
+      "step": 39480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.845048164848823e-05,
+      "loss": 1.2558,
+      "step": 39490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.843644347886109e-05,
+      "loss": 1.2234,
+      "step": 39500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.842240362703714e-05,
+      "loss": 1.1507,
+      "step": 39510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.84083620942974e-05,
+      "loss": 1.3053,
+      "step": 39520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.839431888192314e-05,
+      "loss": 1.0501,
+      "step": 39530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.838027399119561e-05,
+      "loss": 1.2126,
+      "step": 39540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.836622742339636e-05,
+      "loss": 1.1734,
+      "step": 39550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.835217917980704e-05,
+      "loss": 1.113,
+      "step": 39560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.833812926170943e-05,
+      "loss": 1.2672,
+      "step": 39570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.832407767038554e-05,
+      "loss": 0.9944,
+      "step": 39580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.83100244071174e-05,
+      "loss": 1.3661,
+      "step": 39590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.829596947318734e-05,
+      "loss": 1.3116,
+      "step": 39600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.828191286987774e-05,
+      "loss": 1.3279,
+      "step": 39610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.826785459847119e-05,
+      "loss": 1.1673,
+      "step": 39620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.825379466025039e-05,
+      "loss": 1.4262,
+      "step": 39630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.823973305649822e-05,
+      "loss": 1.3877,
+      "step": 39640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.822566978849769e-05,
+      "loss": 1.2213,
+      "step": 39650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.821160485753198e-05,
+      "loss": 1.1957,
+      "step": 39660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.819753826488444e-05,
+      "loss": 1.3698,
+      "step": 39670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.818347001183854e-05,
+      "loss": 1.2701,
+      "step": 39680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.816940009967787e-05,
+      "loss": 1.3569,
+      "step": 39690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.815532852968627e-05,
+      "loss": 1.2205,
+      "step": 39700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.814125530314762e-05,
+      "loss": 1.2081,
+      "step": 39710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.812718042134604e-05,
+      "loss": 1.2491,
+      "step": 39720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.811310388556575e-05,
+      "loss": 1.1939,
+      "step": 39730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.809902569709116e-05,
+      "loss": 1.0998,
+      "step": 39740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.808494585720677e-05,
+      "loss": 1.4934,
+      "step": 39750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.807086436719727e-05,
+      "loss": 1.5166,
+      "step": 39760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.805678122834752e-05,
+      "loss": 1.2678,
+      "step": 39770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.80426964419425e-05,
+      "loss": 1.1983,
+      "step": 39780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.802861000926735e-05,
+      "loss": 1.3412,
+      "step": 39790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.801452193160735e-05,
+      "loss": 1.2861,
+      "step": 39800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.800043221024794e-05,
+      "loss": 1.3857,
+      "step": 39810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.798634084647471e-05,
+      "loss": 1.1595,
+      "step": 39820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.797224784157343e-05,
+      "loss": 1.358,
+      "step": 39830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.795815319682993e-05,
+      "loss": 1.3683,
+      "step": 39840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.79440569135303e-05,
+      "loss": 1.2679,
+      "step": 39850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.792995899296068e-05,
+      "loss": 1.2655,
+      "step": 39860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.791585943640743e-05,
+      "loss": 1.197,
+      "step": 39870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.790175824515706e-05,
+      "loss": 1.4707,
+      "step": 39880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.788765542049618e-05,
+      "loss": 1.0409,
+      "step": 39890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.787355096371158e-05,
+      "loss": 1.4909,
+      "step": 39900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.785944487609018e-05,
+      "loss": 1.4464,
+      "step": 39910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.784533715891908e-05,
+      "loss": 1.3778,
+      "step": 39920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.783122781348549e-05,
+      "loss": 1.1798,
+      "step": 39930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.781711684107682e-05,
+      "loss": 1.3959,
+      "step": 39940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.780300424298059e-05,
+      "loss": 1.4152,
+      "step": 39950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.778889002048443e-05,
+      "loss": 1.2672,
+      "step": 39960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.777477417487623e-05,
+      "loss": 1.204,
+      "step": 39970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.776065670744392e-05,
+      "loss": 1.2221,
+      "step": 39980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.774653761947566e-05,
+      "loss": 1.3792,
+      "step": 39990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.773241691225966e-05,
+      "loss": 1.2647,
+      "step": 40000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.771829458708439e-05,
+      "loss": 1.3701,
+      "step": 40010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.770417064523838e-05,
+      "loss": 1.3269,
+      "step": 40020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.769004508801036e-05,
+      "loss": 1.4423,
+      "step": 40030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.767591791668918e-05,
+      "loss": 1.3957,
+      "step": 40040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.766178913256383e-05,
+      "loss": 1.169,
+      "step": 40050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.764765873692351e-05,
+      "loss": 1.135,
+      "step": 40060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.763352673105747e-05,
+      "loss": 1.2183,
+      "step": 40070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.761939311625518e-05,
+      "loss": 1.2422,
+      "step": 40080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.760525789380625e-05,
+      "loss": 1.2527,
+      "step": 40090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.759112106500039e-05,
+      "loss": 1.276,
+      "step": 40100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.757698263112749e-05,
+      "loss": 1.0917,
+      "step": 40110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.756284259347759e-05,
+      "loss": 1.3492,
+      "step": 40120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.754870095334087e-05,
+      "loss": 1.4433,
+      "step": 40130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.753455771200769e-05,
+      "loss": 1.3454,
+      "step": 40140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.752041287076845e-05,
+      "loss": 1.1753,
+      "step": 40150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.750626643091382e-05,
+      "loss": 1.2438,
+      "step": 40160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.749211839373456e-05,
+      "loss": 1.2434,
+      "step": 40170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.747796876052155e-05,
+      "loss": 1.3341,
+      "step": 40180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.74638175325659e-05,
+      "loss": 1.1336,
+      "step": 40190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.744966471115874e-05,
+      "loss": 1.4933,
+      "step": 40200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.743551029759149e-05,
+      "loss": 1.1898,
+      "step": 40210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.742135429315558e-05,
+      "loss": 1.3852,
+      "step": 40220
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.740719669914268e-05,
+      "loss": 1.208,
+      "step": 40230
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.739303751684455e-05,
+      "loss": 1.2554,
+      "step": 40240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.737887674755314e-05,
+      "loss": 1.2452,
+      "step": 40250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.736471439256053e-05,
+      "loss": 1.2203,
+      "step": 40260
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.73505504531589e-05,
+      "loss": 1.144,
+      "step": 40270
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.73363849306406e-05,
+      "loss": 1.3685,
+      "step": 40280
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.732221782629822e-05,
+      "loss": 1.1862,
+      "step": 40290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.730804914142431e-05,
+      "loss": 1.3339,
+      "step": 40300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.729387887731171e-05,
+      "loss": 1.2099,
+      "step": 40310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.727970703525336e-05,
+      "loss": 1.2817,
+      "step": 40320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.726553361654234e-05,
+      "loss": 1.2084,
+      "step": 40330
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.725135862247189e-05,
+      "loss": 1.2536,
+      "step": 40340
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.723718205433532e-05,
+      "loss": 1.2808,
+      "step": 40350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.72230039134262e-05,
+      "loss": 1.3546,
+      "step": 40360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.720882420103816e-05,
+      "loss": 1.2609,
+      "step": 40370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.719464291846503e-05,
+      "loss": 1.0784,
+      "step": 40380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.718046006700072e-05,
+      "loss": 1.2334,
+      "step": 40390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.716627564793933e-05,
+      "loss": 1.1831,
+      "step": 40400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.715208966257509e-05,
+      "loss": 1.4768,
+      "step": 40410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.713790211220235e-05,
+      "loss": 1.1456,
+      "step": 40420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.712371299811567e-05,
+      "loss": 1.3119,
+      "step": 40430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.710952232160968e-05,
+      "loss": 1.2658,
+      "step": 40440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.709533008397917e-05,
+      "loss": 1.2845,
+      "step": 40450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.70811362865191e-05,
+      "loss": 1.1218,
+      "step": 40460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.706694093052455e-05,
+      "loss": 1.1409,
+      "step": 40470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.705274401729076e-05,
+      "loss": 1.3249,
+      "step": 40480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.703854554811308e-05,
+      "loss": 0.9968,
+      "step": 40490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.702434552428703e-05,
+      "loss": 1.2545,
+      "step": 40500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.701014394710825e-05,
+      "loss": 1.366,
+      "step": 40510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.699594081787253e-05,
+      "loss": 1.2327,
+      "step": 40520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.698173613787586e-05,
+      "loss": 1.2537,
+      "step": 40530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.696752990841427e-05,
+      "loss": 1.2219,
+      "step": 40540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.695332213078398e-05,
+      "loss": 1.3962,
+      "step": 40550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.693911280628136e-05,
+      "loss": 1.0257,
+      "step": 40560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.692490193620291e-05,
+      "loss": 1.4341,
+      "step": 40570
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.691068952184527e-05,
+      "loss": 1.3693,
+      "step": 40580
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.689647556450525e-05,
+      "loss": 1.2385,
+      "step": 40590
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.688226006547973e-05,
+      "loss": 1.2903,
+      "step": 40600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.686804302606582e-05,
+      "loss": 1.2775,
+      "step": 40610
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.685382444756068e-05,
+      "loss": 1.2731,
+      "step": 40620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.683960433126168e-05,
+      "loss": 1.1435,
+      "step": 40630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.682538267846632e-05,
+      "loss": 1.2533,
+      "step": 40640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.68111594904722e-05,
+      "loss": 1.2658,
+      "step": 40650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.679693476857711e-05,
+      "loss": 1.2501,
+      "step": 40660
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.678270851407893e-05,
+      "loss": 1.3579,
+      "step": 40670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.676848072827574e-05,
+      "loss": 1.1907,
+      "step": 40680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.675425141246571e-05,
+      "loss": 1.1671,
+      "step": 40690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.674002056794715e-05,
+      "loss": 1.1534,
+      "step": 40700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.672578819601854e-05,
+      "loss": 1.0911,
+      "step": 40710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.671155429797848e-05,
+      "loss": 1.4156,
+      "step": 40720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.669731887512574e-05,
+      "loss": 1.2791,
+      "step": 40730
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.668308192875916e-05,
+      "loss": 1.1994,
+      "step": 40740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.666884346017779e-05,
+      "loss": 1.3724,
+      "step": 40750
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.665460347068078e-05,
+      "loss": 1.192,
+      "step": 40760
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.664036196156742e-05,
+      "loss": 1.2747,
+      "step": 40770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.662611893413718e-05,
+      "loss": 1.2047,
+      "step": 40780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.661187438968962e-05,
+      "loss": 1.2685,
+      "step": 40790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.659762832952445e-05,
+      "loss": 1.3452,
+      "step": 40800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.658338075494151e-05,
+      "loss": 1.2788,
+      "step": 40810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.656913166724082e-05,
+      "loss": 1.0796,
+      "step": 40820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.655488106772251e-05,
+      "loss": 1.2695,
+      "step": 40830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.654062895768682e-05,
+      "loss": 1.3074,
+      "step": 40840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.65263753384342e-05,
+      "loss": 1.2268,
+      "step": 40850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.651212021126513e-05,
+      "loss": 1.4,
+      "step": 40860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.649786357748033e-05,
+      "loss": 1.2685,
+      "step": 40870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.648360543838063e-05,
+      "loss": 1.3608,
+      "step": 40880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.646934579526696e-05,
+      "loss": 1.1392,
+      "step": 40890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.645508464944044e-05,
+      "loss": 1.1834,
+      "step": 40900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.644082200220226e-05,
+      "loss": 1.3361,
+      "step": 40910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.642655785485383e-05,
+      "loss": 1.2986,
+      "step": 40920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.641229220869663e-05,
+      "loss": 1.2302,
+      "step": 40930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.639802506503228e-05,
+      "loss": 1.2555,
+      "step": 40940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.638375642516261e-05,
+      "loss": 1.1616,
+      "step": 40950
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.63694862903895e-05,
+      "loss": 1.6905,
+      "step": 40960
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.6355214662015e-05,
+      "loss": 1.2973,
+      "step": 40970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.634094154134129e-05,
+      "loss": 1.3576,
+      "step": 40980
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.632666692967073e-05,
+      "loss": 1.3233,
+      "step": 40990
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.631239082830575e-05,
+      "loss": 1.2005,
+      "step": 41000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.629811323854893e-05,
+      "loss": 1.0917,
+      "step": 41010
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.628383416170302e-05,
+      "loss": 1.1465,
+      "step": 41020
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.62695535990709e-05,
+      "loss": 1.2488,
+      "step": 41030
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.625527155195556e-05,
+      "loss": 1.2618,
+      "step": 41040
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.624098802166012e-05,
+      "loss": 1.2121,
+      "step": 41050
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.622670300948788e-05,
+      "loss": 1.2653,
+      "step": 41060
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.621241651674222e-05,
+      "loss": 1.3095,
+      "step": 41070
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.61981285447267e-05,
+      "loss": 1.1277,
+      "step": 41080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.6183839094745e-05,
+      "loss": 1.413,
+      "step": 41090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.616954816810093e-05,
+      "loss": 1.2168,
+      "step": 41100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.615525576609842e-05,
+      "loss": 1.1659,
+      "step": 41110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.614096189004158e-05,
+      "loss": 1.157,
+      "step": 41120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.61266665412346e-05,
+      "loss": 1.5425,
+      "step": 41130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.611236972098185e-05,
+      "loss": 1.375,
+      "step": 41140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.609807143058782e-05,
+      "loss": 1.1938,
+      "step": 41150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.60837716713571e-05,
+      "loss": 1.1169,
+      "step": 41160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.606947044459446e-05,
+      "loss": 1.3525,
+      "step": 41170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.605516775160479e-05,
+      "loss": 1.3685,
+      "step": 41180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.604086359369312e-05,
+      "loss": 1.1121,
+      "step": 41190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.602655797216459e-05,
+      "loss": 1.2515,
+      "step": 41200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.601225088832448e-05,
+      "loss": 1.3406,
+      "step": 41210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.599794234347825e-05,
+      "loss": 1.3324,
+      "step": 41220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.59836323389314e-05,
+      "loss": 1.322,
+      "step": 41230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.596932087598966e-05,
+      "loss": 1.3533,
+      "step": 41240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.595500795595884e-05,
+      "loss": 1.2722,
+      "step": 41250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.594069358014489e-05,
+      "loss": 1.5871,
+      "step": 41260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.592637774985389e-05,
+      "loss": 1.2411,
+      "step": 41270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.591206046639207e-05,
+      "loss": 1.405,
+      "step": 41280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.589774173106578e-05,
+      "loss": 1.101,
+      "step": 41290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.58834215451815e-05,
+      "loss": 1.0542,
+      "step": 41300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.586909991004586e-05,
+      "loss": 1.3303,
+      "step": 41310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.58547768269656e-05,
+      "loss": 1.5177,
+      "step": 41320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.58404522972476e-05,
+      "loss": 1.4671,
+      "step": 41330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.582612632219888e-05,
+      "loss": 1.2435,
+      "step": 41340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.581179890312658e-05,
+      "loss": 1.3567,
+      "step": 41350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.579747004133797e-05,
+      "loss": 1.2182,
+      "step": 41360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.578313973814048e-05,
+      "loss": 1.4492,
+      "step": 41370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.576880799484162e-05,
+      "loss": 1.2936,
+      "step": 41380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.575447481274909e-05,
+      "loss": 1.2785,
+      "step": 41390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.574014019317069e-05,
+      "loss": 1.3432,
+      "step": 41400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.572580413741433e-05,
+      "loss": 1.3113,
+      "step": 41410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.571146664678809e-05,
+      "loss": 1.2227,
+      "step": 41420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.569712772260017e-05,
+      "loss": 1.2008,
+      "step": 41430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.56827873661589e-05,
+      "loss": 1.3844,
+      "step": 41440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.566844557877271e-05,
+      "loss": 1.2153,
+      "step": 41450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.565410236175024e-05,
+      "loss": 1.2094,
+      "step": 41460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.563975771640013e-05,
+      "loss": 1.3363,
+      "step": 41470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.562541164403132e-05,
+      "loss": 1.1893,
+      "step": 41480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.561106414595273e-05,
+      "loss": 1.3546,
+      "step": 41490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.559671522347348e-05,
+      "loss": 1.2487,
+      "step": 41500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.558236487790282e-05,
+      "loss": 1.2374,
+      "step": 41510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.556801311055011e-05,
+      "loss": 1.2347,
+      "step": 41520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.555365992272484e-05,
+      "loss": 1.2922,
+      "step": 41530
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.553930531573667e-05,
+      "loss": 1.3913,
+      "step": 41540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.552494929089533e-05,
+      "loss": 1.2582,
+      "step": 41550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.55105918495107e-05,
+      "loss": 1.35,
+      "step": 41560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.54962329928928e-05,
+      "loss": 1.2943,
+      "step": 41570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.54818727223518e-05,
+      "loss": 1.2681,
+      "step": 41580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.546751103919797e-05,
+      "loss": 1.2389,
+      "step": 41590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.545314794474169e-05,
+      "loss": 1.1527,
+      "step": 41600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.543878344029351e-05,
+      "loss": 1.3653,
+      "step": 41610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.542441752716408e-05,
+      "loss": 1.065,
+      "step": 41620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.54100502066642e-05,
+      "loss": 1.3513,
+      "step": 41630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.539568148010476e-05,
+      "loss": 1.3165,
+      "step": 41640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.538131134879687e-05,
+      "loss": 1.2392,
+      "step": 41650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.536693981405164e-05,
+      "loss": 1.6127,
+      "step": 41660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.535256687718039e-05,
+      "loss": 1.0471,
+      "step": 41670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.533819253949456e-05,
+      "loss": 1.1159,
+      "step": 41680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.532381680230571e-05,
+      "loss": 1.388,
+      "step": 41690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.530943966692553e-05,
+      "loss": 1.4454,
+      "step": 41700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.529506113466579e-05,
+      "loss": 1.1383,
+      "step": 41710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.52806812068385e-05,
+      "loss": 1.6383,
+      "step": 41720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.526629988475567e-05,
+      "loss": 1.4252,
+      "step": 41730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.52519171697295e-05,
+      "loss": 1.2667,
+      "step": 41740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.523753306307235e-05,
+      "loss": 1.3625,
+      "step": 41750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.522314756609665e-05,
+      "loss": 1.242,
+      "step": 41760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.520876068011498e-05,
+      "loss": 1.2406,
+      "step": 41770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.519437240644002e-05,
+      "loss": 1.1479,
+      "step": 41780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.517998274638462e-05,
+      "loss": 1.1082,
+      "step": 41790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.516559170126174e-05,
+      "loss": 1.5322,
+      "step": 41800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.515119927238446e-05,
+      "loss": 1.3253,
+      "step": 41810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.513680546106598e-05,
+      "loss": 1.3413,
+      "step": 41820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.512241026861963e-05,
+      "loss": 1.3393,
+      "step": 41830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.510801369635888e-05,
+      "loss": 1.1759,
+      "step": 41840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.509361574559733e-05,
+      "loss": 1.2599,
+      "step": 41850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.507921641764867e-05,
+      "loss": 1.2834,
+      "step": 41860
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.506481571382674e-05,
+      "loss": 1.2137,
+      "step": 41870
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.505041363544553e-05,
+      "loss": 1.2684,
+      "step": 41880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.50360101838191e-05,
+      "loss": 1.2873,
+      "step": 41890
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.502160536026167e-05,
+      "loss": 1.3106,
+      "step": 41900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.500719916608759e-05,
+      "loss": 1.3021,
+      "step": 41910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.499279160261132e-05,
+      "loss": 1.3459,
+      "step": 41920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.497838267114744e-05,
+      "loss": 1.1842,
+      "step": 41930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.496397237301068e-05,
+      "loss": 1.2094,
+      "step": 41940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.494956070951586e-05,
+      "loss": 1.2897,
+      "step": 41950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.493514768197798e-05,
+      "loss": 1.4416,
+      "step": 41960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.492073329171208e-05,
+      "loss": 1.1711,
+      "step": 41970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.49063175400334e-05,
+      "loss": 1.144,
+      "step": 41980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.489190042825729e-05,
+      "loss": 1.2351,
+      "step": 41990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.487748195769918e-05,
+      "loss": 1.2817,
+      "step": 42000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.486306212967467e-05,
+      "loss": 1.178,
+      "step": 42010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.484864094549949e-05,
+      "loss": 1.2329,
+      "step": 42020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.483421840648941e-05,
+      "loss": 1.2751,
+      "step": 42030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.481979451396046e-05,
+      "loss": 1.064,
+      "step": 42040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.480536926922867e-05,
+      "loss": 1.4848,
+      "step": 42050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.479094267361025e-05,
+      "loss": 1.0586,
+      "step": 42060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.477651472842154e-05,
+      "loss": 1.2421,
+      "step": 42070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.4762085434979e-05,
+      "loss": 1.2807,
+      "step": 42080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.474765479459917e-05,
+      "loss": 1.4826,
+      "step": 42090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.473322280859877e-05,
+      "loss": 1.2489,
+      "step": 42100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.47187894782946e-05,
+      "loss": 1.3564,
+      "step": 42110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.470435480500362e-05,
+      "loss": 1.2247,
+      "step": 42120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.468991879004287e-05,
+      "loss": 1.2853,
+      "step": 42130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.467548143472955e-05,
+      "loss": 1.2756,
+      "step": 42140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.466104274038099e-05,
+      "loss": 1.4013,
+      "step": 42150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.464660270831458e-05,
+      "loss": 1.2552,
+      "step": 42160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.463216133984788e-05,
+      "loss": 1.2458,
+      "step": 42170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.461771863629859e-05,
+      "loss": 1.1625,
+      "step": 42180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.460327459898448e-05,
+      "loss": 1.3145,
+      "step": 42190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.458882922922348e-05,
+      "loss": 1.1647,
+      "step": 42200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.457438252833362e-05,
+      "loss": 1.2926,
+      "step": 42210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.455993449763307e-05,
+      "loss": 1.3975,
+      "step": 42220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.454548513844011e-05,
+      "loss": 1.56,
+      "step": 42230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.453103445207313e-05,
+      "loss": 1.3542,
+      "step": 42240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.451658243985068e-05,
+      "loss": 1.2432,
+      "step": 42250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.45021291030914e-05,
+      "loss": 1.3524,
+      "step": 42260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.448767444311404e-05,
+      "loss": 1.3627,
+      "step": 42270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.447321846123748e-05,
+      "loss": 1.2118,
+      "step": 42280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.445876115878075e-05,
+      "loss": 1.2069,
+      "step": 42290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.444430253706296e-05,
+      "loss": 1.1955,
+      "step": 42300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.442984259740339e-05,
+      "loss": 1.5169,
+      "step": 42310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.441538134112139e-05,
+      "loss": 1.1561,
+      "step": 42320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.440091876953641e-05,
+      "loss": 1.2089,
+      "step": 42330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.438645488396812e-05,
+      "loss": 1.2732,
+      "step": 42340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.43719896857362e-05,
+      "loss": 1.1662,
+      "step": 42350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.435752317616056e-05,
+      "loss": 1.306,
+      "step": 42360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.43430553565611e-05,
+      "loss": 1.2564,
+      "step": 42370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.432858622825794e-05,
+      "loss": 1.3372,
+      "step": 42380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.43141157925713e-05,
+      "loss": 1.4228,
+      "step": 42390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.429964405082145e-05,
+      "loss": 1.3389,
+      "step": 42400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.428517100432891e-05,
+      "loss": 1.2351,
+      "step": 42410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.42706966544142e-05,
+      "loss": 1.489,
+      "step": 42420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.425622100239803e-05,
+      "loss": 1.2079,
+      "step": 42430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.424174404960116e-05,
+      "loss": 1.4864,
+      "step": 42440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.422726579734454e-05,
+      "loss": 1.3409,
+      "step": 42450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.421278624694922e-05,
+      "loss": 1.3894,
+      "step": 42460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.419830539973633e-05,
+      "loss": 1.1874,
+      "step": 42470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.418382325702717e-05,
+      "loss": 1.3227,
+      "step": 42480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.416933982014311e-05,
+      "loss": 1.2721,
+      "step": 42490
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.415485509040569e-05,
+      "loss": 1.1866,
+      "step": 42500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.414036906913653e-05,
+      "loss": 1.186,
+      "step": 42510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.412588175765735e-05,
+      "loss": 1.3188,
+      "step": 42520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.411139315729007e-05,
+      "loss": 1.0726,
+      "step": 42530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.409690326935663e-05,
+      "loss": 1.3239,
+      "step": 42540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.408241209517916e-05,
+      "loss": 1.1542,
+      "step": 42550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.406791963607986e-05,
+      "loss": 1.2571,
+      "step": 42560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.405342589338106e-05,
+      "loss": 1.2236,
+      "step": 42570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.403893086840524e-05,
+      "loss": 1.2927,
+      "step": 42580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.402443456247495e-05,
+      "loss": 1.2065,
+      "step": 42590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.400993697691286e-05,
+      "loss": 1.3056,
+      "step": 42600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.399543811304182e-05,
+      "loss": 1.3691,
+      "step": 42610
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.398093797218471e-05,
+      "loss": 1.3675,
+      "step": 42620
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.396643655566459e-05,
+      "loss": 1.4081,
+      "step": 42630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.395193386480458e-05,
+      "loss": 1.4473,
+      "step": 42640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.3937429900928e-05,
+      "loss": 1.3952,
+      "step": 42650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.392292466535821e-05,
+      "loss": 1.1626,
+      "step": 42660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.39084181594187e-05,
+      "loss": 1.1766,
+      "step": 42670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.38939103844331e-05,
+      "loss": 1.1239,
+      "step": 42680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.387940134172512e-05,
+      "loss": 1.141,
+      "step": 42690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.386489103261866e-05,
+      "loss": 1.2631,
+      "step": 42700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.385037945843763e-05,
+      "loss": 1.2475,
+      "step": 42710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.383586662050615e-05,
+      "loss": 1.1423,
+      "step": 42720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.38213525201484e-05,
+      "loss": 1.1956,
+      "step": 42730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.380683715868868e-05,
+      "loss": 1.4148,
+      "step": 42740
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.379232053745143e-05,
+      "loss": 1.3062,
+      "step": 42750
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.377780265776119e-05,
+      "loss": 1.181,
+      "step": 42760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.376328352094261e-05,
+      "loss": 1.3373,
+      "step": 42770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.374876312832045e-05,
+      "loss": 1.0841,
+      "step": 42780
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.37342414812196e-05,
+      "loss": 1.2136,
+      "step": 42790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.371971858096508e-05,
+      "loss": 1.321,
+      "step": 42800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.370519442888198e-05,
+      "loss": 1.0532,
+      "step": 42810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.369066902629554e-05,
+      "loss": 1.4423,
+      "step": 42820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.367614237453109e-05,
+      "loss": 1.0898,
+      "step": 42830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.366161447491409e-05,
+      "loss": 1.2214,
+      "step": 42840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.364708532877011e-05,
+      "loss": 1.4078,
+      "step": 42850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.363255493742485e-05,
+      "loss": 1.2411,
+      "step": 42860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.361802330220408e-05,
+      "loss": 1.5121,
+      "step": 42870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.360349042443372e-05,
+      "loss": 1.274,
+      "step": 42880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.35889563054398e-05,
+      "loss": 0.9965,
+      "step": 42890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.357442094654844e-05,
+      "loss": 1.4703,
+      "step": 42900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.355988434908591e-05,
+      "loss": 1.2478,
+      "step": 42910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.354534651437858e-05,
+      "loss": 1.2753,
+      "step": 42920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.353080744375289e-05,
+      "loss": 1.2868,
+      "step": 42930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.351626713853546e-05,
+      "loss": 1.5025,
+      "step": 42940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.350172560005297e-05,
+      "loss": 1.1236,
+      "step": 42950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.348718282963225e-05,
+      "loss": 1.2961,
+      "step": 42960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.347263882860024e-05,
+      "loss": 1.2968,
+      "step": 42970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.345809359828394e-05,
+      "loss": 1.2247,
+      "step": 42980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.344354714001053e-05,
+      "loss": 1.3518,
+      "step": 42990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.342899945510726e-05,
+      "loss": 1.218,
+      "step": 43000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.34144505449015e-05,
+      "loss": 1.4158,
+      "step": 43010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.339990041072076e-05,
+      "loss": 1.2148,
+      "step": 43020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.338534905389263e-05,
+      "loss": 1.2138,
+      "step": 43030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.33707964757448e-05,
+      "loss": 1.4384,
+      "step": 43040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.335624267760511e-05,
+      "loss": 1.2074,
+      "step": 43050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.334168766080146e-05,
+      "loss": 1.1725,
+      "step": 43060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.332713142666196e-05,
+      "loss": 1.6475,
+      "step": 43070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.33125739765147e-05,
+      "loss": 1.2646,
+      "step": 43080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.329801531168798e-05,
+      "loss": 1.3386,
+      "step": 43090
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.328345543351017e-05,
+      "loss": 1.2757,
+      "step": 43100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.326889434330974e-05,
+      "loss": 1.1713,
+      "step": 43110
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.325433204241533e-05,
+      "loss": 1.0476,
+      "step": 43120
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.323976853215558e-05,
+      "loss": 1.2316,
+      "step": 43130
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.322520381385939e-05,
+      "loss": 1.1872,
+      "step": 43140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.321063788885563e-05,
+      "loss": 1.2907,
+      "step": 43150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.319607075847336e-05,
+      "loss": 1.2398,
+      "step": 43160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.318150242404173e-05,
+      "loss": 1.3694,
+      "step": 43170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.316693288688998e-05,
+      "loss": 1.4406,
+      "step": 43180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.31523621483475e-05,
+      "loss": 1.3017,
+      "step": 43190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.313779020974378e-05,
+      "loss": 1.1724,
+      "step": 43200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.312321707240837e-05,
+      "loss": 1.2338,
+      "step": 43210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.310864273767101e-05,
+      "loss": 1.313,
+      "step": 43220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.309406720686146e-05,
+      "loss": 1.2621,
+      "step": 43230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.307949048130966e-05,
+      "loss": 1.3434,
+      "step": 43240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.306491256234564e-05,
+      "loss": 1.24,
+      "step": 43250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.305033345129952e-05,
+      "loss": 1.3646,
+      "step": 43260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.303575314950157e-05,
+      "loss": 1.3708,
+      "step": 43270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.302117165828209e-05,
+      "loss": 1.3543,
+      "step": 43280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.300658897897158e-05,
+      "loss": 1.0116,
+      "step": 43290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.29920051129006e-05,
+      "loss": 1.4362,
+      "step": 43300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.297742006139981e-05,
+      "loss": 1.4389,
+      "step": 43310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.296283382580004e-05,
+      "loss": 1.3645,
+      "step": 43320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.29482464074321e-05,
+      "loss": 1.1605,
+      "step": 43330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.293365780762708e-05,
+      "loss": 1.1094,
+      "step": 43340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.291906802771603e-05,
+      "loss": 1.3623,
+      "step": 43350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.290447706903017e-05,
+      "loss": 1.0782,
+      "step": 43360
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.288988493290087e-05,
+      "loss": 1.3556,
+      "step": 43370
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.28752916206595e-05,
+      "loss": 1.2664,
+      "step": 43380
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.286069713363763e-05,
+      "loss": 1.2094,
+      "step": 43390
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.284610147316689e-05,
+      "loss": 1.1874,
+      "step": 43400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.283150464057905e-05,
+      "loss": 1.4138,
+      "step": 43410
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.281690663720597e-05,
+      "loss": 1.4448,
+      "step": 43420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.280230746437961e-05,
+      "loss": 1.4742,
+      "step": 43430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.278770712343203e-05,
+      "loss": 1.4536,
+      "step": 43440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.27731056156954e-05,
+      "loss": 1.299,
+      "step": 43450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.275850294250204e-05,
+      "loss": 1.2921,
+      "step": 43460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.274389910518433e-05,
+      "loss": 1.2291,
+      "step": 43470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.272929410507476e-05,
+      "loss": 1.4033,
+      "step": 43480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.271468794350595e-05,
+      "loss": 1.4159,
+      "step": 43490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.270008062181056e-05,
+      "loss": 1.2164,
+      "step": 43500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.268547214132148e-05,
+      "loss": 1.0954,
+      "step": 43510
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.26708625033716e-05,
+      "loss": 1.4972,
+      "step": 43520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.265625170929393e-05,
+      "loss": 1.3133,
+      "step": 43530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.264163976042163e-05,
+      "loss": 1.3053,
+      "step": 43540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.26270266580879e-05,
+      "loss": 1.2617,
+      "step": 43550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.261241240362612e-05,
+      "loss": 1.2006,
+      "step": 43560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.259779699836972e-05,
+      "loss": 1.2123,
+      "step": 43570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.25831804436523e-05,
+      "loss": 1.316,
+      "step": 43580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.256856274080746e-05,
+      "loss": 1.228,
+      "step": 43590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.255394389116897e-05,
+      "loss": 1.1667,
+      "step": 43600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.253932389607072e-05,
+      "loss": 1.4566,
+      "step": 43610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.252470275684668e-05,
+      "loss": 1.3635,
+      "step": 43620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.251008047483092e-05,
+      "loss": 1.2669,
+      "step": 43630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.249545705135761e-05,
+      "loss": 1.2264,
+      "step": 43640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.248083248776107e-05,
+      "loss": 1.5117,
+      "step": 43650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.246620678537564e-05,
+      "loss": 1.1437,
+      "step": 43660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.245157994553585e-05,
+      "loss": 1.2719,
+      "step": 43670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.243695196957631e-05,
+      "loss": 1.2965,
+      "step": 43680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.242232285883168e-05,
+      "loss": 1.1071,
+      "step": 43690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.240769261463677e-05,
+      "loss": 1.2552,
+      "step": 43700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.239306123832653e-05,
+      "loss": 1.103,
+      "step": 43710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.23784287312359e-05,
+      "loss": 1.4342,
+      "step": 43720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.236379509470007e-05,
+      "loss": 1.1748,
+      "step": 43730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.234916033005421e-05,
+      "loss": 1.3993,
+      "step": 43740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.233452443863365e-05,
+      "loss": 1.2131,
+      "step": 43750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.231988742177383e-05,
+      "loss": 1.2789,
+      "step": 43760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.230524928081023e-05,
+      "loss": 1.3823,
+      "step": 43770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.229061001707856e-05,
+      "loss": 1.2152,
+      "step": 43780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.227596963191446e-05,
+      "loss": 1.3213,
+      "step": 43790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.226132812665381e-05,
+      "loss": 1.251,
+      "step": 43800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.224668550263254e-05,
+      "loss": 1.1005,
+      "step": 43810
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.223204176118668e-05,
+      "loss": 1.0784,
+      "step": 43820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.22173969036524e-05,
+      "loss": 1.2598,
+      "step": 43830
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.220275093136589e-05,
+      "loss": 1.3252,
+      "step": 43840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.218810384566352e-05,
+      "loss": 1.4837,
+      "step": 43850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.217345564788175e-05,
+      "loss": 1.1643,
+      "step": 43860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.215880633935708e-05,
+      "loss": 1.5224,
+      "step": 43870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.214415592142622e-05,
+      "loss": 1.1262,
+      "step": 43880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.212950439542586e-05,
+      "loss": 1.2885,
+      "step": 43890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.211485176269288e-05,
+      "loss": 1.2138,
+      "step": 43900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.210019802456423e-05,
+      "loss": 1.2473,
+      "step": 43910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.208554318237694e-05,
+      "loss": 1.138,
+      "step": 43920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.20708872374682e-05,
+      "loss": 1.2687,
+      "step": 43930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.205623019117522e-05,
+      "loss": 1.2513,
+      "step": 43940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.204157204483538e-05,
+      "loss": 1.382,
+      "step": 43950
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.202691279978613e-05,
+      "loss": 1.4971,
+      "step": 43960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.201225245736502e-05,
+      "loss": 1.4358,
+      "step": 43970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.199759101890971e-05,
+      "loss": 1.4085,
+      "step": 43980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.198292848575794e-05,
+      "loss": 1.2412,
+      "step": 43990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.196826485924758e-05,
+      "loss": 1.3564,
+      "step": 44000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.195360014071656e-05,
+      "loss": 1.1461,
+      "step": 44010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.193893433150297e-05,
+      "loss": 1.3831,
+      "step": 44020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.192426743294494e-05,
+      "loss": 1.3223,
+      "step": 44030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.190959944638072e-05,
+      "loss": 1.2222,
+      "step": 44040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.189493037314867e-05,
+      "loss": 1.2465,
+      "step": 44050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.188026021458722e-05,
+      "loss": 1.2473,
+      "step": 44060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.186558897203494e-05,
+      "loss": 1.3676,
+      "step": 44070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.185091664683049e-05,
+      "loss": 1.2396,
+      "step": 44080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.183624324031259e-05,
+      "loss": 1.2457,
+      "step": 44090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.182156875382009e-05,
+      "loss": 1.0893,
+      "step": 44100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.180689318869195e-05,
+      "loss": 1.1623,
+      "step": 44110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.17922165462672e-05,
+      "loss": 1.3069,
+      "step": 44120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1777538827885e-05,
+      "loss": 1.2221,
+      "step": 44130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.176286003488456e-05,
+      "loss": 1.1399,
+      "step": 44140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.174818016860525e-05,
+      "loss": 1.1009,
+      "step": 44150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.173349923038647e-05,
+      "loss": 1.3744,
+      "step": 44160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.171881722156778e-05,
+      "loss": 1.168,
+      "step": 44170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.17041341434888e-05,
+      "loss": 1.3851,
+      "step": 44180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.168944999748926e-05,
+      "loss": 1.1353,
+      "step": 44190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.167476478490899e-05,
+      "loss": 1.0066,
+      "step": 44200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.16600785070879e-05,
+      "loss": 1.293,
+      "step": 44210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.164539116536603e-05,
+      "loss": 1.2638,
+      "step": 44220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.163070276108349e-05,
+      "loss": 1.2328,
+      "step": 44230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.161601329558048e-05,
+      "loss": 1.3789,
+      "step": 44240
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.160132277019735e-05,
+      "loss": 1.4871,
+      "step": 44250
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.158663118627445e-05,
+      "loss": 1.1888,
+      "step": 44260
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.157193854515233e-05,
+      "loss": 1.358,
+      "step": 44270
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.155724484817157e-05,
+      "loss": 1.2292,
+      "step": 44280
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.154255009667289e-05,
+      "loss": 1.3027,
+      "step": 44290
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.152785429199703e-05,
+      "loss": 1.4257,
+      "step": 44300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.151315743548494e-05,
+      "loss": 1.2583,
+      "step": 44310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.14984595284776e-05,
+      "loss": 1.4678,
+      "step": 44320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.148376057231605e-05,
+      "loss": 1.253,
+      "step": 44330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.14690605683415e-05,
+      "loss": 1.2914,
+      "step": 44340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.14543595178952e-05,
+      "loss": 1.3877,
+      "step": 44350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.143965742231855e-05,
+      "loss": 1.3844,
+      "step": 44360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.142495428295298e-05,
+      "loss": 1.2758,
+      "step": 44370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.141025010114007e-05,
+      "loss": 1.2928,
+      "step": 44380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.139554487822149e-05,
+      "loss": 1.365,
+      "step": 44390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.138083861553894e-05,
+      "loss": 1.2985,
+      "step": 44400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.136613131443431e-05,
+      "loss": 1.6401,
+      "step": 44410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.13514229762495e-05,
+      "loss": 1.274,
+      "step": 44420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.133671360232657e-05,
+      "loss": 1.2728,
+      "step": 44430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.132200319400766e-05,
+      "loss": 1.3959,
+      "step": 44440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.130729175263494e-05,
+      "loss": 1.1998,
+      "step": 44450
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.129257927955078e-05,
+      "loss": 1.0954,
+      "step": 44460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.127786577609757e-05,
+      "loss": 1.3108,
+      "step": 44470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.126315124361781e-05,
+      "loss": 1.1588,
+      "step": 44480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.124843568345412e-05,
+      "loss": 1.3148,
+      "step": 44490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.123371909694914e-05,
+      "loss": 1.5334,
+      "step": 44500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.121900148544572e-05,
+      "loss": 1.4102,
+      "step": 44510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.12042828502867e-05,
+      "loss": 1.3504,
+      "step": 44520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.118956319281504e-05,
+      "loss": 1.2352,
+      "step": 44530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.117484251437388e-05,
+      "loss": 1.3443,
+      "step": 44540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.116012081630629e-05,
+      "loss": 1.3207,
+      "step": 44550
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.114539809995556e-05,
+      "loss": 1.2148,
+      "step": 44560
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.113067436666504e-05,
+      "loss": 1.3666,
+      "step": 44570
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.111594961777817e-05,
+      "loss": 1.0812,
+      "step": 44580
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.110122385463848e-05,
+      "loss": 1.2021,
+      "step": 44590
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.108649707858957e-05,
+      "loss": 1.185,
+      "step": 44600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.107176929097517e-05,
+      "loss": 1.2596,
+      "step": 44610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.105704049313911e-05,
+      "loss": 1.1397,
+      "step": 44620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.104231068642525e-05,
+      "loss": 1.2869,
+      "step": 44630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.102757987217762e-05,
+      "loss": 1.2228,
+      "step": 44640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.101284805174028e-05,
+      "loss": 1.3499,
+      "step": 44650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.099811522645742e-05,
+      "loss": 1.2121,
+      "step": 44660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.098338139767328e-05,
+      "loss": 1.4148,
+      "step": 44670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.0968646566732245e-05,
+      "loss": 1.3671,
+      "step": 44680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.095391073497877e-05,
+      "loss": 1.3327,
+      "step": 44690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.093917390375738e-05,
+      "loss": 1.1579,
+      "step": 44700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.092443607441272e-05,
+      "loss": 1.1987,
+      "step": 44710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.090969724828949e-05,
+      "loss": 1.523,
+      "step": 44720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.089495742673254e-05,
+      "loss": 1.0621,
+      "step": 44730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.088021661108675e-05,
+      "loss": 1.2743,
+      "step": 44740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.086547480269713e-05,
+      "loss": 1.377,
+      "step": 44750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.085073200290876e-05,
+      "loss": 1.4048,
+      "step": 44760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.083598821306682e-05,
+      "loss": 1.2827,
+      "step": 44770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.082124343451658e-05,
+      "loss": 1.2124,
+      "step": 44780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.0806497668603403e-05,
+      "loss": 1.3011,
+      "step": 44790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.0791750916672716e-05,
+      "loss": 1.4225,
+      "step": 44800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.077700318007009e-05,
+      "loss": 1.2,
+      "step": 44810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.076225446014111e-05,
+      "loss": 1.1264,
+      "step": 44820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.0747504758231536e-05,
+      "loss": 1.1907,
+      "step": 44830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.073275407568716e-05,
+      "loss": 1.1106,
+      "step": 44840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.071800241385387e-05,
+      "loss": 1.2078,
+      "step": 44850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.070324977407767e-05,
+      "loss": 1.1648,
+      "step": 44860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.06884961577046e-05,
+      "loss": 1.2619,
+      "step": 44870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.0673741566080854e-05,
+      "loss": 1.131,
+      "step": 44880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.06589860005527e-05,
+      "loss": 1.3873,
+      "step": 44890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.0644229462466454e-05,
+      "loss": 1.3012,
+      "step": 44900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.062947195316855e-05,
+      "loss": 1.4414,
+      "step": 44910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.06147134740055e-05,
+      "loss": 1.1923,
+      "step": 44920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.059995402632394e-05,
+      "loss": 1.1917,
+      "step": 44930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.058519361147055e-05,
+      "loss": 1.3279,
+      "step": 44940
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.057043223079211e-05,
+      "loss": 1.3363,
+      "step": 44950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.0555669885635493e-05,
+      "loss": 1.1519,
+      "step": 44960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.054090657734767e-05,
+      "loss": 1.4093,
+      "step": 44970
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.052614230727569e-05,
+      "loss": 1.29,
+      "step": 44980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.0511377076766686e-05,
+      "loss": 1.0764,
+      "step": 44990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.049661088716787e-05,
+      "loss": 1.2517,
+      "step": 45000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.048184373982656e-05,
+      "loss": 1.2806,
+      "step": 45010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.0467075636090174e-05,
+      "loss": 1.2712,
+      "step": 45020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.045230657730618e-05,
+      "loss": 1.3277,
+      "step": 45030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.043753656482216e-05,
+      "loss": 1.2622,
+      "step": 45040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.0422765599985764e-05,
+      "loss": 1.1962,
+      "step": 45050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.040799368414474e-05,
+      "loss": 1.2458,
+      "step": 45060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.0393220818646946e-05,
+      "loss": 1.2178,
+      "step": 45070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.037844700484028e-05,
+      "loss": 1.3134,
+      "step": 45080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.036367224407274e-05,
+      "loss": 1.226,
+      "step": 45090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.0348896537692446e-05,
+      "loss": 1.1045,
+      "step": 45100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.033411988704756e-05,
+      "loss": 1.0974,
+      "step": 45110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.031934229348636e-05,
+      "loss": 1.3956,
+      "step": 45120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.0304563758357194e-05,
+      "loss": 1.2923,
+      "step": 45130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.028978428300849e-05,
+      "loss": 1.0647,
+      "step": 45140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.0275003868788795e-05,
+      "loss": 1.3697,
+      "step": 45150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.026022251704668e-05,
+      "loss": 1.0517,
+      "step": 45160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.0245440229130876e-05,
+      "loss": 1.217,
+      "step": 45170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.023065700639017e-05,
+      "loss": 1.3212,
+      "step": 45180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.02158728501734e-05,
+      "loss": 1.2887,
+      "step": 45190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.020108776182952e-05,
+      "loss": 1.1666,
+      "step": 45200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.018630174270757e-05,
+      "loss": 1.2987,
+      "step": 45210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.017151479415668e-05,
+      "loss": 1.4526,
+      "step": 45220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.015672691752605e-05,
+      "loss": 1.288,
+      "step": 45230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.0141938114164976e-05,
+      "loss": 1.1831,
+      "step": 45240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.012714838542284e-05,
+      "loss": 1.1942,
+      "step": 45250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.011235773264906e-05,
+      "loss": 1.3256,
+      "step": 45260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.009756615719323e-05,
+      "loss": 1.3161,
+      "step": 45270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.008277366040493e-05,
+      "loss": 1.2851,
+      "step": 45280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.006798024363394e-05,
+      "loss": 1.4009,
+      "step": 45290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.005318590822999e-05,
+      "loss": 1.3426,
+      "step": 45300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.0038390655543e-05,
+      "loss": 1.1841,
+      "step": 45310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.002359448692291e-05,
+      "loss": 1.132,
+      "step": 45320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.000879740371976e-05,
+      "loss": 1.4012,
+      "step": 45330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.9993999407283717e-05,
+      "loss": 1.4221,
+      "step": 45340
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.9979200498964975e-05,
+      "loss": 1.2859,
+      "step": 45350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.996440068011383e-05,
+      "loss": 1.2323,
+      "step": 45360
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.994959995208066e-05,
+      "loss": 1.3655,
+      "step": 45370
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.993479831621592e-05,
+      "loss": 1.1984,
+      "step": 45380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.9919995773870173e-05,
+      "loss": 1.3473,
+      "step": 45390
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.990519232639406e-05,
+      "loss": 1.4132,
+      "step": 45400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.9890387975138254e-05,
+      "loss": 1.1861,
+      "step": 45410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.987558272145358e-05,
+      "loss": 1.468,
+      "step": 45420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.986077656669089e-05,
+      "loss": 1.2724,
+      "step": 45430
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.984596951220116e-05,
+      "loss": 1.3259,
+      "step": 45440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.983116155933543e-05,
+      "loss": 1.2835,
+      "step": 45450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.981635270944481e-05,
+      "loss": 1.4084,
+      "step": 45460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.980154296388053e-05,
+      "loss": 1.2773,
+      "step": 45470
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.978673232399383e-05,
+      "loss": 1.3504,
+      "step": 45480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.977192079113613e-05,
+      "loss": 1.27,
+      "step": 45490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.9757108366658844e-05,
+      "loss": 1.1241,
+      "step": 45500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.974229505191351e-05,
+      "loss": 1.2899,
+      "step": 45510
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.972748084825175e-05,
+      "loss": 1.4009,
+      "step": 45520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.971266575702523e-05,
+      "loss": 1.3871,
+      "step": 45530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.969784977958577e-05,
+      "loss": 1.226,
+      "step": 45540
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.968303291728519e-05,
+      "loss": 1.238,
+      "step": 45550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.966821517147543e-05,
+      "loss": 1.0819,
+      "step": 45560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.96533965435085e-05,
+      "loss": 1.1433,
+      "step": 45570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9638577034736495e-05,
+      "loss": 1.3714,
+      "step": 45580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.962375664651163e-05,
+      "loss": 1.2654,
+      "step": 45590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.960893538018612e-05,
+      "loss": 1.2269,
+      "step": 45600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.959411323711231e-05,
+      "loss": 1.0243,
+      "step": 45610
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.957929021864262e-05,
+      "loss": 1.5088,
+      "step": 45620
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9564466326129574e-05,
+      "loss": 1.2709,
+      "step": 45630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9549641560925704e-05,
+      "loss": 1.2747,
+      "step": 45640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9534815924383704e-05,
+      "loss": 1.2387,
+      "step": 45650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9519989417856284e-05,
+      "loss": 1.4057,
+      "step": 45660
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.950516204269625e-05,
+      "loss": 1.2609,
+      "step": 45670
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9490333800256534e-05,
+      "loss": 1.3847,
+      "step": 45680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.947550469189008e-05,
+      "loss": 1.4188,
+      "step": 45690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.946067471894996e-05,
+      "loss": 1.4945,
+      "step": 45700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.94458438827893e-05,
+      "loss": 1.366,
+      "step": 45710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.943101218476129e-05,
+      "loss": 1.3879,
+      "step": 45720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.941617962621925e-05,
+      "loss": 1.0624,
+      "step": 45730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.940134620851653e-05,
+      "loss": 1.1251,
+      "step": 45740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.938651193300658e-05,
+      "loss": 1.2618,
+      "step": 45750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.937167680104292e-05,
+      "loss": 1.1823,
+      "step": 45760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.935684081397915e-05,
+      "loss": 1.3304,
+      "step": 45770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.934200397316896e-05,
+      "loss": 1.1374,
+      "step": 45780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9327166279966104e-05,
+      "loss": 1.2547,
+      "step": 45790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.931232773572443e-05,
+      "loss": 1.229,
+      "step": 45800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9297488341797825e-05,
+      "loss": 1.1247,
+      "step": 45810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9282648099540295e-05,
+      "loss": 1.3211,
+      "step": 45820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.926780701030591e-05,
+      "loss": 1.1822,
+      "step": 45830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.925296507544882e-05,
+      "loss": 1.4273,
+      "step": 45840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9238122296323226e-05,
+      "loss": 1.2723,
+      "step": 45850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9223278674283463e-05,
+      "loss": 1.3384,
+      "step": 45860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9208434210683874e-05,
+      "loss": 1.1264,
+      "step": 45870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9193588906878924e-05,
+      "loss": 1.2568,
+      "step": 45880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.917874276422315e-05,
+      "loss": 1.2978,
+      "step": 45890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9163895784071176e-05,
+      "loss": 1.1092,
+      "step": 45900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.914904796777765e-05,
+      "loss": 1.432,
+      "step": 45910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.913419931669735e-05,
+      "loss": 1.2197,
+      "step": 45920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.91193498321851e-05,
+      "loss": 1.2969,
+      "step": 45930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.910449951559581e-05,
+      "loss": 1.2116,
+      "step": 45940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9089648368284513e-05,
+      "loss": 1.3858,
+      "step": 45950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9074796391606216e-05,
+      "loss": 1.1823,
+      "step": 45960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.905994358691608e-05,
+      "loss": 1.4225,
+      "step": 45970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.904508995556932e-05,
+      "loss": 1.3216,
+      "step": 45980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9030235498921215e-05,
+      "loss": 1.4712,
+      "step": 45990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.901538021832716e-05,
+      "loss": 1.5056,
+      "step": 46000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.900052411514257e-05,
+      "loss": 1.457,
+      "step": 46010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.898566719072297e-05,
+      "loss": 1.2062,
+      "step": 46020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8970809446423944e-05,
+      "loss": 1.4465,
+      "step": 46030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.895595088360116e-05,
+      "loss": 1.2134,
+      "step": 46040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.894109150361038e-05,
+      "loss": 1.2719,
+      "step": 46050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8926231307807365e-05,
+      "loss": 1.3344,
+      "step": 46060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.891137029754806e-05,
+      "loss": 1.4586,
+      "step": 46070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8896508474188407e-05,
+      "loss": 1.4692,
+      "step": 46080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.888164583908443e-05,
+      "loss": 1.2117,
+      "step": 46090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.886678239359227e-05,
+      "loss": 1.3408,
+      "step": 46100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8851918139068086e-05,
+      "loss": 1.2608,
+      "step": 46110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.883705307686814e-05,
+      "loss": 1.1391,
+      "step": 46120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.882218720834879e-05,
+      "loss": 1.2741,
+      "step": 46130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8807320534866405e-05,
+      "loss": 1.1477,
+      "step": 46140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.879245305777751e-05,
+      "loss": 1.1418,
+      "step": 46150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.877758477843862e-05,
+      "loss": 1.1826,
+      "step": 46160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8762715698206385e-05,
+      "loss": 1.2785,
+      "step": 46170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.87478458184375e-05,
+      "loss": 1.1505,
+      "step": 46180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.873297514048871e-05,
+      "loss": 1.2996,
+      "step": 46190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.87181036657169e-05,
+      "loss": 1.0374,
+      "step": 46200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8703231395478974e-05,
+      "loss": 1.2424,
+      "step": 46210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.868835833113192e-05,
+      "loss": 1.3822,
+      "step": 46220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.86734844740328e-05,
+      "loss": 1.1697,
+      "step": 46230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.865860982553876e-05,
+      "loss": 1.4085,
+      "step": 46240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.864373438700699e-05,
+      "loss": 1.2021,
+      "step": 46250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.862885815979479e-05,
+      "loss": 1.332,
+      "step": 46260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.86139811452595e-05,
+      "loss": 1.275,
+      "step": 46270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.859910334475853e-05,
+      "loss": 1.296,
+      "step": 46280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.858422475964941e-05,
+      "loss": 1.2308,
+      "step": 46290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.856934539128969e-05,
+      "loss": 1.2633,
+      "step": 46300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8554465241037e-05,
+      "loss": 1.3655,
+      "step": 46310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.853958431024906e-05,
+      "loss": 1.1736,
+      "step": 46320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.852470260028364e-05,
+      "loss": 1.3742,
+      "step": 46330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.85098201124986e-05,
+      "loss": 1.2078,
+      "step": 46340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.849493684825188e-05,
+      "loss": 0.969,
+      "step": 46350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.848005280890145e-05,
+      "loss": 1.4297,
+      "step": 46360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.846516799580538e-05,
+      "loss": 1.2545,
+      "step": 46370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.84502824103218e-05,
+      "loss": 1.3457,
+      "step": 46380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8435396053808936e-05,
+      "loss": 1.1778,
+      "step": 46390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.842050892762504e-05,
+      "loss": 1.2627,
+      "step": 46400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8405621033128474e-05,
+      "loss": 1.2545,
+      "step": 46410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.839073237167766e-05,
+      "loss": 1.2288,
+      "step": 46420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.837584294463105e-05,
+      "loss": 1.4969,
+      "step": 46430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.836095275334724e-05,
+      "loss": 1.2948,
+      "step": 46440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.834606179918485e-05,
+      "loss": 1.1889,
+      "step": 46450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.833117008350256e-05,
+      "loss": 1.2606,
+      "step": 46460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.831627760765914e-05,
+      "loss": 1.3676,
+      "step": 46470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8301384373013415e-05,
+      "loss": 1.1777,
+      "step": 46480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.828649038092431e-05,
+      "loss": 1.4203,
+      "step": 46490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8271595632750795e-05,
+      "loss": 1.2019,
+      "step": 46500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.825670012985189e-05,
+      "loss": 1.2948,
+      "step": 46510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.824180387358674e-05,
+      "loss": 1.2322,
+      "step": 46520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.822690686531449e-05,
+      "loss": 1.2143,
+      "step": 46530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.821200910639441e-05,
+      "loss": 1.3434,
+      "step": 46540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.819711059818579e-05,
+      "loss": 1.1232,
+      "step": 46550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.818221134204806e-05,
+      "loss": 1.3862,
+      "step": 46560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.8167311339340634e-05,
+      "loss": 1.3379,
+      "step": 46570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.8152410591423045e-05,
+      "loss": 1.2829,
+      "step": 46580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.8137509099654885e-05,
+      "loss": 1.0943,
+      "step": 46590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.812260686539581e-05,
+      "loss": 1.3873,
+      "step": 46600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.810770389000556e-05,
+      "loss": 1.2307,
+      "step": 46610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.8092800174843884e-05,
+      "loss": 1.3081,
+      "step": 46620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.8077895721270694e-05,
+      "loss": 1.2491,
+      "step": 46630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.806299053064589e-05,
+      "loss": 1.2393,
+      "step": 46640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.804808460432945e-05,
+      "loss": 1.2828,
+      "step": 46650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.8033177943681496e-05,
+      "loss": 1.1027,
+      "step": 46660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.801827055006208e-05,
+      "loss": 1.4728,
+      "step": 46670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.800336242483145e-05,
+      "loss": 1.1929,
+      "step": 46680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.798845356934985e-05,
+      "loss": 1.283,
+      "step": 46690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.797354398497761e-05,
+      "loss": 1.5657,
+      "step": 46700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.795863367307515e-05,
+      "loss": 1.2292,
+      "step": 46710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.794372263500287e-05,
+      "loss": 1.2504,
+      "step": 46720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7928810872121366e-05,
+      "loss": 1.3442,
+      "step": 46730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.791389838579119e-05,
+      "loss": 1.2235,
+      "step": 46740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.789898517737301e-05,
+      "loss": 1.2179,
+      "step": 46750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.788407124822759e-05,
+      "loss": 1.1723,
+      "step": 46760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7869156599715655e-05,
+      "loss": 1.4367,
+      "step": 46770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7854241233198116e-05,
+      "loss": 1.4473,
+      "step": 46780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.783932515003587e-05,
+      "loss": 1.3816,
+      "step": 46790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.782440835158991e-05,
+      "loss": 1.2566,
+      "step": 46800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.780949083922131e-05,
+      "loss": 1.132,
+      "step": 46810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.779457261429115e-05,
+      "loss": 1.2748,
+      "step": 46820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.777965367816066e-05,
+      "loss": 1.3042,
+      "step": 46830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7764734032191046e-05,
+      "loss": 1.3043,
+      "step": 46840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7749813677743644e-05,
+      "loss": 1.1063,
+      "step": 46850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.773489261617985e-05,
+      "loss": 1.296,
+      "step": 46860
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.771997084886105e-05,
+      "loss": 1.3258,
+      "step": 46870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.770504837714881e-05,
+      "loss": 1.3864,
+      "step": 46880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.769012520240467e-05,
+      "loss": 1.3008,
+      "step": 46890
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.767520132599027e-05,
+      "loss": 1.1481,
+      "step": 46900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.766027674926733e-05,
+      "loss": 1.3018,
+      "step": 46910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7645351473597576e-05,
+      "loss": 1.2343,
+      "step": 46920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.763042550034287e-05,
+      "loss": 1.1662,
+      "step": 46930
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7615498830865066e-05,
+      "loss": 1.4862,
+      "step": 46940
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.760057146652616e-05,
+      "loss": 1.172,
+      "step": 46950
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.758564340868815e-05,
+      "loss": 1.1812,
+      "step": 46960
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.757071465871312e-05,
+      "loss": 1.1832,
+      "step": 46970
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.755578521796321e-05,
+      "loss": 1.3304,
+      "step": 46980
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.754085508780063e-05,
+      "loss": 1.3528,
+      "step": 46990
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.752592426958764e-05,
+      "loss": 1.1995,
+      "step": 47000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.751099276468659e-05,
+      "loss": 1.3179,
+      "step": 47010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7496060574459877e-05,
+      "loss": 1.2798,
+      "step": 47020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.748112770026994e-05,
+      "loss": 1.2668,
+      "step": 47030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.74661941434793e-05,
+      "loss": 1.1862,
+      "step": 47040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.745125990545056e-05,
+      "loss": 1.3306,
+      "step": 47050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7436324987546365e-05,
+      "loss": 1.3527,
+      "step": 47060
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.74213893911294e-05,
+      "loss": 1.2552,
+      "step": 47070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.740645311756245e-05,
+      "loss": 1.3138,
+      "step": 47080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.739151616820833e-05,
+      "loss": 1.1528,
+      "step": 47090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.737657854442996e-05,
+      "loss": 1.3394,
+      "step": 47100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.736164024759028e-05,
+      "loss": 1.3675,
+      "step": 47110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.734670127905228e-05,
+      "loss": 1.1041,
+      "step": 47120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.7331761640179084e-05,
+      "loss": 1.2755,
+      "step": 47130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.731682133233378e-05,
+      "loss": 1.1449,
+      "step": 47140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.730188035687961e-05,
+      "loss": 1.2011,
+      "step": 47150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.7286938715179814e-05,
+      "loss": 1.314,
+      "step": 47160
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.727199640859773e-05,
+      "loss": 1.3122,
+      "step": 47170
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.72570534384967e-05,
+      "loss": 1.308,
+      "step": 47180
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.724210980624018e-05,
+      "loss": 1.3599,
+      "step": 47190
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.7227165513191714e-05,
+      "loss": 1.058,
+      "step": 47200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.72122205607148e-05,
+      "loss": 1.2585,
+      "step": 47210
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.719727495017313e-05,
+      "loss": 1.2368,
+      "step": 47220
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.718232868293033e-05,
+      "loss": 1.1802,
+      "step": 47230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.716738176035015e-05,
+      "loss": 0.9619,
+      "step": 47240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.7152434183796424e-05,
+      "loss": 1.2784,
+      "step": 47250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.713748595463298e-05,
+      "loss": 1.3124,
+      "step": 47260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.712253707422377e-05,
+      "loss": 1.3039,
+      "step": 47270
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.710758754393274e-05,
+      "loss": 1.5125,
+      "step": 47280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.709263736512396e-05,
+      "loss": 1.0233,
+      "step": 47290
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.7077686539161515e-05,
+      "loss": 1.0014,
+      "step": 47300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.706273506740956e-05,
+      "loss": 1.1457,
+      "step": 47310
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.704778295123234e-05,
+      "loss": 1.4484,
+      "step": 47320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.7032830191994094e-05,
+      "loss": 1.2449,
+      "step": 47330
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.7017876791059176e-05,
+      "loss": 1.2105,
+      "step": 47340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.700292274979199e-05,
+      "loss": 1.4322,
+      "step": 47350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.698796806955695e-05,
+      "loss": 1.2236,
+      "step": 47360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.6973012751718624e-05,
+      "loss": 1.1892,
+      "step": 47370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.695805679764153e-05,
+      "loss": 1.5505,
+      "step": 47380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.694310020869031e-05,
+      "loss": 1.1827,
+      "step": 47390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.692814298622966e-05,
+      "loss": 1.1189,
+      "step": 47400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.691318513162429e-05,
+      "loss": 1.2189,
+      "step": 47410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.689822664623906e-05,
+      "loss": 1.2616,
+      "step": 47420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.6883267531438765e-05,
+      "loss": 1.2732,
+      "step": 47430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.686830778858835e-05,
+      "loss": 1.2872,
+      "step": 47440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.685334741905278e-05,
+      "loss": 1.278,
+      "step": 47450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.6838386424197084e-05,
+      "loss": 1.3536,
+      "step": 47460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.682342480538637e-05,
+      "loss": 1.2619,
+      "step": 47470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.680846256398574e-05,
+      "loss": 1.2299,
+      "step": 47480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.679349970136042e-05,
+      "loss": 1.3944,
+      "step": 47490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.6778536218875676e-05,
+      "loss": 1.2899,
+      "step": 47500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.676357211789679e-05,
+      "loss": 1.2917,
+      "step": 47510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.674860739978917e-05,
+      "loss": 1.061,
+      "step": 47520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.67336420659182e-05,
+      "loss": 1.1928,
+      "step": 47530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.6718676117649404e-05,
+      "loss": 1.3854,
+      "step": 47540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.6703709556348284e-05,
+      "loss": 0.9909,
+      "step": 47550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.6688742383380454e-05,
+      "loss": 1.3523,
+      "step": 47560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.667377460011158e-05,
+      "loss": 1.2299,
+      "step": 47570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.665880620790733e-05,
+      "loss": 1.2417,
+      "step": 47580
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.664383720813349e-05,
+      "loss": 1.3399,
+      "step": 47590
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.6628867602155863e-05,
+      "loss": 1.0172,
+      "step": 47600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.6613897391340343e-05,
+      "loss": 1.3082,
+      "step": 47610
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.659892657705283e-05,
+      "loss": 1.3448,
+      "step": 47620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.658395516065933e-05,
+      "loss": 1.2026,
+      "step": 47630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.656898314352588e-05,
+      "loss": 1.2983,
+      "step": 47640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.655401052701853e-05,
+      "loss": 1.2483,
+      "step": 47650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.6539037312503485e-05,
+      "loss": 1.2147,
+      "step": 47660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.652406350134691e-05,
+      "loss": 1.1359,
+      "step": 47670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.650908909491508e-05,
+      "loss": 1.2807,
+      "step": 47680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.649411409457429e-05,
+      "loss": 1.4558,
+      "step": 47690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.6479138501690896e-05,
+      "loss": 1.4579,
+      "step": 47700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.6464162317631355e-05,
+      "loss": 1.2372,
+      "step": 47710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.6449185543762105e-05,
+      "loss": 0.9895,
+      "step": 47720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.6434208181449674e-05,
+      "loss": 1.2814,
+      "step": 47730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.641923023206066e-05,
+      "loss": 1.2588,
+      "step": 47740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.640425169696167e-05,
+      "loss": 1.2651,
+      "step": 47750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.6389272577519426e-05,
+      "loss": 1.1983,
+      "step": 47760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.637429287510064e-05,
+      "loss": 1.2406,
+      "step": 47770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.635931259107212e-05,
+      "loss": 1.1165,
+      "step": 47780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.634433172680071e-05,
+      "loss": 1.5445,
+      "step": 47790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.6329350283653295e-05,
+      "loss": 1.3053,
+      "step": 47800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.631436826299684e-05,
+      "loss": 1.1891,
+      "step": 47810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.6299385666198356e-05,
+      "loss": 1.1232,
+      "step": 47820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.628440249462489e-05,
+      "loss": 1.15,
+      "step": 47830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.626941874964354e-05,
+      "loss": 1.2495,
+      "step": 47840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.6254434432621496e-05,
+      "loss": 1.4001,
+      "step": 47850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.623944954492595e-05,
+      "loss": 1.321,
+      "step": 47860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.622446408792419e-05,
+      "loss": 1.0874,
+      "step": 47870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.620947806298351e-05,
+      "loss": 1.2628,
+      "step": 47880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.61944914714713e-05,
+      "loss": 1.2606,
+      "step": 47890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.617950431475496e-05,
+      "loss": 1.3286,
+      "step": 47900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.6164516594201986e-05,
+      "loss": 1.2909,
+      "step": 47910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.6149528311179886e-05,
+      "loss": 1.2387,
+      "step": 47920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.613453946705627e-05,
+      "loss": 1.3549,
+      "step": 47930
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.6119550063198714e-05,
+      "loss": 1.2433,
+      "step": 47940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.6104560100974935e-05,
+      "loss": 1.2987,
+      "step": 47950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.6089569581752644e-05,
+      "loss": 1.4034,
+      "step": 47960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.607457850689962e-05,
+      "loss": 1.2956,
+      "step": 47970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.605958687778372e-05,
+      "loss": 1.5,
+      "step": 47980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.604459469577279e-05,
+      "loss": 1.1507,
+      "step": 47990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.602960196223478e-05,
+      "loss": 1.3989,
+      "step": 48000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.601460867853767e-05,
+      "loss": 1.3516,
+      "step": 48010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.59996148460495e-05,
+      "loss": 1.2525,
+      "step": 48020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.5984620466138346e-05,
+      "loss": 1.3429,
+      "step": 48030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.596962554017233e-05,
+      "loss": 1.3668,
+      "step": 48040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.5954630069519657e-05,
+      "loss": 1.2796,
+      "step": 48050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.593963405554854e-05,
+      "loss": 1.2219,
+      "step": 48060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.5924637499627254e-05,
+      "loss": 1.2098,
+      "step": 48070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.590964040312416e-05,
+      "loss": 1.2002,
+      "step": 48080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.5894642767407615e-05,
+      "loss": 1.2234,
+      "step": 48090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.587964459384606e-05,
+      "loss": 1.1679,
+      "step": 48100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.586464588380798e-05,
+      "loss": 1.1805,
+      "step": 48110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5849646638661853e-05,
+      "loss": 1.3435,
+      "step": 48120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.583464685977634e-05,
+      "loss": 1.2777,
+      "step": 48130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5819646548519986e-05,
+      "loss": 1.3938,
+      "step": 48140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5804645706261514e-05,
+      "loss": 1.1454,
+      "step": 48150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5789644334369636e-05,
+      "loss": 1.6416,
+      "step": 48160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.57746424342131e-05,
+      "loss": 1.3286,
+      "step": 48170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5759640007160766e-05,
+      "loss": 1.3975,
+      "step": 48180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.574463705458145e-05,
+      "loss": 1.4875,
+      "step": 48190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.572963357784411e-05,
+      "loss": 1.5109,
+      "step": 48200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.571462957831768e-05,
+      "loss": 1.1065,
+      "step": 48210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5699625057371194e-05,
+      "loss": 1.2431,
+      "step": 48220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.568462001637371e-05,
+      "loss": 1.1877,
+      "step": 48230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.56696144566943e-05,
+      "loss": 1.2013,
+      "step": 48240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5654608379702145e-05,
+      "loss": 1.2246,
+      "step": 48250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5639601786766436e-05,
+      "loss": 0.999,
+      "step": 48260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.562459467925643e-05,
+      "loss": 1.3148,
+      "step": 48270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.56095870585414e-05,
+      "loss": 1.366,
+      "step": 48280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5594578925990714e-05,
+      "loss": 1.018,
+      "step": 48290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5579570282973735e-05,
+      "loss": 1.1258,
+      "step": 48300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.55645611308599e-05,
+      "loss": 1.2357,
+      "step": 48310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.554955147101871e-05,
+      "loss": 1.1582,
+      "step": 48320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5534541304819674e-05,
+      "loss": 1.3045,
+      "step": 48330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5519530633632366e-05,
+      "loss": 1.266,
+      "step": 48340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5504519458826406e-05,
+      "loss": 1.296,
+      "step": 48350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5489507781771454e-05,
+      "loss": 1.3603,
+      "step": 48360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.547449560383724e-05,
+      "loss": 1.2973,
+      "step": 48370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5459482926393515e-05,
+      "loss": 1.3142,
+      "step": 48380
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5444469750810055e-05,
+      "loss": 1.3363,
+      "step": 48390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.542945607845674e-05,
+      "loss": 1.1919,
+      "step": 48400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.541444191070343e-05,
+      "loss": 1.3416,
+      "step": 48410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.53994272489201e-05,
+      "loss": 1.1561,
+      "step": 48420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.538441209447671e-05,
+      "loss": 1.2199,
+      "step": 48430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.536939644874329e-05,
+      "loss": 1.3491,
+      "step": 48440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5354380313089925e-05,
+      "loss": 1.2251,
+      "step": 48450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.533936368888671e-05,
+      "loss": 1.1191,
+      "step": 48460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.532434657750384e-05,
+      "loss": 1.2545,
+      "step": 48470
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5309328980311493e-05,
+      "loss": 1.303,
+      "step": 48480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5294310898679925e-05,
+      "loss": 1.1856,
+      "step": 48490
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5279292333979436e-05,
+      "loss": 1.2177,
+      "step": 48500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.526427328758036e-05,
+      "loss": 1.336,
+      "step": 48510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.52492537608531e-05,
+      "loss": 1.6078,
+      "step": 48520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.523423375516806e-05,
+      "loss": 1.2114,
+      "step": 48530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.521921327189572e-05,
+      "loss": 1.2961,
+      "step": 48540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.52041923124066e-05,
+      "loss": 1.3463,
+      "step": 48550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.518917087807123e-05,
+      "loss": 1.5425,
+      "step": 48560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.517414897026025e-05,
+      "loss": 1.0238,
+      "step": 48570
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.515912659034428e-05,
+      "loss": 1.1294,
+      "step": 48580
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.514410373969402e-05,
+      "loss": 1.2454,
+      "step": 48590
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.512908041968018e-05,
+      "loss": 1.2068,
+      "step": 48600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.511405663167355e-05,
+      "loss": 1.3181,
+      "step": 48610
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.509903237704494e-05,
+      "loss": 1.3094,
+      "step": 48620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.508400765716523e-05,
+      "loss": 1.208,
+      "step": 48630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.506898247340529e-05,
+      "loss": 1.288,
+      "step": 48640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.505395682713607e-05,
+      "loss": 1.4243,
+      "step": 48650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.5038930719728565e-05,
+      "loss": 1.3259,
+      "step": 48660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.5023904152553796e-05,
+      "loss": 1.2286,
+      "step": 48670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.500887712698284e-05,
+      "loss": 1.3015,
+      "step": 48680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4993849644386795e-05,
+      "loss": 1.3733,
+      "step": 48690
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4978821706136816e-05,
+      "loss": 1.2165,
+      "step": 48700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4963793313604114e-05,
+      "loss": 1.4424,
+      "step": 48710
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.494876446815992e-05,
+      "loss": 1.1533,
+      "step": 48720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4933735171175494e-05,
+      "loss": 1.1578,
+      "step": 48730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4918705424022166e-05,
+      "loss": 1.1721,
+      "step": 48740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4903675228071284e-05,
+      "loss": 1.4644,
+      "step": 48750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.488864458469427e-05,
+      "loss": 1.4237,
+      "step": 48760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.487361349526257e-05,
+      "loss": 1.2883,
+      "step": 48770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4858581961147616e-05,
+      "loss": 1.1601,
+      "step": 48780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4843549983721e-05,
+      "loss": 1.1388,
+      "step": 48790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.482851756435423e-05,
+      "loss": 1.3727,
+      "step": 48800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.481348470441894e-05,
+      "loss": 1.1535,
+      "step": 48810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.479845140528675e-05,
+      "loss": 0.9777,
+      "step": 48820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.478341766832936e-05,
+      "loss": 1.2298,
+      "step": 48830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.476838349491851e-05,
+      "loss": 1.3673,
+      "step": 48840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4753348886425916e-05,
+      "loss": 1.2185,
+      "step": 48850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4738313844223433e-05,
+      "loss": 1.3933,
+      "step": 48860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.472327836968285e-05,
+      "loss": 1.0845,
+      "step": 48870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.470824246417611e-05,
+      "loss": 1.2873,
+      "step": 48880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4693206129075105e-05,
+      "loss": 1.4896,
+      "step": 48890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4678169365751764e-05,
+      "loss": 1.3495,
+      "step": 48900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4663132175578135e-05,
+      "loss": 1.2707,
+      "step": 48910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4648094559926223e-05,
+      "loss": 1.26,
+      "step": 48920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4633056520168134e-05,
+      "loss": 1.3458,
+      "step": 48930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.461801805767598e-05,
+      "loss": 1.3453,
+      "step": 48940
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.46029791738219e-05,
+      "loss": 1.1922,
+      "step": 48950
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.458793986997809e-05,
+      "loss": 1.4255,
+      "step": 48960
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.457290014751677e-05,
+      "loss": 1.3661,
+      "step": 48970
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4557860007810246e-05,
+      "loss": 1.2956,
+      "step": 48980
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.454281945223081e-05,
+      "loss": 1.1534,
+      "step": 48990
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.452777848215078e-05,
+      "loss": 1.1886,
+      "step": 49000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.451273709894258e-05,
+      "loss": 1.2293,
+      "step": 49010
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.44976953039786e-05,
+      "loss": 1.1047,
+      "step": 49020
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.448265309863132e-05,
+      "loss": 1.3792,
+      "step": 49030
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.446761048427323e-05,
+      "loss": 1.3,
+      "step": 49040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.445256746227687e-05,
+      "loss": 1.1171,
+      "step": 49050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4437524034014784e-05,
+      "loss": 1.1444,
+      "step": 49060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.442248020085961e-05,
+      "loss": 1.362,
+      "step": 49070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.4407435964183986e-05,
+      "loss": 1.282,
+      "step": 49080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.439239132536058e-05,
+      "loss": 1.2146,
+      "step": 49090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.437734628576212e-05,
+      "loss": 1.5993,
+      "step": 49100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.436230084676136e-05,
+      "loss": 1.1151,
+      "step": 49110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.434725500973109e-05,
+      "loss": 1.3419,
+      "step": 49120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.433220877604415e-05,
+      "loss": 1.5298,
+      "step": 49130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.431716214707338e-05,
+      "loss": 1.1774,
+      "step": 49140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.43021151241917e-05,
+      "loss": 1.4273,
+      "step": 49150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.428706770877203e-05,
+      "loss": 1.2264,
+      "step": 49160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.427201990218735e-05,
+      "loss": 1.0585,
+      "step": 49170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.4256971705810674e-05,
+      "loss": 1.4064,
+      "step": 49180
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.424192312101505e-05,
+      "loss": 1.3312,
+      "step": 49190
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.422687414917354e-05,
+      "loss": 1.1178,
+      "step": 49200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.4211824791659246e-05,
+      "loss": 1.181,
+      "step": 49210
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.419677504984534e-05,
+      "loss": 1.5156,
+      "step": 49220
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.4181724925105006e-05,
+      "loss": 1.4541,
+      "step": 49230
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.4166674418811456e-05,
+      "loss": 1.3231,
+      "step": 49240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.415162353233795e-05,
+      "loss": 1.2137,
+      "step": 49250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.413657226705775e-05,
+      "loss": 1.1452,
+      "step": 49260
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.412152062434421e-05,
+      "loss": 1.3388,
+      "step": 49270
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.410646860557068e-05,
+      "loss": 1.2464,
+      "step": 49280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.409141621211055e-05,
+      "loss": 1.2869,
+      "step": 49290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.407636344533724e-05,
+      "loss": 1.2331,
+      "step": 49300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.40613103066242e-05,
+      "loss": 1.287,
+      "step": 49310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.404625679734495e-05,
+      "loss": 1.1255,
+      "step": 49320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.4031202918873014e-05,
+      "loss": 1.2168,
+      "step": 49330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.4016148672581935e-05,
+      "loss": 1.1822,
+      "step": 49340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.400109405984533e-05,
+      "loss": 1.0959,
+      "step": 49350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.398603908203679e-05,
+      "loss": 1.1623,
+      "step": 49360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.3970983740530015e-05,
+      "loss": 1.2668,
+      "step": 49370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.395592803669869e-05,
+      "loss": 1.2985,
+      "step": 49380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.394087197191654e-05,
+      "loss": 1.3535,
+      "step": 49390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.3925815547557326e-05,
+      "loss": 1.4672,
+      "step": 49400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.391075876499483e-05,
+      "loss": 1.1933,
+      "step": 49410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.389570162560289e-05,
+      "loss": 1.1542,
+      "step": 49420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.388064413075537e-05,
+      "loss": 1.3368,
+      "step": 49430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.386558628182616e-05,
+      "loss": 1.3574,
+      "step": 49440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.385052808018918e-05,
+      "loss": 1.5783,
+      "step": 49450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.383546952721838e-05,
+      "loss": 1.2639,
+      "step": 49460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.382041062428774e-05,
+      "loss": 1.2551,
+      "step": 49470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.380535137277132e-05,
+      "loss": 1.2416,
+      "step": 49480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.379029177404312e-05,
+      "loss": 1.3656,
+      "step": 49490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.377523182947726e-05,
+      "loss": 1.4326,
+      "step": 49500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.3760171540447835e-05,
+      "loss": 1.0762,
+      "step": 49510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.3745110908329e-05,
+      "loss": 1.2656,
+      "step": 49520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.373004993449493e-05,
+      "loss": 1.2891,
+      "step": 49530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.371498862031985e-05,
+      "loss": 1.15,
+      "step": 49540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.369992696717797e-05,
+      "loss": 1.3196,
+      "step": 49550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.368486497644357e-05,
+      "loss": 1.2544,
+      "step": 49560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.366980264949098e-05,
+      "loss": 1.1986,
+      "step": 49570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.365473998769448e-05,
+      "loss": 1.3477,
+      "step": 49580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.363967699242849e-05,
+      "loss": 1.2976,
+      "step": 49590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.362461366506735e-05,
+      "loss": 1.1523,
+      "step": 49600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.360955000698552e-05,
+      "loss": 1.3105,
+      "step": 49610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.359448601955745e-05,
+      "loss": 1.2679,
+      "step": 49620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.357942170415759e-05,
+      "loss": 1.2758,
+      "step": 49630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.356435706216051e-05,
+      "loss": 1.2136,
+      "step": 49640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.354929209494071e-05,
+      "loss": 1.1862,
+      "step": 49650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.353422680387278e-05,
+      "loss": 1.2517,
+      "step": 49660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.351916119033131e-05,
+      "loss": 1.5123,
+      "step": 49670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3504095255690944e-05,
+      "loss": 1.1377,
+      "step": 49680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3489029001326354e-05,
+      "loss": 1.3096,
+      "step": 49690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.34739624286122e-05,
+      "loss": 1.301,
+      "step": 49700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3458895538923226e-05,
+      "loss": 1.2684,
+      "step": 49710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3443828333634184e-05,
+      "loss": 1.175,
+      "step": 49720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.342876081411982e-05,
+      "loss": 1.2661,
+      "step": 49730
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.341369298175497e-05,
+      "loss": 1.3877,
+      "step": 49740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3398624837914454e-05,
+      "loss": 1.0107,
+      "step": 49750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3383556383973155e-05,
+      "loss": 1.2434,
+      "step": 49760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.336848762130594e-05,
+      "loss": 1.3279,
+      "step": 49770
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.335341855128773e-05,
+      "loss": 1.4296,
+      "step": 49780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.333834917529348e-05,
+      "loss": 1.2114,
+      "step": 49790
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3323279494698176e-05,
+      "loss": 1.2811,
+      "step": 49800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3308209510876806e-05,
+      "loss": 1.2556,
+      "step": 49810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.329313922520442e-05,
+      "loss": 1.1493,
+      "step": 49820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3278068639056035e-05,
+      "loss": 1.1834,
+      "step": 49830
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.326299775380677e-05,
+      "loss": 1.1801,
+      "step": 49840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.324792657083175e-05,
+      "loss": 1.1585,
+      "step": 49850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.323285509150609e-05,
+      "loss": 1.2594,
+      "step": 49860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.321778331720495e-05,
+      "loss": 1.3213,
+      "step": 49870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.320271124930355e-05,
+      "loss": 1.2639,
+      "step": 49880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.318763888917711e-05,
+      "loss": 1.181,
+      "step": 49890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.317256623820085e-05,
+      "loss": 1.0944,
+      "step": 49900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.315749329775007e-05,
+      "loss": 1.0444,
+      "step": 49910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.314242006920005e-05,
+      "loss": 1.3899,
+      "step": 49920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3127346553926135e-05,
+      "loss": 1.4084,
+      "step": 49930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.311227275330368e-05,
+      "loss": 1.3878,
+      "step": 49940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.309719866870806e-05,
+      "loss": 1.3662,
+      "step": 49950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.308212430151467e-05,
+      "loss": 1.1183,
+      "step": 49960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.306704965309893e-05,
+      "loss": 1.2649,
+      "step": 49970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.305197472483634e-05,
+      "loss": 1.2816,
+      "step": 49980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.303689951810234e-05,
+      "loss": 1.306,
+      "step": 49990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.302182403427247e-05,
+      "loss": 1.1412,
+      "step": 50000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.300674827472223e-05,
+      "loss": 1.4391,
+      "step": 50010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2991672240827194e-05,
+      "loss": 1.1872,
+      "step": 50020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2976595933962946e-05,
+      "loss": 1.1904,
+      "step": 50030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.29615193555051e-05,
+      "loss": 1.3628,
+      "step": 50040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.294644250682927e-05,
+      "loss": 1.0709,
+      "step": 50050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.293136538931114e-05,
+      "loss": 1.4487,
+      "step": 50060
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2916288004326354e-05,
+      "loss": 1.4438,
+      "step": 50070
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2901210353250654e-05,
+      "loss": 1.4214,
+      "step": 50080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.288613243745974e-05,
+      "loss": 1.2682,
+      "step": 50090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.287105425832941e-05,
+      "loss": 1.4497,
+      "step": 50100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.28559758172354e-05,
+      "loss": 1.2125,
+      "step": 50110
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2840897115553525e-05,
+      "loss": 1.2984,
+      "step": 50120
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.282581815465961e-05,
+      "loss": 1.2022,
+      "step": 50130
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2810738935929526e-05,
+      "loss": 1.163,
+      "step": 50140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.279565946073912e-05,
+      "loss": 1.2733,
+      "step": 50150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.278057973046431e-05,
+      "loss": 1.1092,
+      "step": 50160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.276549974648099e-05,
+      "loss": 1.2164,
+      "step": 50170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.275041951016512e-05,
+      "loss": 1.4442,
+      "step": 50180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.273533902289266e-05,
+      "loss": 1.274,
+      "step": 50190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.272025828603963e-05,
+      "loss": 1.2361,
+      "step": 50200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.270517730098199e-05,
+      "loss": 1.3723,
+      "step": 50210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.269009606909579e-05,
+      "loss": 1.0709,
+      "step": 50220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2675014591757124e-05,
+      "loss": 1.2967,
+      "step": 50230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2659932870342035e-05,
+      "loss": 1.3665,
+      "step": 50240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2644850906226637e-05,
+      "loss": 1.2112,
+      "step": 50250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.262976870078704e-05,
+      "loss": 1.4384,
+      "step": 50260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2614686255399425e-05,
+      "loss": 1.3267,
+      "step": 50270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.259960357143993e-05,
+      "loss": 1.5372,
+      "step": 50280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.258452065028473e-05,
+      "loss": 1.4847,
+      "step": 50290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2569437493310084e-05,
+      "loss": 1.2418,
+      "step": 50300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2554354101892175e-05,
+      "loss": 1.3173,
+      "step": 50310
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2539270477407296e-05,
+      "loss": 1.2097,
+      "step": 50320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2524186621231707e-05,
+      "loss": 1.2524,
+      "step": 50330
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.250910253474169e-05,
+      "loss": 1.1287,
+      "step": 50340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.24940182193136e-05,
+      "loss": 1.3282,
+      "step": 50350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2478933676323737e-05,
+      "loss": 1.2453,
+      "step": 50360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2463848907148484e-05,
+      "loss": 1.2691,
+      "step": 50370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2448763913164215e-05,
+      "loss": 1.3254,
+      "step": 50380
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.243367869574731e-05,
+      "loss": 1.3074,
+      "step": 50390
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.241859325627423e-05,
+      "loss": 1.2026,
+      "step": 50400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2403507596121385e-05,
+      "loss": 1.3197,
+      "step": 50410
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.238842171666524e-05,
+      "loss": 1.2421,
+      "step": 50420
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.237333561928229e-05,
+      "loss": 1.4447,
+      "step": 50430
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.235824930534902e-05,
+      "loss": 1.0364,
+      "step": 50440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.234316277624197e-05,
+      "loss": 1.3153,
+      "step": 50450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.232807603333766e-05,
+      "loss": 1.1723,
+      "step": 50460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2312989078012665e-05,
+      "loss": 1.2589,
+      "step": 50470
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.229790191164355e-05,
+      "loss": 1.4622,
+      "step": 50480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.228281453560693e-05,
+      "loss": 1.3886,
+      "step": 50490
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2267726951279425e-05,
+      "loss": 1.1611,
+      "step": 50500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.225263916003764e-05,
+      "loss": 1.3335,
+      "step": 50510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.223755116325827e-05,
+      "loss": 1.1589,
+      "step": 50520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.222246296231795e-05,
+      "loss": 1.1977,
+      "step": 50530
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2207374558593414e-05,
+      "loss": 1.3142,
+      "step": 50540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2192285953461354e-05,
+      "loss": 1.2057,
+      "step": 50550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.21771971482985e-05,
+      "loss": 1.3651,
+      "step": 50560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.216210814448159e-05,
+      "loss": 1.4046,
+      "step": 50570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.214701894338741e-05,
+      "loss": 1.3616,
+      "step": 50580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.213192954639273e-05,
+      "loss": 1.3452,
+      "step": 50590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2116839954874365e-05,
+      "loss": 1.3,
+      "step": 50600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2101750170209116e-05,
+      "loss": 1.4873,
+      "step": 50610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.208666019377384e-05,
+      "loss": 1.1537,
+      "step": 50620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2071570026945374e-05,
+      "loss": 1.3788,
+      "step": 50630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2056479671100614e-05,
+      "loss": 1.3261,
+      "step": 50640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2041389127616426e-05,
+      "loss": 1.4346,
+      "step": 50650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2026298397869734e-05,
+      "loss": 1.3263,
+      "step": 50660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.201120748323745e-05,
+      "loss": 1.4491,
+      "step": 50670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1996116385096515e-05,
+      "loss": 1.2181,
+      "step": 50680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.19810251048239e-05,
+      "loss": 1.152,
+      "step": 50690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.196593364379656e-05,
+      "loss": 1.3625,
+      "step": 50700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.195084200339151e-05,
+      "loss": 1.2004,
+      "step": 50710
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.193575018498573e-05,
+      "loss": 1.3543,
+      "step": 50720
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.192065818995625e-05,
+      "loss": 1.6968,
+      "step": 50730
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.190556601968012e-05,
+      "loss": 1.3563,
+      "step": 50740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.189047367553439e-05,
+      "loss": 1.2273,
+      "step": 50750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.187538115889613e-05,
+      "loss": 1.2597,
+      "step": 50760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1860288471142435e-05,
+      "loss": 1.2388,
+      "step": 50770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1845195613650375e-05,
+      "loss": 1.2576,
+      "step": 50780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1830102587797114e-05,
+      "loss": 1.1169,
+      "step": 50790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1815009394959756e-05,
+      "loss": 1.2102,
+      "step": 50800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.179991603651547e-05,
+      "loss": 1.3828,
+      "step": 50810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1784822513841404e-05,
+      "loss": 1.2758,
+      "step": 50820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.176972882831472e-05,
+      "loss": 1.1581,
+      "step": 50830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.175463498131265e-05,
+      "loss": 1.3271,
+      "step": 50840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.173954097421237e-05,
+      "loss": 1.4305,
+      "step": 50850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1724446808391136e-05,
+      "loss": 1.4932,
+      "step": 50860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.170935248522615e-05,
+      "loss": 1.2778,
+      "step": 50870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1694258006094676e-05,
+      "loss": 1.1861,
+      "step": 50880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1679163372374e-05,
+      "loss": 1.4853,
+      "step": 50890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.166406858544135e-05,
+      "loss": 1.3197,
+      "step": 50900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.164897364667409e-05,
+      "loss": 1.1906,
+      "step": 50910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.163387855744948e-05,
+      "loss": 1.3196,
+      "step": 50920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.161878331914486e-05,
+      "loss": 1.2306,
+      "step": 50930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.160368793313756e-05,
+      "loss": 1.2837,
+      "step": 50940
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1588592400804905e-05,
+      "loss": 1.3014,
+      "step": 50950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1573496723524314e-05,
+      "loss": 1.1992,
+      "step": 50960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1558400902673096e-05,
+      "loss": 1.1938,
+      "step": 50970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.15433049396287e-05,
+      "loss": 1.2387,
+      "step": 50980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.152820883576849e-05,
+      "loss": 1.1937,
+      "step": 50990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.151311259246987e-05,
+      "loss": 1.1036,
+      "step": 51000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.149801621111031e-05,
+      "loss": 1.1642,
+      "step": 51010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.148291969306721e-05,
+      "loss": 1.2889,
+      "step": 51020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1467823039718044e-05,
+      "loss": 1.2455,
+      "step": 51030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1452726252440274e-05,
+      "loss": 1.3417,
+      "step": 51040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1437629332611356e-05,
+      "loss": 1.425,
+      "step": 51050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1422532281608814e-05,
+      "loss": 1.2619,
+      "step": 51060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.140743510081011e-05,
+      "loss": 1.1612,
+      "step": 51070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.139233779159279e-05,
+      "loss": 1.1906,
+      "step": 51080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1377240355334356e-05,
+      "loss": 1.2625,
+      "step": 51090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.136214279341235e-05,
+      "loss": 1.1298,
+      "step": 51100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.134704510720433e-05,
+      "loss": 1.2264,
+      "step": 51110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.133194729808783e-05,
+      "loss": 1.2326,
+      "step": 51120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.131684936744044e-05,
+      "loss": 1.2895,
+      "step": 51130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1301751316639743e-05,
+      "loss": 1.2394,
+      "step": 51140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.128665314706331e-05,
+      "loss": 1.335,
+      "step": 51150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.127155486008878e-05,
+      "loss": 1.2355,
+      "step": 51160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.125645645709371e-05,
+      "loss": 1.1116,
+      "step": 51170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.124135793945578e-05,
+      "loss": 1.2612,
+      "step": 51180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.122625930855259e-05,
+      "loss": 1.1132,
+      "step": 51190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1211160565761806e-05,
+      "loss": 1.3731,
+      "step": 51200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1196061712461085e-05,
+      "loss": 1.4521,
+      "step": 51210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.118096275002805e-05,
+      "loss": 1.4079,
+      "step": 51220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.116586367984042e-05,
+      "loss": 1.2368,
+      "step": 51230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.115076450327586e-05,
+      "loss": 1.2638,
+      "step": 51240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.113566522171207e-05,
+      "loss": 1.1036,
+      "step": 51250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.112056583652677e-05,
+      "loss": 1.1808,
+      "step": 51260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1105466349097644e-05,
+      "loss": 1.2167,
+      "step": 51270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1090366760802433e-05,
+      "loss": 1.2339,
+      "step": 51280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.107526707301885e-05,
+      "loss": 1.4341,
+      "step": 51290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.106016728712467e-05,
+      "loss": 1.3669,
+      "step": 51300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1045067404497616e-05,
+      "loss": 1.4424,
+      "step": 51310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.102996742651546e-05,
+      "loss": 1.1136,
+      "step": 51320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.101486735455596e-05,
+      "loss": 1.1393,
+      "step": 51330
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.099976718999689e-05,
+      "loss": 1.54,
+      "step": 51340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.098466693421605e-05,
+      "loss": 1.3059,
+      "step": 51350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.096956658859122e-05,
+      "loss": 1.0878,
+      "step": 51360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0954466154500214e-05,
+      "loss": 1.3417,
+      "step": 51370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.093936563332082e-05,
+      "loss": 1.3559,
+      "step": 51380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.092426502643087e-05,
+      "loss": 1.2429,
+      "step": 51390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.090916433520819e-05,
+      "loss": 1.2171,
+      "step": 51400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0894063561030616e-05,
+      "loss": 1.2403,
+      "step": 51410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.087896270527598e-05,
+      "loss": 1.3424,
+      "step": 51420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.086386176932212e-05,
+      "loss": 1.2661,
+      "step": 51430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.08487607545469e-05,
+      "loss": 1.3462,
+      "step": 51440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.083365966232819e-05,
+      "loss": 1.143,
+      "step": 51450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.081855849404385e-05,
+      "loss": 1.1827,
+      "step": 51460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0803457251071785e-05,
+      "loss": 1.2999,
+      "step": 51470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.078835593478983e-05,
+      "loss": 1.0715,
+      "step": 51480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.07732545465759e-05,
+      "loss": 1.409,
+      "step": 51490
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.075815308780789e-05,
+      "loss": 1.0834,
+      "step": 51500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.07430515598637e-05,
+      "loss": 1.3987,
+      "step": 51510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.072794996412126e-05,
+      "loss": 1.3186,
+      "step": 51520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0712848301958435e-05,
+      "loss": 1.1885,
+      "step": 51530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.069774657475319e-05,
+      "loss": 1.2342,
+      "step": 51540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.068264478388345e-05,
+      "loss": 1.4019,
+      "step": 51550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.066754293072712e-05,
+      "loss": 1.3622,
+      "step": 51560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.065244101666218e-05,
+      "loss": 1.3211,
+      "step": 51570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.063733904306652e-05,
+      "loss": 1.3383,
+      "step": 51580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.062223701131814e-05,
+      "loss": 1.1558,
+      "step": 51590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0607134922794974e-05,
+      "loss": 1.3788,
+      "step": 51600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.059203277887498e-05,
+      "loss": 1.1399,
+      "step": 51610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.057693058093613e-05,
+      "loss": 1.1176,
+      "step": 51620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0561828330356386e-05,
+      "loss": 1.1965,
+      "step": 51630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0546726028513724e-05,
+      "loss": 1.3427,
+      "step": 51640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.053162367678615e-05,
+      "loss": 1.3388,
+      "step": 51650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.051652127655159e-05,
+      "loss": 1.2379,
+      "step": 51660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.05014188291881e-05,
+      "loss": 1.3619,
+      "step": 51670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0486316336073626e-05,
+      "loss": 1.1364,
+      "step": 51680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.047121379858618e-05,
+      "loss": 1.2849,
+      "step": 51690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0456111218103764e-05,
+      "loss": 1.2234,
+      "step": 51700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.044100859600438e-05,
+      "loss": 1.3065,
+      "step": 51710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.042590593366605e-05,
+      "loss": 1.1896,
+      "step": 51720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0410803232466766e-05,
+      "loss": 1.3563,
+      "step": 51730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.039570049378456e-05,
+      "loss": 1.3118,
+      "step": 51740
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.038059771899745e-05,
+      "loss": 1.1589,
+      "step": 51750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.0365494909483434e-05,
+      "loss": 1.321,
+      "step": 51760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.035039206662059e-05,
+      "loss": 1.181,
+      "step": 51770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.033528919178688e-05,
+      "loss": 1.1617,
+      "step": 51780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.0320186286360394e-05,
+      "loss": 1.3233,
+      "step": 51790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.030508335171914e-05,
+      "loss": 1.1979,
+      "step": 51800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.028998038924114e-05,
+      "loss": 1.4353,
+      "step": 51810
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.027487740030449e-05,
+      "loss": 1.3359,
+      "step": 51820
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.025977438628717e-05,
+      "loss": 1.0127,
+      "step": 51830
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.024467134856725e-05,
+      "loss": 1.2917,
+      "step": 51840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.0229568288522776e-05,
+      "loss": 1.1555,
+      "step": 51850
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.0214465207531816e-05,
+      "loss": 1.2205,
+      "step": 51860
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.01993621069724e-05,
+      "loss": 1.4418,
+      "step": 51870
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.0184258988222564e-05,
+      "loss": 1.1585,
+      "step": 51880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.01691558526604e-05,
+      "loss": 1.1614,
+      "step": 51890
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.015405270166392e-05,
+      "loss": 1.0997,
+      "step": 51900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.0138949536611225e-05,
+      "loss": 1.5178,
+      "step": 51910
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.0123846358880354e-05,
+      "loss": 1.1863,
+      "step": 51920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.010874316984937e-05,
+      "loss": 1.3089,
+      "step": 51930
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.0093639970896335e-05,
+      "loss": 1.2836,
+      "step": 51940
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.007853676339929e-05,
+      "loss": 1.3021,
+      "step": 51950
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.006343354873632e-05,
+      "loss": 1.4353,
+      "step": 51960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.0048330328285476e-05,
+      "loss": 1.1568,
+      "step": 51970
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.003322710342484e-05,
+      "loss": 1.3908,
+      "step": 51980
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.001812387553245e-05,
+      "loss": 1.2602,
+      "step": 51990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.000302064598639e-05,
+      "loss": 1.0821,
+      "step": 52000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 104004,
+  "num_train_epochs": 2,
+  "save_steps": 500,
+  "total_flos": 6.124293860268442e+16,
+  "trial_name": null,
+  "trial_params": null
+}